The End of Software?

The Angle Issue #227

The end of software?
David Peterson

If you follow the public markets at all then you probably read a bunch of hair-on-fire takes this past week about how the software market is in trouble.

And to be fair, public enterprise software companies did get hammered last week. Salesforce fell 15%, MongoDB fell 24%, UIPath fell 36%, just to name a few.

What’s happening? The simple explanation is that these companies all revised their revenue projections down. So of course valuations fell as a result!

That being said, this is a sector-wide issue, as highlighted by Chetan Puttagunta:

And if you compare performance to the past few years, you’ll notice that growth has indeed meaningfully shifted. As Tomasz Tunguz details here, the 25th, 50th, & 75th percentiles for public software company growth rates have all halved in the last 18-24 months.

So where did all the growth go? I think there are a few potential explanations. One is that a lot of growth was pulled forward during the pandemic. I’m not very convinced of this one for a variety of reasons, but you can certainly find examples where this is the case (e.g. Zoom).

Another is that we’re reaching the “top of the S curve” of software, as explained by Jared Sleeper in this piece. I think there’s something to this idea, but S curves exist for technologies, not business models. And so while it feels right as an explanation in general, I think it falls apart if you look at it too closely.

More convincing is the argument that the index is just full of huge companies that went public years ago and are, understandably, growing more slowly as they reach saturation (e.g. how fast do we expect Salesforce to grow at 150K customers and $35B in revenue?). It’s not that there are zero software companies growing fast. It’s just that the fast growing ones are all private. In other words, if companies went public earlier, public software companies would, on average, be growing faster, and my guess is we’d be having a very different conversation now.

My personal favorite explanation is simply that easy wins are getting harder to come by. Back in 2008, you could meaningfully improve a product by rebuilding it in the cloud. Not to say that was all you needed to do, but if you were aiming for that “10x” improvement over the incumbent, you could get a lot of that improvement for free if you pulled off the cloud transition. This predictability led to the playbook-ification of venture, or “entrepreneurship on autopilot,” as my partner Gil likes to say. And this dynamic has defined the startup world for the past decade.

Today, without the wave of the cloud to ride, everything feels a bit more incremental. You need a meaningful edge in either technology or distribution to break through. And I think that fact is starting to catch up with a lot of companies, both public and private. How many companies are facing growth challenges simply because their product is better, but not that much better? How often are buyers deciding to stick with their current vendor because the switching cost is just too high and the benefit isn’t there?

For what it’s worth, that’s why, these days, I’m so focused on talking to founders with ideas that truly break the mold. I want to spend time with founders who have ideas that are so out there they take me a few meetings to understand. I want to go deep on sectors that investors have all long since ignored. The cloud transition tricked us all into believing that if you built something that looked more or less the same as what came before, but was just a bit better, that’d be enough to win. No longer.

All of which is to say, software isn’t over. And maybe there are some private companies waiting in the wings that will show public market investors what growth looks like once again. But it's my bet that the next great public company, even if it’s software, won’t look anything like what came before.



Revenue Durability in the LLM World
Everything about LLMs seems to make revenue durability more challenging than ever.

A Digital Fabric for Maritime Trade
Why we invested in Portchain.

Three Keys to the Kingdom
The sometimes-competing and sometimes-aligned goals that early-stage founders must manage.



Enterprise software malaise. The biggest news in tech this week seems to be a widening sense of capitulation around enterprise software valuations. Jamin Ball wrote that software sentiment had crumbled. There are concerns around AI, but the main issue seems to be a slow down in growth in general. As Jamin writes, “growth hasn’t rebounded - there was no v shaped recovery. In fact, forward estimates are starting to go down, not up. Sure, the strong headwinds have abated, but the headwinds haven’t been replaced with tailwinds. Everyone is still waiting for growth to “re-accelerate,” and a number of companies are implying this will come in the back half of the year, but will it? Feels like investors are broadly more comfortable waiting for the data to show up. What’s more likely is the current state we find ourselves in is likely to be the equilibrium.” Anita Ramaswamy, writing in The Information, argues that the selloff and the collapse of sentiment is overdone: “The market’s panicked reaction may be overdone. It’s far too early to judge how much companies will shift their spending from existing enterprise software firms to new AI services, and when. Given questions about large language models’ accuracy and cost, which have caused many businesses to hold off spending on AI-powered services so far, the near-term impact on software spending may be minimal. And, of course, software companies such as Salesforce are adding AI tech to their existing products in an effort to cash in on the AI boom. That complicates the issue, as some software believers think AI can help drive growth for older companies, while others think it will kill them. Whatever happens, the shift is likely to take years.” Jared Sleeper pointed out that software is just naturally growing slower as it goes logarithmic. “As bad as all of that sounds, there is nothing to fret about here per se over the long run. It isn’t surprising that an adjustment away from behaviors learned over two decades. It is actually an altogether more benign explanation for the current malaise than some of the others circulating. In this scenario, SaaS still has a healthy decade of well-above-GDP growth ahead of it (without considering AI’s impact). There are still plenty of exciting startups which for one reason or another are less mature than the sector overall and merit investment levels that harken back to headier days. We're experiencing a normal adjustment period of a disruptive, exponentially growing technology- not an existential crisis.” Tomasz Tunguