software b_老虎社区_美港股上老虎

2021-11-23

software business can scale cheaply but may also face more competition. Key is to devise a strategy to monopolize the market for one' s software products

Are Software Stocks In A Bubble Or A New Paradigm?

免责声明：上述内容仅代表发帖人个人观点，不构成本平台的任何投资建议。

精彩评论

我们需要你的真知灼见来填补这片空白

发表看法

{"i18n":{"language":"zh_CN"},"detailType":1,"isChannel":false,"data":{"magic":2,"id":875652332,"tweetId":"875652332","gmtCreate":1637648198502,"gmtModify":1637648998676,"author":{"id":4093410373855950,"idStr":"4093410373855950","authorId":4093410373855950,"authorIdStr":"4093410373855950","name":"Humbly","avatar":"https://static.tigerbbs.com/cebd1eb0173a5ad49e43baaa7c0252d8","vip":1,"userType":1,"introduction":"","boolIsFan":false,"boolIsHead":false,"crmLevel":6,"crmLevelSwitch":0,"individualDisplayBadges":[],"fanSize":34,"starInvestorFlag":false},"themes":[],"images":[],"coverImages":[],"extraTitle":"","html":"<html><head></head><body>software business can scale cheaply but may also face more competition. Key is to devise a strategy to monopolize the market for one' s software products</body></html>","htmlText":"<html><head></head><body>software business can scale cheaply but may also face more competition. Key is to devise a strategy to monopolize the market for one' s software products</body></html>","text":"software business can scale cheaply but may also face more competition. Key is to devise a strategy to monopolize the market for one' s software products","highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":2,"repostSize":0,"favoriteSize":0,"link":"https://laohu8.com/post/875652332","repostId":1196838355,"repostType":4,"repost":{"id":"1196838355","pubTimestamp":1637645542,"share":"https://www.laohu8.com/m/news/1196838355?lang=&edition=full","pubTime":"2021-11-23 13:32","market":"us","language":"en","title":"Are Software Stocks In A Bubble Or A New Paradigm?","url":"https://stock-news.laohu8.com/highlight/detail?id=1196838355","media":"Seeking Alpha","summary":"Summary\n\nIs slower growth decay and higher gross margins causing the rich valuations seen among soft","content":"Summary\n<ul>\n <li>Is slower growth decay and higher gross margins causing the rich valuations seen among software stocks today?</li>\n <li>Is the long-term market opportunity bigger and more profitable for today’s software stocks and this is why companies’ management decide to spend colossal amounts in opex?</li>\n <li>If the answers to these questions are \"yes\" then could they explain the historically high multiples of today's market?</li>\n <li>We take a deep dive into the characteristics of today’s software companies to open up a discussion for explaining current valuations.</li>\n</ul>\n<img src=\"https://static.tigerbbs.com/bee61fbc78b90df059fde3958cfad089\" tg-width=\"1536\" tg-height=\"864\" width=\"100%\" height=\"auto\">metamorworks/iStock via Getty Images\nAbout This Article\nIn this article we share some of our recent research about software stock valuations. It may come across as controversial to some of the more conservative/value investors but we’re looking to open up the conversation rather than make outlandish claims. Going into the research, given that we specialize in software investing and run a subscription service dedicated to that, we acknowledge that we have a bias toward justifying the current rich valuations within the tech sector. We try and address such biases whilst making rationales for the findings from our research.\nThroughout the article we attempt to pack in lots of qualitative and quantitative insights to help the reader better understand SaaS stocks and consider different viewpoints for understanding today’s market. Stocks discussed/referenced are Cloudflare(NYSE:NET), SentinelOne(NYSE:S), Asana(NYSE:ASAN), Datadog(NASDAQ:DDOG), Twilio(NYSE:TWLO), Okta(NASDAQ:OKTA), Zscaler(NASDAQ:ZS), Zoom(NASDAQ:ZM), Dropbox(NASDAQ:DBX), ServiceNow(NYSE:NOW), Dynatrace(NYSE:DT), Elastic(NYSE:ESTC), New Relic(NYSE:NEWR), Check Point(NASDAQ:CHKP), Palo Alto(NASDAQ:PANW), Cisco(NASDAQ:CSCO), and Juniper(NYSE:JNPR). The hypotheses and qualitative stuff are in the first half and the quantitative results are roughly in the second half of the article.\nIntro\nThe‘new paradigm’phrase often gets bandied about. This is especially the case in uncertain fields like finance and investments where there are continual attempts to either rationalize or dismiss the current lofty prices and bullish behavior.\nInvestors who believe the market is nearing bubble-like territory highlight the current multiples that are very extended compared to historical averages. This camp also reminds others that‘new paradigm’and‘this time is different’narratives were repeatedly used during bubbles like the dotcom in the late 1990s and the housing market in the mid-2000s, and we all know how those unfolded. And such groups of investors tend to be the more experienced that have been investing long enough tovividlyrecall the boom and bust of previous cycles and the ensuing despair. Also, they are likely to be predominately value-oriented with experience investing in companies producing something physical rather than information/virtual. However, some of these investors might have become overly cautious in the current market due to behavioral biases such as oversimplification [comparing today to the dotcom era just because of extended multiples driven by tech stocks], or even confirmation bias [where bubble believers are selectively looking for characteristics that compare today’s market to the dotcom era without regard for the differences].\nInvestors who believe the market has entered a new era tend to be the younger investors who are aware of the previous bubbles but didn’t experience the subsequent crashes at first hand. Despite many younger investors’ efforts to research historical events, this doesn’t compensate for the inexperience. As a result, this camp is probably more affected by representativeness bias [whereby recent results/data are extrapolated far out into the future] and hence it will be a natural tendency for most to assume that this time is different. Also, this camp is probably overly influenced by the likes of Cathy Woods [we also take note of Wood’s view], the plethora of YouTubers pumping up hot stocks, and the general FOMO.\nFigure 1 - An Interesting Statement that Applies to Several Investing Biases\n<img src=\"https://static.tigerbbs.com/9a2d7270c804f2e953a9164d2e214545\" tg-width=\"406\" tg-height=\"277\" width=\"100%\" height=\"auto\">Source: cruxinvestor.com\nAfter contemplating these different investor views and reconciling them with our own views whilst also being mindful to limit our own biases, we decided to look for quantitative evidence [other than traditional metrics] that may assist in understanding whether today’s market is indeed&nbsp;in&nbsp;or&nbsp;heading&nbsp;toward bubble territory or whether valuations can be rationalized. And as the tech sector holds the majority of the rich valuations as well as our area of focus [growth and value tech], we decided to focus the research on U.S. software stocks.\nHypothesis\nWe must admit we didn’t go into this research entirely neutral. We published an&nbsp;articles&nbsp;even months ago to present a similar research exercise [scroll down to the Tech subheading] in which we hypothesized that U.S. software stocks in 2021 had statistically&nbsp;higher gross marginsand statistically&nbsp;lower operating margins&nbsp;than U.S. software stocks in 2007. And if the null hypotheses were rejected [statistical parlance for saying they are true] we had an explanation at the ready which could potentially [and partly] explain the then lofty valuations. We had the same hypotheses going into the research we’re about to present in this article.\nGross Margins\nOur hypothesis and explanations are related to the emergence of the cloud and SaaS businesses. We hypothesize that gross margins are higher today because of software companies’ utilization of public cloud infrastructures that equip their businesses with greater scalability thus driving higher revenue growth, and at the same time the multitenancy that brings down the cost of revenue percentage. However, it's our thinking that the multitenancy benefit declines with size, which is why some SaaS vendors switch to colocation data centres when they reach a certain threshold of revenue – DBX is a case in point here. Though as this is an industry-wide focused research effort and larger SaaS vendors [say $2bn in revenue] are the minority, we predict the average [median and mean] gross margin is higher today.\nIn the earlier&nbsp;research&nbsp;[scroll down to Tech subheading] we discovered an 850 basis-point median gross margin differential between U.S. software stocks in 2021 and 2007. This wasn’t actuallystatisticallysignificant but it grabbed our attention given how difficult it is for firms to improve their gross margins. Although more research is required – hence the reason we present this article today - it got us thinking why gross marginsmaybe higher today.\nThe obvious reason, which we previously alluded to, is the economies of scale that the cloud and SaaS business models produce. It’s so much easier for SaaS firms to onboard customers compared to pre-cloud days. Gone are the days when users had only the option of installing the software on their PC in order to access the product. Also gone are the days when users needed to periodically update the software, in fact, SaaS users are completely hands-off with updates to the point they might not even notice it’s happened. All these greatly contributes to faster customer onboarding/retention, faster growth, quicker upsells, and greater scale economies.\nThe more nuanced reasons for possibly explaining higher gross margins [lower cost of revenue] today, relate to the degree of virtualization and software-defined approaches that the top SaaS firms utilize. Virtualization/software-defined approaches reduce the number of servers required and hence lowers data centre costs, hosting costs, energy costs, and depreciation. It also enables software firms to create better algos, for example, to compress data packets to optimize bandwidth thus bringing down network costs.\nWe think comparing NET and AKAM – two Content Delivery Network [CDN] providers that have both ventured off into security [SASE, ZTNA, WAF, etc.] – is an ideal example for thinking about the possible cost of revenue benefits of cloud/SaaS because NET is the epitome of the cloud-native/SaaS business model and AKAM is a company working hard to shift to a more modern architecture but is entangled in an abundance of technical debt.\nFigure 2 - Comparing Two CDN Providers: Pure SaaS [Cloudflare] vs Legacy Rooted [Akamai]\n<img src=\"https://static.tigerbbs.com/5fbf2b270c21a0ddc80e0b33eb0000a5\" tg-width=\"383\" tg-height=\"245\" width=\"100%\" height=\"auto\">Source: Convequity\nThis difference when comparing pure SaaS firms to legacy firms isn’t always there because in reality there are lots of interconnected variables at play. However, given the similarities between these two firms – the high number of PoPs [points of presence], the content delivery, DDoS protection, and the security offerings – it is a great like-for-like comparison for isolating the effects of virtualization/software-defined approaches on gross margins. What isn’t reflected in the gross margin differential, however, is NET giving DDoS protection to millions for free, including us. This makes the difference even more astounding.\nOperating Margins\nAdditionally, we hypothesize that operating margins are lower today because of the customer LTV [Lifetime Value] concept introduced by recurring revenue in the SaaS business model. We surmise that to capture this larger revenue opportunity management teams are spending more opex to drive faster growth. This opex is in the form of S&amp;M, G&amp;A, and SBC [stock-based compensation] which is distributed across the two expense categories. [Note cost of revenue also includes SBC but it is typically a negligible amount].\nCompanies that are evidently firm believers in the new opportunities via the customer LTV concept are S and ASAN. Respectively, these two SaaS vendors most-recent-quarter operating margins were -147% and -67%. This didn’t happen in pre-cloud eras. And, from observations, it seems as though the most aggressive SaaS companies also have the most visionary founder leaders occupying the CEO position.\nSuch SaaS leaders have grand visions for the market opportunities that lay ahead that will retain existing customers and win new ones – and the recurring SaaS revenue model is what makes such visions realizable and [potentially] highly profitable. Cloud/SaaS makes it much easier than yesteryear to create new products [because of the software abstraction attributed to software-defined approaches] and enter new markets [thanks to the scalability offered by IaaS]. Moreover, the recurring revenue model makes it far easier to retain customers compared to the software license model that entails customers coughing up all the cost up front when renewal is due or to buy the new version. Plus, at renewal S&amp;M costs are lower because sales personnel have to do much less work to persuade a customer to stay on with monthly billing, thus longer-term profitability should be higher vs license-based firms of the past.\nIn S’s case, we agree with management that they have the best NGAV [next-gen antivirus] software available, and their TAM [Total Addressable Market] is in the range of $10bn to $20bn according to market research estimates. This presents a huge opportunity – having the best software within a large TAM with currently c. 1% penetration [ARR of $198m]. However, CRWD is the NGAV market leader with a strong product offering and highly effective sales teams. There is no other way for S to capitalize on their superior software other than to go all-in with -147% operating margin to drive market share. They could play the slow game but in the dynamic world of software their NGAV could quickly lose its edge – so they need to strike while the iron is hot. And once they’ve built a solid market presence and reputation, they should be able to cool off the opex and benefit from natural market momentum.For more on S and to get an overview of the AV industry&nbsp;check out this article.\nBut the point is that the prize for a BoB [best-of-breed] SaaS company is huge in the modern landscape. To reiterate S’s case – the best NGAV in a huge [and quickly expanding] TAM with c. 1% penetration. One can begin to at least partly see the reasoning behind the colossal opex. And at NTM P/S 76x we won’t be buying S shares but we appreciate its business, competitive advantages, and market opportunity and hence will act quickly on the dip – which could very well occur following the lockup expiry in December.\nWe first covered ASAN in April 2021 for our Asymmetric Tech Investments subscription members, and to us it was clear as to why management was spending so aggressively. ASAN’s software is super-easy to install [literally a couple of clicks], and it is great for coordinating projects and tasks across large teams. ASAN’s software is built on a new abstraction layer called the coordination layer that is actually highly differentiated and novel yet, in our opinion, not fully recognized by others and often oversimplified as a weak competitor in the crowded productivity tool space.\nASAN’s S&amp;M expense is 70%+ of revenue to outcompete the rivalry, but once users are onboarded there begins a very organic land-and-expand throughout the respective customer organizations due to the quality and ease of using the software, which presumably generates a contribution margin trajectory per customer organization that is very steep. ASAN’s Net Dollar-Based Retention Rate of 125% for accounts spending over $5k indicates this may be true. Also, ASAN’s gross margin-adjusted Magic Number [gross margin * QoQ net new revenue / prior quarter’s S&amp;M] is quite high among SaaS cohorts, which is another sign of its contribution margin economics.\nProductivity companies like ASAN and MNDY have gross margins close to 90% because a lot of work the software does is storing and retrieving text-based information. This is a highly organized and structured form of data, therefore, compute requirements per operation are much less, and as text-based data is super lightweight it doesn’t cost much to transmit, thus reducing bandwidth costs. Moreover, employees using productivity software will generally accept a little latency because it’s not ultra-time sensitive [unlike the deep packet inspection of real-time application data that cybersecurity firms like PANW and FTNT have to do] and hence ASAN/MNDY will be making even further savings in bandwidth. Indeed, assuming these two dominate the coordination market,when their opex normalizes/matures they will be very profitable businesses.\nFor the most software-defined-oriented SaaS firms the land and expand is much less effort – just think ASAN, NET, DDOG, OKTA – but at the same time competition is intense. So, the SaaS firms with the best products look to capitalize on their technological edge with huge opex to win the new logos, knowing that retaining and growing their wallet will be the easier and the more profitable part. And naturally, customer service is of a very high quality because otherwise customers will leave before the SaaS companies has recouped its S&amp;M costs.\nThis contrasts with the software license model whereby the high contribution margin stage is the beginning of the customer relationship, and as a result may not motivate software companies to have their finger on the pulse so much in regards to customer service, and in general not be as focused in nurturing long-term relationships.\nBelow we speculate the generalized peaks and troughs of sales and customer service effort/expense and customer satisfaction for the license model and SaaS. And from these diagrams it's easier to grasp why SaaS and customer LTV go hand-in-hand – SaaS is just more conducive to consistency and long-term customer satisfaction.\nFigure 3 - Comparing Software License Model to SaaS Model: Fluctuations in Effort/Expense &amp; Customer Satisfaction\n<img src=\"https://static.tigerbbs.com/5daa7636643e83cf48230165464d3c53\" tg-width=\"640\" tg-height=\"194\" width=\"100%\" height=\"auto\">Source: Convequity\nThe above diagram also illustrates why it's worth it for SaaS firms to spend more in opex today than license-based software firms did in the past. Higher customer satisfaction equals more profitable long-term relationships. And it’s the delivery and usability of SaaS products that enables this higher customer satisfaction. We’ll also like to add that many SaaS firms are treating their huge opex as the new capex. Indeed, the economic benefits of much of the opex occurs long after 12 months, so in our opinion, it’s time accounting standards changed to reflect this by shifting the expense from income statement to cash flow statement. NYU Corporate Finance Professor Damodaran is a prominent advocate for changing the accounting standards to better reflect economic reality.\nBesides pure S&amp;M expenditure, SBC is a big contributor to opex, distributed across all opex categories. But it’s kind of clear as to why SBC has become so important for leading software companies. Due to deep-seated issues pertaining to education curriculums and the rapid pace of software innovations in the past decade – thanks to things like the cloud, containers, and microservices – a huge skills shortage across IT has emerged. One of the most prominent shortages is that within the cybersecurity industry, though there are many other software sectors experiencing this demand/supply mismatch. Hence, there's a fight to win the best talent within a limited pool and SBC is one of the biggest incentives to use. It also improves alignment of long-term interest, among employees, management and the shareholders.\nHowever, it looks as though the significant portion of the colossal SBC generates poor returns. The following table shows that there is a moderate correlation [0.39] between SBC and revenue growth. And the differences between the lower, median, upper quartiles of the distribution of 95 stocks indicates there is a moderate correlation that wepresumeis a cause-and-effect relationship. But given the aggression in SaaS companies SBC policies, one would assume the revenue growth, and as a by-product, ROI, would be more likely – the r-squared indicates that SBC only explains 15% of revenue growth variance.\nFigure 4 - Moderate Correlation between SBC and Growth\n<img src=\"https://static.tigerbbs.com/c361ac06e4cc4be70ec5f77b590047ea\" tg-width=\"640\" tg-height=\"156\" width=\"100%\" height=\"auto\">Source: Convequity\nWe think this highlights the struggles that SaaS firms are encountering. They are forced to offer high SBC to attract limited talent but the resulting business performance is far from probable. And, of course, this is because great SaaS firms have a bunch of characteristics needed to make them a success – such as product led growth, great leadership, empowering culture, a huge freemium base [e.g., NET, DDOG, ASAN], the use of open protocols, software-defined approaches, software design, CI/CD, and microservices architectures. So, as investors there is a lot of opportunity in identifying the truly BoB SaaS players that have these characteristics to ensure a good ROI in SBC is generated to avoid a net negative impact on shareholder value.\nFrom the 95-stock peer group shown above, we compared four observability players’ SBC % and TTM revenue growth – DDOG, ESTC, DT, and NEWR – shown in the following table. We would strongly surmise that DDOG has plenty more of the aforementioned characteristics than the other three. So, it makes one wonder why the management, of say NEWR, will decide to pay out 21% of revenue in SBC before they’ve addressed their fit with the aforementioned characteristics and other leading SaaS attributes. The SBC above and beyond the market rate is likely to be a waste of investment because SBC alone is not enough.\nFigure 5 - Leading Observability Players: SBC % vs TTM Growth\n<img src=\"https://static.tigerbbs.com/adabb209ab58eea92422cf5d036fdb12\" tg-width=\"273\" tg-height=\"127\" width=\"100%\" height=\"auto\">Source: Convequity\nIt’s also noteworthy that as cloud has lowered the entry barriers for software companies, more opex is necessary to flourish in a more competitive environment versus pre-cloud eras. As we show in the forthcoming data, there are double the amount of public software companies today compared to 2007. And this is in the context of SaaS reducing vendor lock-in thereby putting even more pressure on companies to keep customers happy and loyal in order to maximize the customer LTV potential. It’s very much a winner-takes-all in the majority of SaaS markets, which is why a big component of our investing approach is to invest in the BoB SaaS vendors at a reasonable price.\nIn statistical parlance, the hypotheses were as follows:\n<ul>\n <li>Gross margin</li>\n</ul>\n<ul>\n <ul>\n <li>H0: median&nbsp;gross margintoday&nbsp;and median&nbsp;gross margin in 2007&nbsp;are the same.</li>\n <li>H1: median&nbsp;gross margin today&nbsp;and median&nbsp;gross margin in 2007&nbsp;are different.</li>\n </ul>\n <li>Operating margin</li>\n</ul>\n<ul>\n <ul>\n <li>H0: median&nbsp;operating margin today&nbsp;and median&nbsp;gross margin in 2007&nbsp;are the same.</li>\n <li>H1: median&nbsp;operating margin today&nbsp;and median&nbsp;gross margin in 2007&nbsp;are different.</li>\n </ul>\n</ul>\nIf the gross margin H0 [null hypothesis] is rejected, meaning H1 is true, then perhaps this is causing the richer valuations today. Companies with higher gross margins have better core economics within their business and hence long-term, mature-stage profitability would probably be comparatively higher. Could this potentially and/or partly explain the current valuations?\nIf the operating margin H0 is rejected, then the notion of how the cloud and SaaS is affecting opex could partly explain the relatively lower earnings for software stocks versus earlier periods.\nIn essence, if our theory can be rationalized then higher gross margins are pushing up the numerator of P/E [and other ratios] and lower operating margins are pushing down the denominator of P/E [and other ratios].\nAs a sidenote, there is also a strong argument that this depressing of earnings is unwarranted because a lot of S&amp;M expense, as earlier touched on, is usually generating economic benefits [revenue] for longer than 12 months – and the same as with R&amp;D. Therefore, in our view accounting standards are long overdue a change to better reflect the economic benefits of expenditures, and if/when this occurs it would have a big impact on the multiples were witnessing today [not the valuations though because opex is just shifting to capex] as such expense will be transferred from the income statement to the cash flow statement. Though, we’re not confident such change will happen anytime soon – it could easily take another decade. Just think how long it took to change lease accounting. Anyway, we believe there is a solid argument that today’s higher multiples for software stocks can bepartlyjustified by the pervasiveness of the cloud and SaaS business.\nThe caveat, however, is that justification of today’s multiples will be considerably higher for those SaaS stocks that represent BoB characteristics and are not throwing away SBC. And this loops back to the winner-takes-all idea within SaaS markets – multiples are stretched across the board but likely justified for those stocks that will win their market, carve out a long runway of growth, and have opportunities to increase margins.\nRevenue Growth and Revenue Growth Decay\nAs previously pointed out, in our earlier&nbsp;research&nbsp;we only made hypotheses for gross margin and operating margin but this time we’re also predicting that TTM revenue growth is statistically higher and revenue growth decay is statistically smaller today versus 2007. In fact, for the revenue growth decay we used stocks from the 2011-2013 and 2019-2021 time periods and simply calculated the decay in growth – for example, Company XYZ growth in 2013 subtract Company XYZ growth in 2011.\n<ul>\n <li>Revenue growth</li>\n</ul>\n<ul>\n <ul>\n <li>H0: median&nbsp;TTMrevenue growth today&nbsp;and median&nbsp;TTMrevenue growth in 2007&nbsp;are the same.</li>\n <li>H1: median&nbsp;TTMrevenue growth today&nbsp;and median&nbsp;TTMrevenue growth in 2007&nbsp;are different.</li>\n </ul>\n</ul>\n<ul>\n <li>Revenue growth decay</li>\n</ul>\n<ul>\n <ul>\n <li>H0: median&nbsp;revenue growth decay during 2019-2021&nbsp;and median&nbsp;revenue growth decay during 2011-2013&nbsp;are the same.</li>\n <li>H1: median&nbsp;revenue growth decay during 2019-2021&nbsp;and median&nbsp;revenue growth decay during 2011-2013&nbsp;are different.</li>\n </ul>\n</ul>\nOur reasoning for the revenue growth hypothesis is that in using the hyperscalers IaaS, software companies can scale their business faster today than in yesteryear and hence can drive faster revenue growth. Software-defined-everything also enables SaaS firms to produce new products with more frequency which underscored this hypothesis.\nThe logic behind speculating that today’s software companies experience a smaller delta in revenue growth decay, or slower growth deceleration, is that SaaS/cloud-native companies are able to produce new products and enter new markets much quicker than in earlier pre-cloud periods [2011-2013 would be pre-cloud for the high majority of stocks selected]. Hence, we surmise that growth decay is less today.\nNET, ZS, TWLO, OKTA, and NOW are just a few SaaS firms that have demonstrated tremendous TAM expansion via prolific new product releases which have helped slow down growth deceleration. The extensible software design and architecture adopted by these SaaS firms enables them to avoid getting bogged down in technical debt. This is very different to earlier legacy approaches that entailed an application being tied toonecode base consisting of tens of thousands of lines of code with poor annotation/explanation linked to them. Each line of code is dependent on thousands of others, making troubleshooting a daunting task, and may result in the application being out-of-action for 24 hours at least. Just imagine the reputational and financial damage a top software firm would experience if their app went offline for such a time period.\nLegacy firms are often aware for the need for change to keep up with the fast-paced software industry but are stifled into inertia because of the monolithic structure of their apps. Changing one line of code may bring the entire application offline and troubleshooting a 100k line code base is painstakingly difficult to do. So, just imagine the feeling of dismay when facing the need to radically alter the app and how it's delivered. Often the pain is too much to bear and just quick fixes and improvements will be opted for instead.\nThis cycle continues until the once-leading software firm is now categorized as legacy and revenue growth is diminished to low-single-digit or even flat. CHKP and JNPR are two cases in which we strongly surmise this is the case. Many software firms become victim of this scenario because of a lack of leadership whereby the company is not led by the innovative founder, or are led by a conservative founder [like in CHKP’s case]. Consequently, very often a new owner is brought in like STG [Symphony Technology Group] that privatizes the business with a predominant focus on profits over innovation – and then there is certainly no return to the cutting edge.\nIt really is easier to start from scratch. Computer chip guru,Jim Keller, advocates doing just this every 3-5 years – starting again with a blank canvas, a completely clean slate, even a new company altogether. Otherwise, the change is too painful and the risk is too high. But how can a large software company manage this whilst protecting its revenue and profitable cash flows? It’s nigh-on impossible with old software architectures. It requires a combination of modern architecture, modern DevOps, and modern organizational processes, to combat the loom of legacy in order for software firms to S-Curve stack their way to being perennial pioneers. It appears the likes of the FAANGM stocks as well as CRM and NOW have this blend of attributes to make sure they stay on the steep part of the S-Curve.\nConsider when Eric Yuan [CEO and Founder of Zoom] wanted to radically improve upon the WebEx meeting software, but he couldn’t persuade CSCO executives who didn’t want to risk its high cash flows for a cloud-native project with uncertain visibility into the future. So, it wasn't until 2020, when ZM overtook WebEx as the market leader, that CSCO realized how poor its solution actually is. Now, CSCO is in crisis mode to try and claw back some market share but it’s probably already far too late. Such an example, that happens frequently, illuminates the difficulty for software firms to stay modern.\nDesigning software with built-in extensibility is one approach modern SaaS firms are adopting to avoid the legacy fate. Fully utilizing software-defined approaches, CI/CD, and microservices architectures are other intertwining approaches to enable them to improve their software with less stress and be adaptable to future industry developments. They aren’t legacy-proof but will help them maintain their BoB status, continue to deliver the best software, and slow the growth decay and maintain double-digit growth for many years. Aside from the previously mentioned SaaS stocks, we would suggest the likes of DBX, FSLY, JAMF, SUMO, and IS also design their applications in a way that fits into the description of extensible software.\nFigure 6 - Application Architectures: Monolithic vs Microservices\n<img src=\"https://static.tigerbbs.com/706413663e80097f8f9c4cd336674788\" tg-width=\"557\" tg-height=\"272\" width=\"100%\" height=\"auto\">Source: n-ix.com/microservices-vs-monolith\nThe other reason for postulating that today’s software firms suffer slower growth decay is because the required capex to move into new markets is minimal in comparison to the pre-cloud era because they don’t need to purchase and implement more servers. Once an application is developed and ready to be deployed, they can scale it across the elastic supply of AWS, Azure, Alibaba Cloud, or GCP servers that provide all the necessary compute, networking, and storage requirements – and ramp-up or decrease capacity according to demand. There have been several other industry developments that allow further specialization and outsourcing of the work without the need for one vendor to do all the back-end operations by itself. That includes services like colocation and better virtualization technologies that allow SaaS vendors to have confidential information running in untrusted physical locations.\nThe statistical tests we used to compare all the medians of the different stock groups are called the Mood Test and the Kruskal-Wallis Test. We also did median tests for FCF margin, the Rule of 40, and FCF/EV but we didn’t form prior hypotheses. The tests were all based on a level of 95% confidence.\nCriteria for Stocks in 2007 and 2021\nAs October has just passed, it made sense to use market prices and financials as of October 2007 so we could compare the 2021 and 2007 groups based on the same time of the year. And this worked out well because prior to GFC the Nasdaq 100 peaked on 31stOctober 2007. So, we’re comparing today’s software stocks to a cycle peak which seems conservative for the tests [we presumed historically high gross and operating margins, and high revenue growth in late 2007 just before GFC] and unfavourable for our hypotheses.\nFigure 7 - Weekly Chart of Nasdaq 100\n<img src=\"https://static.tigerbbs.com/b9dd349d51b5f58225cfdc7788d07f98\" tg-width=\"640\" tg-height=\"378\" width=\"100%\" height=\"auto\">Source: TradingView, Convequity annotation\nSo, based on market price and financial data as of the end of October in 2007 and end of October in 2021 we set selection criteria using GuruFocus. In fact, we did the 2007 vs 2021 comparison based on two criteria sets.\n<ul>\n <li>Firstly, we searched for U.S. headquartered software stocks with&nbsp;revenue above $100m&nbsp;and a market cap above $1bn for both 2007 and 2021 [as of end of October]. The lower bound for revenue and market cap was to exclude microcap stocks.</li>\n <li>Secondly, to refine the research we ran the search process again to look for U.S. headquartered software stocks with&nbsp;revenue between $200m and $2bn&nbsp;and a minimum market cap of $1bn for both 2007 and 2021. We thought it would be useful to compare 2007 and 2021 based on this revenue bracket because from our experience it seems as though most of today’s richer valuations tend to be assigned to stocks with revenue in this ballpark.</li>\n</ul>\nWe did some data cleansing to remove the most extreme outliers and financials that didn’t make sense [like a 123% gross margin, for example]. And after some trial and error we found that searching for only U.S. headquartered companies removed some of the questionable companies and financials in both periods – hence the reason for this criterion outlined above.\nFindings\nAnnual Revenue Above $100m\nBased on the criterion of annual revenue above $100m, we found 70 software stocks in 2007 and 151 software stocks in the 2021 cohort. The gross margin and revenue growth hypotheses were not proven accurate [couldn’t reject the H0] but the operating margin and revenue growth decay hypotheses were proven accurate [rejected the H0]. The following table shows the results [note we didn’t test the means because of uncertainty of the presence of normal distributions but have included them for the readers perusal].\nFigure 8 - October 2007 vs October 2021: Annual Revenue Above $100m\n<img src=\"https://static.tigerbbs.com/6728d4448fe29bd07ba6216274daa258\" tg-width=\"640\" tg-height=\"490\" width=\"100%\" height=\"auto\">Source: Convequity\nWhilst not statistically significant, the 480-basis point higher median gross margin for the 2021 cohort is notable considering how difficult it is for companies to improve gross margins. Unlike margins further down the income statement, gross margins don’t vary much as they reflect the core economics and the core of a business model. When we run this test earlier this year, the gross margin differential was 850-basis points. We think the results from both our earlier and today’s research on gross margin warrants continued examination.\nGiven that companies are still rather early on in the cloud journey, it’s likely that the gross margin delta will be greater if tested in the future once companies have further optimized their operations in the clouds. It’s also important to consider what we mentioned earlier, in that to counter the diminishing effects of the multitenancy, software firms that have already scaled into a $1bn-$2bn revenue business will likely be better off switching to colocation data centers to drive up gross margins. So, if other SaaS firms follow the likes of DBX we think its probable gross margins will improve in the long-term.\nWhilst the Mood and Kruskal-Wallis tests proved us statistically accurate for the operating margin hypothesis, the increase in stock-based compensation [SBC] expense over the years does muddy the waters. Therefore, in future research we shall remove the SBC effects [originally, we did want to search for S&amp;M expense only but we couldn’t retrieve this from GuruFocus]. However, as earlier mentioned, software firms have been increasing SBC to attract and retain top talent required to win customers [a S&amp;M expense] and develop good SaaS products and features [a R&amp;D expense]. So, there is a pretty strong case to claim that the rising SBC is also closely tied to companies aiming to maximize the customer LTV opportunity.\nWe were surprised in the equality between the 2007 and 2021 TTM revenue growth rates [for both mean and median]. There isn’t much of a counter here, our hypothesis is wrong. But it would be interesting to do the TTM growth rate comparison for two periods with the same business cycle stage. October 2007 was late stage whilst currently we’re in the recovery stage. Just food for thought for if we were to do this analysis in the future.\nThe differences [both the mean and median] in revenue growth decay look small but the median difference is in fact statistically significant according to the tests. As is shown, we also refined the revenue growth decay testing to just the top revenue growth quintile [based on 2011 and 2019 revenue growth rates]. The median difference is not significant but we would say it is non-negligible, and the difference in the means would probably be statistically significant had we tested it.\nFor both the revenue growth and revenue growth decay we feel like further research in the future is needed. But it does look like it’s possible that, whilst the cloud may not be enabling software companies to drive faster growth, it may be enabling them to maintain growth for longer.\nWithout formally declaring a hypothesis, we did anticipate the significant [and obvious] differences in FCF/EV. However, we thought that what we believed to be higher expected revenue growth in 2021 would have closed the difference in EV/FCF/Growth. In isolation, the EV/FCF/Growth does indicate today’s software stocks are very overvalued. Though we would contend that this metric doesn’t account for future growth [it’s based on TTM growth because we couldn’t retrieve forward growth estimates for 2007] and hence won’t reflect any view on slower growth decay in today’s environment. In other words, the metric doesn’t reflect the compounding growth of the future, and as the decay is smaller and the growth outlook is more visible and probable thanks to recurring revenue models, and the aforementioned ease of producing new products and entering new markets, investors may be attaching premiums to reflect this.\nAnnual Revenue Between $200m and $2bn\nBased on the criterion of annual revenue between $200m and $2bn, we found 45 software stocks in 2007 and 93 software stocks in the 2021 cohort. Aside from selecting this criterion to focus on the richest software valuations, companies in this revenue bracket tend to experience the fastest growth and are BoB software companies that are rising to be industry-defining leaders – and these types of stocks feature in a sizeable portion of our portfolio.\nAs shown in the following table, the results are similar to the previous criteria set but the gross margins are pretty much equal and the difference in revenue growth decay is not statistically significant. So, on the whole our set of hypotheses are less accurate for this revenue bracket. However, the median revenue growth rates do show a meaningful difference.\nFigure 9 - October 2007 vs October 2021: Annual Revenue between $200m and $2bn\n<img src=\"https://static.tigerbbs.com/ae2abf0893e35b6a5d9d23b8b46c6295\" tg-width=\"640\" tg-height=\"493\" width=\"100%\" height=\"auto\">Source: Convequity\nWe are really surprised that there isn’t a notable difference in the median gross margins. And we’re trying to avoid coming up with what may seem as excuses and remain objective, but it is possible that the pandemic-induced remote working environments of 2020 increased SaaS-based cost of revenue percentages and has thus lowered the TTM gross margin for 2021. Not all SaaS-like firms experienced this but quite a few did, like ZS and RNG and other per-user [not usage based], cloud-native SaaS firms that position themselves in the middle of traffic and are thus sometimes victim of higher bandwidth costs amid sudden surges in demand. So, it will be worth running this gross margin testing for this revenue bracket again in the future.\nThe operating margin differential is much wider than the previous group which could be explained by this revenue bracket containing more of the aggressive companies, and this greater aggression manifests itself in higher opex in the form of S&amp;M and SBC.\nThere is no statistical significance in the difference between the medians for revenue growth decay and the same for the top quintile. Although, the differences are non-negligible, especially for the top quintile. From this we infer that the highest growers in the 2019-2021 period have been able to better minimize the growth decay than software companies achieved in the 2011-2013 period. These highest growth rate stocks are invariably the most cloud-native and hence they can produce new products and enter new markets with speed – just think of the likes of NET and DDOG – so the results here come as no surprise.\nThe FCF/EV and EV/FCF/Growth results draw similar reasoning/conclusions to the previous [TTM revenue &gt; $100m] group, but they are just more extended for the 2021 cohort. Indeed, the median 7.05 EV/FCF/Growth for the 2021 group is way higher than the 1.58 for the 2007 group, however, as we mentioned earlier this metric doesn’t account for future compounding growth. We’re not justifying the 7.05 but is the market implicitly increasing growth forecasts and extending explicit forecast periods in DCF valuations to factor in the slower growth decay in today’s environment? If so, then this may be boosting the numerator andpartlyexplain the differential.\nWithin the context of the winner-takes-all idea that we’ve highlighted in this article, we view the hyper-extended multiples, especially in this $200m-$2bn revenue bracket, as big bets by investors that they know who the winners are going to be. This in stark contrast to more traditional investing predicated on forecasting a company’s cash flow generation over the next five years which is not necessarily about picking winners but how the company in question can improve its business over time. Both are valid but very different approaches, and rather than outright say paying 60x P/S for a SaaS stock can be justified, here we are merely trying to find possible reasoning behind current market behavior rather than taking the easy view that the market is in a bubble.\nWith the winner-takes-all idea in mind, not all these stock multiples can be justified because most will not meet long-term expectations. Again, this is why an understanding of a software company’s technology, culture, application design, etc., in our opinion, is important to identify the winners and not just rely on an extrapolation of current growth.\nGross Margin Distribution Comparisons\nAs we contemplated the difference in median gross margins being statistically&nbsp;insignificant, but in light of the stickiness of gross margins and hence feeling that the differences are noteworthy, we decided to look at the distributions. The 2007 and 2021 distributions for the first selection [revenue above $100m] and the second selection [revenue between $200m and $2bn] were similar, therefore, for brevity we’ll only show the first selection.\nThe first chart below shows that the 2007 distribution looks more evenly spread whilst the 2021 distribution has a high peak in the 70%-75% gross margin level. However, the difference in the number of stocks selected [70 for 2007 and 151 for 2021] makes any conclusion difficult.\nFigure 10 - 2007 vs 2021: Gross Margin Distribution Comparison\n<img src=\"https://static.tigerbbs.com/0c246518ab933f9f85bbaac0dcce06ad\" tg-width=\"640\" tg-height=\"290\" width=\"100%\" height=\"auto\">Source: Convequity\nThe next chart shows the gross margin proportions across each distribution. We can see that almost 17% of the 2021 cohort had gross margins in the 70%-75% bucket. This contrasts with only 6% of the 2007 cohort that had gross margins in the 70%-75% bucket. This is certainly a big difference that illuminates the positive impact of software firms utilizing IaaS. And from our experience in analyzing cloud-native/SaaS stocks this does make sense as this gross margin bucket tends to be common, especially for firms that are achieving scale.\nFigure 11 - 2007 vs 2021: Gross Margin Distribution in Percentage Terms Comparison\n<img src=\"https://static.tigerbbs.com/b95c89fd472aecee6e2e0132e81bfea8\" tg-width=\"640\" tg-height=\"282\" width=\"100%\" height=\"auto\">Source: Convequity\nBased on the median gross margin differences [not statistically significant but notable] and the disparities in the distributions highlighting a boost in the proportion of stocks achieving 70%-75% gross margin, we believe future research is warranted. It also opens up the conversation that perhaps today’s software companies, or leading software companies, have better underlying economics and prospects for long-term, mature-stage profitability than software companies did in pre-cloud eras. If future research provided stronger evidence that this is the case, then the currently rich valuations [often construed as grossly overvalued] could be partly justified.\nOur last note on gross margins, is that as touched on earlier, gross margin levels are very dictated by the type of software sector. For example, cybersecurity companies rarely have gross margins above 80% because of the high-power computing needed to decrypt and re-encrypt traffic flowing through inspection points. As we discussed earlier, productivity companies like ASAN and MNDY have gross margins close to 90% because a lot of work the software does is storing and retrieving text-based information. Such cost of revenue differences between different software sectors, such as cybersecurity vs productivity, can be attributed to differences in information density. In high information density software like productivity, the entire list of tasks might take less than 1MB to store yet the daily inspected data for cybersecurity firm PANW could be more than 1GB per seat per day.\nTherefore, in future research we may look into controlling the software sector variable to test differences within specific sectors. For example, if there is a higher proportion of cybersecurity firms in the 2021 cohort versus the 2007 cohort, or vice versa, it may obscure the true results.\nConclusion\nAcross the testing we believe there is enough evidence to at least warrant further research by ourselves and/or others to get closer in determining or dismissing that software stocks are in a new paradigm.\nIn the broader software selection represented by all U.S. headquartered stocks with revenue above $100m and market cap above $1bn, our set of hypotheses are supported by either statistical differences or notable/meaningful differences – statistical differences were operating margin and revenue growth decay and the notable difference was gross margin. We found very similar results for the gross margin and operating margin differences in the research we published earlier in the year, which somewhat encourages us with the line of thinking outlined in this article.\nThe narrower software selection that honed in on the richer valued stocks, represented by all U.S. headquartered stocks with revenue between $200m and $2bn, was less supportive to our set of hypotheses. The only supportive statistical difference was the operating margin, but revenue growth and revenue growth decay differences were notable.\nAs expected, the FCF yield [FCF/EV] was statistically significant for both selections, though the growth-adjusted EV/FCF multiple was surprisingly much higher than anticipated in 2021.\nOn the whole, based on the testing results, we believe there is a moderate argument that today’s software firms’ use of IaaS and SaaS business models is helping them achieve increasing gross margins. We also think it’s possible that the cloud/SaaS is enabling higher revenue growth [for certain revenue brackets, i.e., the $200m to $2bn bracket] and slower revenue growth decay, despite not all the statistical tests not supporting this.\nAnd if some of today’s software firms are experiencing higher gross margins, higher growth, and slower growth decay compared to earlier periods, then higher long-term profitability and more sustained growth runways will be realized. And if this is being acknowledged by the market then perhaps this can explain the current valuations.\nIt seems as though higher opex is required in order for today’s software firms to be future winners amid the greater competition, capture the lucrative customer lifetime values up for grabs, leverage the relatively higher gross margins afforded by cloud computing, and mature with greater profitability than software firms of the past.\nIn effect, many software firms are trading short-term profits for long-term winner-takes-all dominance. Software has become more competitive but the prizes are bigger, necessitating huge upfront costs that current accounting standards deem should be included in the income statement despite the economic rewards of such costs lasting far longer than the next 12 months.\nThe higher opex is powering future growth for firms to capture the huge TAMs [Total Addressable Market] that have become available thanks to cloud computing, but in the present it is depressing earnings and extending [income statement-based] multiples and thus making valuations look stretched by traditional metrics.\nIndeed, many of the extended stock multiples cannot be justified because not all will materialize as long-term winners and hence expectations will not be met. We view the broad extended multiples of the software sector as big investor bets on who these winners will turn out to be – but only a portion will be right. This is in stark contrast to the more traditional type of investing entailing the forecasting of a company’s next 5-year cash flows that are not subject to it becoming the market’s winner as such.\nTo summarize all this in two brief sentences. We believe this research offers a moderate argument that today’s software firms have higher gross margins and slower growth decay that present investors with more profitable, sustainable business growth than software firms in pre-cloud periods. To capture this bigger prize among fiercer competition, firms are substantially ramping up opex which is depressing earnings and stretching multiples – some of which can be justified [the market winners] and others which cannot [the runners-up or losers].\nAll we’ve shared in this article is just our opinion on interpreting the data – we’re aware that we might be wrong. Hopefully it can at least spur on some discussion to help us refine or question these theories.","collect":0,"html":"<!DOCTYPE html>\n<html>\n<head>\n<meta http-equiv=\"Content-Type\" content=\"text/html; charset=utf-8\" />\n<meta name=\"viewport\" content=\"width=device-width,initial-scale=1.0,minimum-scale=1.0,maximum-scale=1.0,user-scalable=no\"/>\n<meta name=\"format-detection\" content=\"telephone=no,email=no,address=no\" />\n<title>Are Software Stocks In A Bubble Or A New Paradigm?</title>\n<style type=\"text/css\">\na,abbr,acronym,address,applet,article,aside,audio,b,big,blockquote,body,canvas,caption,center,cite,code,dd,del,details,dfn,div,dl,dt,\nem,embed,fieldset,figcaption,figure,footer,form,h1,h2,h3,h4,h5,h6,header,hgroup,html,i,iframe,img,ins,kbd,label,legend,li,mark,menu,nav,\nobject,ol,output,p,pre,q,ruby,s,samp,section,small,span,strike,strong,sub,summary,sup,table,tbody,td,tfoot,th,thead,time,tr,tt,u,ul,var,video{ font:inherit;margin:0;padding:0;vertical-align:baseline;border:0 }\nbody{ font-size:16px; line-height:1.5; color:#999; background:transparent; }\n.wrapper{ overflow:hidden;word-break:break-all;padding:10px; }\nh1,h2{ font-weight:normal; line-height:1.35; margin-bottom:.6em; }\nh3,h4,h5,h6{ line-height:1.35; margin-bottom:1em; }\nh1{ font-size:24px; }\nh2{ font-size:20px; }\nh3{ font-size:18px; }\nh4{ font-size:16px; }\nh5{ font-size:14px; }\nh6{ font-size:12px; }\np,ul,ol,blockquote,dl,table{ margin:1.2em 0; }\nul,ol{ margin-left:2em; }\nul{ list-style:disc; }\nol{ list-style:decimal; }\nli,li p{ margin:10px 0;}\nimg{ max-width:100%;display:block;margin:0 auto 1em; }\nblockquote{ color:#B5B2B1; border-left:3px solid #aaa; padding:1em; }\nstrong,b{font-weight:bold;}\nem,i{font-style:italic;}\ntable{ width:100%;border-collapse:collapse;border-spacing:1px;margin:1em 0;font-size:.9em; }\nth,td{ padding:5px;text-align:left;border:1px solid #aaa; }\nth{ font-weight:bold;background:#5d5d5d; }\n.symbol-link{font-weight:bold;}\n/* header{ border-bottom:1px solid #494756; } */\n.title{ margin:0 0 8px;line-height:1.3;color:#ddd; }\n.meta {color:#5e5c6d;font-size:13px;margin:0 0 .5em; }\na{text-decoration:none; color:#2a4b87;}\n.meta .head { display: inline-block; overflow: hidden}\n.head .h-thumb { width: 30px; height: 30px; margin: 0; padding: 0; border-radius: 50%; float: left;}\n.head .h-content { margin: 0; padding: 0 0 0 9px; float: left;}\n.head .h-name {font-size: 13px; color: #eee; margin: 0;}\n.head .h-time {font-size: 11px; color: #7E829C; margin: 0;line-height: 11px;}\n.small {font-size: 12.5px; display: inline-block; transform: scale(0.9); -webkit-transform: scale(0.9); transform-origin: left; -webkit-transform-origin: left;}\n.smaller {font-size: 12.5px; display: inline-block; transform: scale(0.8); -webkit-transform: scale(0.8); transform-origin: left; -webkit-transform-origin: left;}\n.bt-text {font-size: 12px;margin: 1.5em 0 0 0}\n.bt-text p {margin: 0}\n</style>\n</head>\n<body>\n<div class=\"wrapper\">\n<header>\n<h2 class=\"title\">\nAre Software Stocks In A Bubble Or A New Paradigm?\n</h2>\n\n<h4 class=\"meta\">\n\n\n2021-11-23 13:32 GMT+8&nbsp;&nbsp;&nbsp;<a href=https://seekingalpha.com/article/4471260-are-software-stocks-in-a-bubble-or-a-new-paradigm>Seeking Alpha</a>\n\n\n</h4>\n\n</header>\n<article>\n<div>\nSummary\n\nIs slower growth decay and higher gross margins causing the rich valuations seen among software stocks today?\nIs the long-term market opportunity bigger and more profitable for today’s ...\n\n<a href=\"https://seekingalpha.com/article/4471260-are-software-stocks-in-a-bubble-or-a-new-paradigm\">Web Link</a>\n\n</div>\n\n\n</article>\n</div>\n</body>\n</html>\n","type":0,"thumbnail":"","relate_stocks":{"NEWR":"New Relic","CHKP":"Check Point软件科技","ESTC":"Elastic N.V.","DBX":"Dropbox Inc.","CSCO":"思科","ZS":"Zscaler Inc.","ASAN":"阿莎娜","MSFT":"微软","ZM":"Zoom","NOW":"ServiceNow","NET":"Cloudflare, Inc.","JNPR":"瞻博网络","S":"SentinelOne, Inc","TWLO":"Twilio Inc","PANW":"Palo Alto Networks","DT":"Dynatrace Holdings LLC","OKTA":"Okta Inc.","DDOG":"Datadog"},"source_url":"https://seekingalpha.com/article/4471260-are-software-stocks-in-a-bubble-or-a-new-paradigm","is_english":true,"share_image_url":"https://static.laohu8.com/e9f99090a1c2ed51c021029395664489","article_id":"1196838355","content_text":"Summary\n\nIs slower growth decay and higher gross margins causing the rich valuations seen among software stocks today?\nIs the long-term market opportunity bigger and more profitable for today’s software stocks and this is why companies’ management decide to spend colossal amounts in opex?\nIf the answers to these questions are \"yes\" then could they explain the historically high multiples of today's market?\nWe take a deep dive into the characteristics of today’s software companies to open up a discussion for explaining current valuations.\n\nmetamorworks/iStock via Getty Images\nAbout This Article\nIn this article we share some of our recent research about software stock valuations. It may come across as controversial to some of the more conservative/value investors but we’re looking to open up the conversation rather than make outlandish claims. Going into the research, given that we specialize in software investing and run a subscription service dedicated to that, we acknowledge that we have a bias toward justifying the current rich valuations within the tech sector. We try and address such biases whilst making rationales for the findings from our research.\nThroughout the article we attempt to pack in lots of qualitative and quantitative insights to help the reader better understand SaaS stocks and consider different viewpoints for understanding today’s market. Stocks discussed/referenced are Cloudflare(NYSE:NET), SentinelOne(NYSE:S), Asana(NYSE:ASAN), Datadog(NASDAQ:DDOG), Twilio(NYSE:TWLO), Okta(NASDAQ:OKTA), Zscaler(NASDAQ:ZS), Zoom(NASDAQ:ZM), Dropbox(NASDAQ:DBX), ServiceNow(NYSE:NOW), Dynatrace(NYSE:DT), Elastic(NYSE:ESTC), New Relic(NYSE:NEWR), Check Point(NASDAQ:CHKP), Palo Alto(NASDAQ:PANW), Cisco(NASDAQ:CSCO), and Juniper(NYSE:JNPR). The hypotheses and qualitative stuff are in the first half and the quantitative results are roughly in the second half of the article.\nIntro\nThe‘new paradigm’phrase often gets bandied about. This is especially the case in uncertain fields like finance and investments where there are continual attempts to either rationalize or dismiss the current lofty prices and bullish behavior.\nInvestors who believe the market is nearing bubble-like territory highlight the current multiples that are very extended compared to historical averages. This camp also reminds others that‘new paradigm’and‘this time is different’narratives were repeatedly used during bubbles like the dotcom in the late 1990s and the housing market in the mid-2000s, and we all know how those unfolded. And such groups of investors tend to be the more experienced that have been investing long enough tovividlyrecall the boom and bust of previous cycles and the ensuing despair. Also, they are likely to be predominately value-oriented with experience investing in companies producing something physical rather than information/virtual. However, some of these investors might have become overly cautious in the current market due to behavioral biases such as oversimplification [comparing today to the dotcom era just because of extended multiples driven by tech stocks], or even confirmation bias [where bubble believers are selectively looking for characteristics that compare today’s market to the dotcom era without regard for the differences].\nInvestors who believe the market has entered a new era tend to be the younger investors who are aware of the previous bubbles but didn’t experience the subsequent crashes at first hand. Despite many younger investors’ efforts to research historical events, this doesn’t compensate for the inexperience. As a result, this camp is probably more affected by representativeness bias [whereby recent results/data are extrapolated far out into the future] and hence it will be a natural tendency for most to assume that this time is different. Also, this camp is probably overly influenced by the likes of Cathy Woods [we also take note of Wood’s view], the plethora of YouTubers pumping up hot stocks, and the general FOMO.\nFigure 1 - An Interesting Statement that Applies to Several Investing Biases\nSource: cruxinvestor.com\nAfter contemplating these different investor views and reconciling them with our own views whilst also being mindful to limit our own biases, we decided to look for quantitative evidence [other than traditional metrics] that may assist in understanding whether today’s market is indeed in or heading toward bubble territory or whether valuations can be rationalized. And as the tech sector holds the majority of the rich valuations as well as our area of focus [growth and value tech], we decided to focus the research on U.S. software stocks.\nHypothesis\nWe must admit we didn’t go into this research entirely neutral. We published an articles even months ago to present a similar research exercise [scroll down to the Tech subheading] in which we hypothesized that U.S. software stocks in 2021 had statistically higher gross marginsand statistically lower operating margins than U.S. software stocks in 2007. And if the null hypotheses were rejected [statistical parlance for saying they are true] we had an explanation at the ready which could potentially [and partly] explain the then lofty valuations. We had the same hypotheses going into the research we’re about to present in this article.\nGross Margins\nOur hypothesis and explanations are related to the emergence of the cloud and SaaS businesses. We hypothesize that gross margins are higher today because of software companies’ utilization of public cloud infrastructures that equip their businesses with greater scalability thus driving higher revenue growth, and at the same time the multitenancy that brings down the cost of revenue percentage. However, it's our thinking that the multitenancy benefit declines with size, which is why some SaaS vendors switch to colocation data centres when they reach a certain threshold of revenue – DBX is a case in point here. Though as this is an industry-wide focused research effort and larger SaaS vendors [say $2bn in revenue] are the minority, we predict the average [median and mean] gross margin is higher today.\nIn the earlier research [scroll down to Tech subheading] we discovered an 850 basis-point median gross margin differential between U.S. software stocks in 2021 and 2007. This wasn’t actuallystatisticallysignificant but it grabbed our attention given how difficult it is for firms to improve their gross margins. Although more research is required – hence the reason we present this article today - it got us thinking why gross marginsmaybe higher today.\nThe obvious reason, which we previously alluded to, is the economies of scale that the cloud and SaaS business models produce. It’s so much easier for SaaS firms to onboard customers compared to pre-cloud days. Gone are the days when users had only the option of installing the software on their PC in order to access the product. Also gone are the days when users needed to periodically update the software, in fact, SaaS users are completely hands-off with updates to the point they might not even notice it’s happened. All these greatly contributes to faster customer onboarding/retention, faster growth, quicker upsells, and greater scale economies.\nThe more nuanced reasons for possibly explaining higher gross margins [lower cost of revenue] today, relate to the degree of virtualization and software-defined approaches that the top SaaS firms utilize. Virtualization/software-defined approaches reduce the number of servers required and hence lowers data centre costs, hosting costs, energy costs, and depreciation. It also enables software firms to create better algos, for example, to compress data packets to optimize bandwidth thus bringing down network costs.\nWe think comparing NET and AKAM – two Content Delivery Network [CDN] providers that have both ventured off into security [SASE, ZTNA, WAF, etc.] – is an ideal example for thinking about the possible cost of revenue benefits of cloud/SaaS because NET is the epitome of the cloud-native/SaaS business model and AKAM is a company working hard to shift to a more modern architecture but is entangled in an abundance of technical debt.\nFigure 2 - Comparing Two CDN Providers: Pure SaaS [Cloudflare] vs Legacy Rooted [Akamai]\nSource: Convequity\nThis difference when comparing pure SaaS firms to legacy firms isn’t always there because in reality there are lots of interconnected variables at play. However, given the similarities between these two firms – the high number of PoPs [points of presence], the content delivery, DDoS protection, and the security offerings – it is a great like-for-like comparison for isolating the effects of virtualization/software-defined approaches on gross margins. What isn’t reflected in the gross margin differential, however, is NET giving DDoS protection to millions for free, including us. This makes the difference even more astounding.\nOperating Margins\nAdditionally, we hypothesize that operating margins are lower today because of the customer LTV [Lifetime Value] concept introduced by recurring revenue in the SaaS business model. We surmise that to capture this larger revenue opportunity management teams are spending more opex to drive faster growth. This opex is in the form of S&M, G&A, and SBC [stock-based compensation] which is distributed across the two expense categories. [Note cost of revenue also includes SBC but it is typically a negligible amount].\nCompanies that are evidently firm believers in the new opportunities via the customer LTV concept are S and ASAN. Respectively, these two SaaS vendors most-recent-quarter operating margins were -147% and -67%. This didn’t happen in pre-cloud eras. And, from observations, it seems as though the most aggressive SaaS companies also have the most visionary founder leaders occupying the CEO position.\nSuch SaaS leaders have grand visions for the market opportunities that lay ahead that will retain existing customers and win new ones – and the recurring SaaS revenue model is what makes such visions realizable and [potentially] highly profitable. Cloud/SaaS makes it much easier than yesteryear to create new products [because of the software abstraction attributed to software-defined approaches] and enter new markets [thanks to the scalability offered by IaaS]. Moreover, the recurring revenue model makes it far easier to retain customers compared to the software license model that entails customers coughing up all the cost up front when renewal is due or to buy the new version. Plus, at renewal S&M costs are lower because sales personnel have to do much less work to persuade a customer to stay on with monthly billing, thus longer-term profitability should be higher vs license-based firms of the past.\nIn S’s case, we agree with management that they have the best NGAV [next-gen antivirus] software available, and their TAM [Total Addressable Market] is in the range of $10bn to $20bn according to market research estimates. This presents a huge opportunity – having the best software within a large TAM with currently c. 1% penetration [ARR of $198m]. However, CRWD is the NGAV market leader with a strong product offering and highly effective sales teams. There is no other way for S to capitalize on their superior software other than to go all-in with -147% operating margin to drive market share. They could play the slow game but in the dynamic world of software their NGAV could quickly lose its edge – so they need to strike while the iron is hot. And once they’ve built a solid market presence and reputation, they should be able to cool off the opex and benefit from natural market momentum.For more on S and to get an overview of the AV industry check out this article.\nBut the point is that the prize for a BoB [best-of-breed] SaaS company is huge in the modern landscape. To reiterate S’s case – the best NGAV in a huge [and quickly expanding] TAM with c. 1% penetration. One can begin to at least partly see the reasoning behind the colossal opex. And at NTM P/S 76x we won’t be buying S shares but we appreciate its business, competitive advantages, and market opportunity and hence will act quickly on the dip – which could very well occur following the lockup expiry in December.\nWe first covered ASAN in April 2021 for our Asymmetric Tech Investments subscription members, and to us it was clear as to why management was spending so aggressively. ASAN’s software is super-easy to install [literally a couple of clicks], and it is great for coordinating projects and tasks across large teams. ASAN’s software is built on a new abstraction layer called the coordination layer that is actually highly differentiated and novel yet, in our opinion, not fully recognized by others and often oversimplified as a weak competitor in the crowded productivity tool space.\nASAN’s S&M expense is 70%+ of revenue to outcompete the rivalry, but once users are onboarded there begins a very organic land-and-expand throughout the respective customer organizations due to the quality and ease of using the software, which presumably generates a contribution margin trajectory per customer organization that is very steep. ASAN’s Net Dollar-Based Retention Rate of 125% for accounts spending over $5k indicates this may be true. Also, ASAN’s gross margin-adjusted Magic Number [gross margin * QoQ net new revenue / prior quarter’s S&M] is quite high among SaaS cohorts, which is another sign of its contribution margin economics.\nProductivity companies like ASAN and MNDY have gross margins close to 90% because a lot of work the software does is storing and retrieving text-based information. This is a highly organized and structured form of data, therefore, compute requirements per operation are much less, and as text-based data is super lightweight it doesn’t cost much to transmit, thus reducing bandwidth costs. Moreover, employees using productivity software will generally accept a little latency because it’s not ultra-time sensitive [unlike the deep packet inspection of real-time application data that cybersecurity firms like PANW and FTNT have to do] and hence ASAN/MNDY will be making even further savings in bandwidth. Indeed, assuming these two dominate the coordination market,when their opex normalizes/matures they will be very profitable businesses.\nFor the most software-defined-oriented SaaS firms the land and expand is much less effort – just think ASAN, NET, DDOG, OKTA – but at the same time competition is intense. So, the SaaS firms with the best products look to capitalize on their technological edge with huge opex to win the new logos, knowing that retaining and growing their wallet will be the easier and the more profitable part. And naturally, customer service is of a very high quality because otherwise customers will leave before the SaaS companies has recouped its S&M costs.\nThis contrasts with the software license model whereby the high contribution margin stage is the beginning of the customer relationship, and as a result may not motivate software companies to have their finger on the pulse so much in regards to customer service, and in general not be as focused in nurturing long-term relationships.\nBelow we speculate the generalized peaks and troughs of sales and customer service effort/expense and customer satisfaction for the license model and SaaS. And from these diagrams it's easier to grasp why SaaS and customer LTV go hand-in-hand – SaaS is just more conducive to consistency and long-term customer satisfaction.\nFigure 3 - Comparing Software License Model to SaaS Model: Fluctuations in Effort/Expense & Customer Satisfaction\nSource: Convequity\nThe above diagram also illustrates why it's worth it for SaaS firms to spend more in opex today than license-based software firms did in the past. Higher customer satisfaction equals more profitable long-term relationships. And it’s the delivery and usability of SaaS products that enables this higher customer satisfaction. We’ll also like to add that many SaaS firms are treating their huge opex as the new capex. Indeed, the economic benefits of much of the opex occurs long after 12 months, so in our opinion, it’s time accounting standards changed to reflect this by shifting the expense from income statement to cash flow statement. NYU Corporate Finance Professor Damodaran is a prominent advocate for changing the accounting standards to better reflect economic reality.\nBesides pure S&M expenditure, SBC is a big contributor to opex, distributed across all opex categories. But it’s kind of clear as to why SBC has become so important for leading software companies. Due to deep-seated issues pertaining to education curriculums and the rapid pace of software innovations in the past decade – thanks to things like the cloud, containers, and microservices – a huge skills shortage across IT has emerged. One of the most prominent shortages is that within the cybersecurity industry, though there are many other software sectors experiencing this demand/supply mismatch. Hence, there's a fight to win the best talent within a limited pool and SBC is one of the biggest incentives to use. It also improves alignment of long-term interest, among employees, management and the shareholders.\nHowever, it looks as though the significant portion of the colossal SBC generates poor returns. The following table shows that there is a moderate correlation [0.39] between SBC and revenue growth. And the differences between the lower, median, upper quartiles of the distribution of 95 stocks indicates there is a moderate correlation that wepresumeis a cause-and-effect relationship. But given the aggression in SaaS companies SBC policies, one would assume the revenue growth, and as a by-product, ROI, would be more likely – the r-squared indicates that SBC only explains 15% of revenue growth variance.\nFigure 4 - Moderate Correlation between SBC and Growth\nSource: Convequity\nWe think this highlights the struggles that SaaS firms are encountering. They are forced to offer high SBC to attract limited talent but the resulting business performance is far from probable. And, of course, this is because great SaaS firms have a bunch of characteristics needed to make them a success – such as product led growth, great leadership, empowering culture, a huge freemium base [e.g., NET, DDOG, ASAN], the use of open protocols, software-defined approaches, software design, CI/CD, and microservices architectures. So, as investors there is a lot of opportunity in identifying the truly BoB SaaS players that have these characteristics to ensure a good ROI in SBC is generated to avoid a net negative impact on shareholder value.\nFrom the 95-stock peer group shown above, we compared four observability players’ SBC % and TTM revenue growth – DDOG, ESTC, DT, and NEWR – shown in the following table. We would strongly surmise that DDOG has plenty more of the aforementioned characteristics than the other three. So, it makes one wonder why the management, of say NEWR, will decide to pay out 21% of revenue in SBC before they’ve addressed their fit with the aforementioned characteristics and other leading SaaS attributes. The SBC above and beyond the market rate is likely to be a waste of investment because SBC alone is not enough.\nFigure 5 - Leading Observability Players: SBC % vs TTM Growth\nSource: Convequity\nIt’s also noteworthy that as cloud has lowered the entry barriers for software companies, more opex is necessary to flourish in a more competitive environment versus pre-cloud eras. As we show in the forthcoming data, there are double the amount of public software companies today compared to 2007. And this is in the context of SaaS reducing vendor lock-in thereby putting even more pressure on companies to keep customers happy and loyal in order to maximize the customer LTV potential. It’s very much a winner-takes-all in the majority of SaaS markets, which is why a big component of our investing approach is to invest in the BoB SaaS vendors at a reasonable price.\nIn statistical parlance, the hypotheses were as follows:\n\nGross margin\n\n\n\nH0: median gross margintoday and median gross margin in 2007 are the same.\nH1: median gross margin today and median gross margin in 2007 are different.\n\nOperating margin\n\n\n\nH0: median operating margin today and median gross margin in 2007 are the same.\nH1: median operating margin today and median gross margin in 2007 are different.\n\n\nIf the gross margin H0 [null hypothesis] is rejected, meaning H1 is true, then perhaps this is causing the richer valuations today. Companies with higher gross margins have better core economics within their business and hence long-term, mature-stage profitability would probably be comparatively higher. Could this potentially and/or partly explain the current valuations?\nIf the operating margin H0 is rejected, then the notion of how the cloud and SaaS is affecting opex could partly explain the relatively lower earnings for software stocks versus earlier periods.\nIn essence, if our theory can be rationalized then higher gross margins are pushing up the numerator of P/E [and other ratios] and lower operating margins are pushing down the denominator of P/E [and other ratios].\nAs a sidenote, there is also a strong argument that this depressing of earnings is unwarranted because a lot of S&M expense, as earlier touched on, is usually generating economic benefits [revenue] for longer than 12 months – and the same as with R&D. Therefore, in our view accounting standards are long overdue a change to better reflect the economic benefits of expenditures, and if/when this occurs it would have a big impact on the multiples were witnessing today [not the valuations though because opex is just shifting to capex] as such expense will be transferred from the income statement to the cash flow statement. Though, we’re not confident such change will happen anytime soon – it could easily take another decade. Just think how long it took to change lease accounting. Anyway, we believe there is a solid argument that today’s higher multiples for software stocks can bepartlyjustified by the pervasiveness of the cloud and SaaS business.\nThe caveat, however, is that justification of today’s multiples will be considerably higher for those SaaS stocks that represent BoB characteristics and are not throwing away SBC. And this loops back to the winner-takes-all idea within SaaS markets – multiples are stretched across the board but likely justified for those stocks that will win their market, carve out a long runway of growth, and have opportunities to increase margins.\nRevenue Growth and Revenue Growth Decay\nAs previously pointed out, in our earlier research we only made hypotheses for gross margin and operating margin but this time we’re also predicting that TTM revenue growth is statistically higher and revenue growth decay is statistically smaller today versus 2007. In fact, for the revenue growth decay we used stocks from the 2011-2013 and 2019-2021 time periods and simply calculated the decay in growth – for example, Company XYZ growth in 2013 subtract Company XYZ growth in 2011.\n\nRevenue growth\n\n\n\nH0: median TTMrevenue growth today and median TTMrevenue growth in 2007 are the same.\nH1: median TTMrevenue growth today and median TTMrevenue growth in 2007 are different.\n\n\n\nRevenue growth decay\n\n\n\nH0: median revenue growth decay during 2019-2021 and median revenue growth decay during 2011-2013 are the same.\nH1: median revenue growth decay during 2019-2021 and median revenue growth decay during 2011-2013 are different.\n\n\nOur reasoning for the revenue growth hypothesis is that in using the hyperscalers IaaS, software companies can scale their business faster today than in yesteryear and hence can drive faster revenue growth. Software-defined-everything also enables SaaS firms to produce new products with more frequency which underscored this hypothesis.\nThe logic behind speculating that today’s software companies experience a smaller delta in revenue growth decay, or slower growth deceleration, is that SaaS/cloud-native companies are able to produce new products and enter new markets much quicker than in earlier pre-cloud periods [2011-2013 would be pre-cloud for the high majority of stocks selected]. Hence, we surmise that growth decay is less today.\nNET, ZS, TWLO, OKTA, and NOW are just a few SaaS firms that have demonstrated tremendous TAM expansion via prolific new product releases which have helped slow down growth deceleration. The extensible software design and architecture adopted by these SaaS firms enables them to avoid getting bogged down in technical debt. This is very different to earlier legacy approaches that entailed an application being tied toonecode base consisting of tens of thousands of lines of code with poor annotation/explanation linked to them. Each line of code is dependent on thousands of others, making troubleshooting a daunting task, and may result in the application being out-of-action for 24 hours at least. Just imagine the reputational and financial damage a top software firm would experience if their app went offline for such a time period.\nLegacy firms are often aware for the need for change to keep up with the fast-paced software industry but are stifled into inertia because of the monolithic structure of their apps. Changing one line of code may bring the entire application offline and troubleshooting a 100k line code base is painstakingly difficult to do. So, just imagine the feeling of dismay when facing the need to radically alter the app and how it's delivered. Often the pain is too much to bear and just quick fixes and improvements will be opted for instead.\nThis cycle continues until the once-leading software firm is now categorized as legacy and revenue growth is diminished to low-single-digit or even flat. CHKP and JNPR are two cases in which we strongly surmise this is the case. Many software firms become victim of this scenario because of a lack of leadership whereby the company is not led by the innovative founder, or are led by a conservative founder [like in CHKP’s case]. Consequently, very often a new owner is brought in like STG [Symphony Technology Group] that privatizes the business with a predominant focus on profits over innovation – and then there is certainly no return to the cutting edge.\nIt really is easier to start from scratch. Computer chip guru,Jim Keller, advocates doing just this every 3-5 years – starting again with a blank canvas, a completely clean slate, even a new company altogether. Otherwise, the change is too painful and the risk is too high. But how can a large software company manage this whilst protecting its revenue and profitable cash flows? It’s nigh-on impossible with old software architectures. It requires a combination of modern architecture, modern DevOps, and modern organizational processes, to combat the loom of legacy in order for software firms to S-Curve stack their way to being perennial pioneers. It appears the likes of the FAANGM stocks as well as CRM and NOW have this blend of attributes to make sure they stay on the steep part of the S-Curve.\nConsider when Eric Yuan [CEO and Founder of Zoom] wanted to radically improve upon the WebEx meeting software, but he couldn’t persuade CSCO executives who didn’t want to risk its high cash flows for a cloud-native project with uncertain visibility into the future. So, it wasn't until 2020, when ZM overtook WebEx as the market leader, that CSCO realized how poor its solution actually is. Now, CSCO is in crisis mode to try and claw back some market share but it’s probably already far too late. Such an example, that happens frequently, illuminates the difficulty for software firms to stay modern.\nDesigning software with built-in extensibility is one approach modern SaaS firms are adopting to avoid the legacy fate. Fully utilizing software-defined approaches, CI/CD, and microservices architectures are other intertwining approaches to enable them to improve their software with less stress and be adaptable to future industry developments. They aren’t legacy-proof but will help them maintain their BoB status, continue to deliver the best software, and slow the growth decay and maintain double-digit growth for many years. Aside from the previously mentioned SaaS stocks, we would suggest the likes of DBX, FSLY, JAMF, SUMO, and IS also design their applications in a way that fits into the description of extensible software.\nFigure 6 - Application Architectures: Monolithic vs Microservices\nSource: n-ix.com/microservices-vs-monolith\nThe other reason for postulating that today’s software firms suffer slower growth decay is because the required capex to move into new markets is minimal in comparison to the pre-cloud era because they don’t need to purchase and implement more servers. Once an application is developed and ready to be deployed, they can scale it across the elastic supply of AWS, Azure, Alibaba Cloud, or GCP servers that provide all the necessary compute, networking, and storage requirements – and ramp-up or decrease capacity according to demand. There have been several other industry developments that allow further specialization and outsourcing of the work without the need for one vendor to do all the back-end operations by itself. That includes services like colocation and better virtualization technologies that allow SaaS vendors to have confidential information running in untrusted physical locations.\nThe statistical tests we used to compare all the medians of the different stock groups are called the Mood Test and the Kruskal-Wallis Test. We also did median tests for FCF margin, the Rule of 40, and FCF/EV but we didn’t form prior hypotheses. The tests were all based on a level of 95% confidence.\nCriteria for Stocks in 2007 and 2021\nAs October has just passed, it made sense to use market prices and financials as of October 2007 so we could compare the 2021 and 2007 groups based on the same time of the year. And this worked out well because prior to GFC the Nasdaq 100 peaked on 31stOctober 2007. So, we’re comparing today’s software stocks to a cycle peak which seems conservative for the tests [we presumed historically high gross and operating margins, and high revenue growth in late 2007 just before GFC] and unfavourable for our hypotheses.\nFigure 7 - Weekly Chart of Nasdaq 100\nSource: TradingView, Convequity annotation\nSo, based on market price and financial data as of the end of October in 2007 and end of October in 2021 we set selection criteria using GuruFocus. In fact, we did the 2007 vs 2021 comparison based on two criteria sets.\n\nFirstly, we searched for U.S. headquartered software stocks with revenue above $100m and a market cap above $1bn for both 2007 and 2021 [as of end of October]. The lower bound for revenue and market cap was to exclude microcap stocks.\nSecondly, to refine the research we ran the search process again to look for U.S. headquartered software stocks with revenue between $200m and $2bn and a minimum market cap of $1bn for both 2007 and 2021. We thought it would be useful to compare 2007 and 2021 based on this revenue bracket because from our experience it seems as though most of today’s richer valuations tend to be assigned to stocks with revenue in this ballpark.\n\nWe did some data cleansing to remove the most extreme outliers and financials that didn’t make sense [like a 123% gross margin, for example]. And after some trial and error we found that searching for only U.S. headquartered companies removed some of the questionable companies and financials in both periods – hence the reason for this criterion outlined above.\nFindings\nAnnual Revenue Above $100m\nBased on the criterion of annual revenue above $100m, we found 70 software stocks in 2007 and 151 software stocks in the 2021 cohort. The gross margin and revenue growth hypotheses were not proven accurate [couldn’t reject the H0] but the operating margin and revenue growth decay hypotheses were proven accurate [rejected the H0]. The following table shows the results [note we didn’t test the means because of uncertainty of the presence of normal distributions but have included them for the readers perusal].\nFigure 8 - October 2007 vs October 2021: Annual Revenue Above $100m\nSource: Convequity\nWhilst not statistically significant, the 480-basis point higher median gross margin for the 2021 cohort is notable considering how difficult it is for companies to improve gross margins. Unlike margins further down the income statement, gross margins don’t vary much as they reflect the core economics and the core of a business model. When we run this test earlier this year, the gross margin differential was 850-basis points. We think the results from both our earlier and today’s research on gross margin warrants continued examination.\nGiven that companies are still rather early on in the cloud journey, it’s likely that the gross margin delta will be greater if tested in the future once companies have further optimized their operations in the clouds. It’s also important to consider what we mentioned earlier, in that to counter the diminishing effects of the multitenancy, software firms that have already scaled into a $1bn-$2bn revenue business will likely be better off switching to colocation data centers to drive up gross margins. So, if other SaaS firms follow the likes of DBX we think its probable gross margins will improve in the long-term.\nWhilst the Mood and Kruskal-Wallis tests proved us statistically accurate for the operating margin hypothesis, the increase in stock-based compensation [SBC] expense over the years does muddy the waters. Therefore, in future research we shall remove the SBC effects [originally, we did want to search for S&M expense only but we couldn’t retrieve this from GuruFocus]. However, as earlier mentioned, software firms have been increasing SBC to attract and retain top talent required to win customers [a S&M expense] and develop good SaaS products and features [a R&D expense]. So, there is a pretty strong case to claim that the rising SBC is also closely tied to companies aiming to maximize the customer LTV opportunity.\nWe were surprised in the equality between the 2007 and 2021 TTM revenue growth rates [for both mean and median]. There isn’t much of a counter here, our hypothesis is wrong. But it would be interesting to do the TTM growth rate comparison for two periods with the same business cycle stage. October 2007 was late stage whilst currently we’re in the recovery stage. Just food for thought for if we were to do this analysis in the future.\nThe differences [both the mean and median] in revenue growth decay look small but the median difference is in fact statistically significant according to the tests. As is shown, we also refined the revenue growth decay testing to just the top revenue growth quintile [based on 2011 and 2019 revenue growth rates]. The median difference is not significant but we would say it is non-negligible, and the difference in the means would probably be statistically significant had we tested it.\nFor both the revenue growth and revenue growth decay we feel like further research in the future is needed. But it does look like it’s possible that, whilst the cloud may not be enabling software companies to drive faster growth, it may be enabling them to maintain growth for longer.\nWithout formally declaring a hypothesis, we did anticipate the significant [and obvious] differences in FCF/EV. However, we thought that what we believed to be higher expected revenue growth in 2021 would have closed the difference in EV/FCF/Growth. In isolation, the EV/FCF/Growth does indicate today’s software stocks are very overvalued. Though we would contend that this metric doesn’t account for future growth [it’s based on TTM growth because we couldn’t retrieve forward growth estimates for 2007] and hence won’t reflect any view on slower growth decay in today’s environment. In other words, the metric doesn’t reflect the compounding growth of the future, and as the decay is smaller and the growth outlook is more visible and probable thanks to recurring revenue models, and the aforementioned ease of producing new products and entering new markets, investors may be attaching premiums to reflect this.\nAnnual Revenue Between $200m and $2bn\nBased on the criterion of annual revenue between $200m and $2bn, we found 45 software stocks in 2007 and 93 software stocks in the 2021 cohort. Aside from selecting this criterion to focus on the richest software valuations, companies in this revenue bracket tend to experience the fastest growth and are BoB software companies that are rising to be industry-defining leaders – and these types of stocks feature in a sizeable portion of our portfolio.\nAs shown in the following table, the results are similar to the previous criteria set but the gross margins are pretty much equal and the difference in revenue growth decay is not statistically significant. So, on the whole our set of hypotheses are less accurate for this revenue bracket. However, the median revenue growth rates do show a meaningful difference.\nFigure 9 - October 2007 vs October 2021: Annual Revenue between $200m and $2bn\nSource: Convequity\nWe are really surprised that there isn’t a notable difference in the median gross margins. And we’re trying to avoid coming up with what may seem as excuses and remain objective, but it is possible that the pandemic-induced remote working environments of 2020 increased SaaS-based cost of revenue percentages and has thus lowered the TTM gross margin for 2021. Not all SaaS-like firms experienced this but quite a few did, like ZS and RNG and other per-user [not usage based], cloud-native SaaS firms that position themselves in the middle of traffic and are thus sometimes victim of higher bandwidth costs amid sudden surges in demand. So, it will be worth running this gross margin testing for this revenue bracket again in the future.\nThe operating margin differential is much wider than the previous group which could be explained by this revenue bracket containing more of the aggressive companies, and this greater aggression manifests itself in higher opex in the form of S&M and SBC.\nThere is no statistical significance in the difference between the medians for revenue growth decay and the same for the top quintile. Although, the differences are non-negligible, especially for the top quintile. From this we infer that the highest growers in the 2019-2021 period have been able to better minimize the growth decay than software companies achieved in the 2011-2013 period. These highest growth rate stocks are invariably the most cloud-native and hence they can produce new products and enter new markets with speed – just think of the likes of NET and DDOG – so the results here come as no surprise.\nThe FCF/EV and EV/FCF/Growth results draw similar reasoning/conclusions to the previous [TTM revenue > $100m] group, but they are just more extended for the 2021 cohort. Indeed, the median 7.05 EV/FCF/Growth for the 2021 group is way higher than the 1.58 for the 2007 group, however, as we mentioned earlier this metric doesn’t account for future compounding growth. We’re not justifying the 7.05 but is the market implicitly increasing growth forecasts and extending explicit forecast periods in DCF valuations to factor in the slower growth decay in today’s environment? If so, then this may be boosting the numerator andpartlyexplain the differential.\nWithin the context of the winner-takes-all idea that we’ve highlighted in this article, we view the hyper-extended multiples, especially in this $200m-$2bn revenue bracket, as big bets by investors that they know who the winners are going to be. This in stark contrast to more traditional investing predicated on forecasting a company’s cash flow generation over the next five years which is not necessarily about picking winners but how the company in question can improve its business over time. Both are valid but very different approaches, and rather than outright say paying 60x P/S for a SaaS stock can be justified, here we are merely trying to find possible reasoning behind current market behavior rather than taking the easy view that the market is in a bubble.\nWith the winner-takes-all idea in mind, not all these stock multiples can be justified because most will not meet long-term expectations. Again, this is why an understanding of a software company’s technology, culture, application design, etc., in our opinion, is important to identify the winners and not just rely on an extrapolation of current growth.\nGross Margin Distribution Comparisons\nAs we contemplated the difference in median gross margins being statistically insignificant, but in light of the stickiness of gross margins and hence feeling that the differences are noteworthy, we decided to look at the distributions. The 2007 and 2021 distributions for the first selection [revenue above $100m] and the second selection [revenue between $200m and $2bn] were similar, therefore, for brevity we’ll only show the first selection.\nThe first chart below shows that the 2007 distribution looks more evenly spread whilst the 2021 distribution has a high peak in the 70%-75% gross margin level. However, the difference in the number of stocks selected [70 for 2007 and 151 for 2021] makes any conclusion difficult.\nFigure 10 - 2007 vs 2021: Gross Margin Distribution Comparison\nSource: Convequity\nThe next chart shows the gross margin proportions across each distribution. We can see that almost 17% of the 2021 cohort had gross margins in the 70%-75% bucket. This contrasts with only 6% of the 2007 cohort that had gross margins in the 70%-75% bucket. This is certainly a big difference that illuminates the positive impact of software firms utilizing IaaS. And from our experience in analyzing cloud-native/SaaS stocks this does make sense as this gross margin bucket tends to be common, especially for firms that are achieving scale.\nFigure 11 - 2007 vs 2021: Gross Margin Distribution in Percentage Terms Comparison\nSource: Convequity\nBased on the median gross margin differences [not statistically significant but notable] and the disparities in the distributions highlighting a boost in the proportion of stocks achieving 70%-75% gross margin, we believe future research is warranted. It also opens up the conversation that perhaps today’s software companies, or leading software companies, have better underlying economics and prospects for long-term, mature-stage profitability than software companies did in pre-cloud eras. If future research provided stronger evidence that this is the case, then the currently rich valuations [often construed as grossly overvalued] could be partly justified.\nOur last note on gross margins, is that as touched on earlier, gross margin levels are very dictated by the type of software sector. For example, cybersecurity companies rarely have gross margins above 80% because of the high-power computing needed to decrypt and re-encrypt traffic flowing through inspection points. As we discussed earlier, productivity companies like ASAN and MNDY have gross margins close to 90% because a lot of work the software does is storing and retrieving text-based information. Such cost of revenue differences between different software sectors, such as cybersecurity vs productivity, can be attributed to differences in information density. In high information density software like productivity, the entire list of tasks might take less than 1MB to store yet the daily inspected data for cybersecurity firm PANW could be more than 1GB per seat per day.\nTherefore, in future research we may look into controlling the software sector variable to test differences within specific sectors. For example, if there is a higher proportion of cybersecurity firms in the 2021 cohort versus the 2007 cohort, or vice versa, it may obscure the true results.\nConclusion\nAcross the testing we believe there is enough evidence to at least warrant further research by ourselves and/or others to get closer in determining or dismissing that software stocks are in a new paradigm.\nIn the broader software selection represented by all U.S. headquartered stocks with revenue above $100m and market cap above $1bn, our set of hypotheses are supported by either statistical differences or notable/meaningful differences – statistical differences were operating margin and revenue growth decay and the notable difference was gross margin. We found very similar results for the gross margin and operating margin differences in the research we published earlier in the year, which somewhat encourages us with the line of thinking outlined in this article.\nThe narrower software selection that honed in on the richer valued stocks, represented by all U.S. headquartered stocks with revenue between $200m and $2bn, was less supportive to our set of hypotheses. The only supportive statistical difference was the operating margin, but revenue growth and revenue growth decay differences were notable.\nAs expected, the FCF yield [FCF/EV] was statistically significant for both selections, though the growth-adjusted EV/FCF multiple was surprisingly much higher than anticipated in 2021.\nOn the whole, based on the testing results, we believe there is a moderate argument that today’s software firms’ use of IaaS and SaaS business models is helping them achieve increasing gross margins. We also think it’s possible that the cloud/SaaS is enabling higher revenue growth [for certain revenue brackets, i.e., the $200m to $2bn bracket] and slower revenue growth decay, despite not all the statistical tests not supporting this.\nAnd if some of today’s software firms are experiencing higher gross margins, higher growth, and slower growth decay compared to earlier periods, then higher long-term profitability and more sustained growth runways will be realized. And if this is being acknowledged by the market then perhaps this can explain the current valuations.\nIt seems as though higher opex is required in order for today’s software firms to be future winners amid the greater competition, capture the lucrative customer lifetime values up for grabs, leverage the relatively higher gross margins afforded by cloud computing, and mature with greater profitability than software firms of the past.\nIn effect, many software firms are trading short-term profits for long-term winner-takes-all dominance. Software has become more competitive but the prizes are bigger, necessitating huge upfront costs that current accounting standards deem should be included in the income statement despite the economic rewards of such costs lasting far longer than the next 12 months.\nThe higher opex is powering future growth for firms to capture the huge TAMs [Total Addressable Market] that have become available thanks to cloud computing, but in the present it is depressing earnings and extending [income statement-based] multiples and thus making valuations look stretched by traditional metrics.\nIndeed, many of the extended stock multiples cannot be justified because not all will materialize as long-term winners and hence expectations will not be met. We view the broad extended multiples of the software sector as big investor bets on who these winners will turn out to be – but only a portion will be right. This is in stark contrast to the more traditional type of investing entailing the forecasting of a company’s next 5-year cash flows that are not subject to it becoming the market’s winner as such.\nTo summarize all this in two brief sentences. We believe this research offers a moderate argument that today’s software firms have higher gross margins and slower growth decay that present investors with more profitable, sustainable business growth than software firms in pre-cloud periods. To capture this bigger prize among fiercer competition, firms are substantially ramping up opex which is depressing earnings and stretching multiples – some of which can be justified [the market winners] and others which cannot [the runners-up or losers].\nAll we’ve shared in this article is just our opinion on interpreting the data – we’re aware that we might be wrong. Hopefully it can at least spur on some discussion to help us refine or question these theories.","news_type":1},"isVote":1,"tweetType":1,"viewCount":225,"commentLimit":10,"likeStatus":false,"favoriteStatus":false,"reportStatus":false,"symbols":[],"verified":2,"subType":0,"readableState":1,"langContent":"CN","currentLanguage":"CN","warmUpFlag":false,"orderFlag":false,"shareable":true,"causeOfNotShareable":"","featuresForAnalytics":[],"commentAndTweetFlag":false,"andRepostAutoSelectedFlag":false,"upFlag":false,"length":128,"xxTargetLangEnum":"ZH_CN"},"commentList":[],"isCommentEnd":true,"isTiger":false,"isWeiXinMini":false,"url":"/m/post/875652332"}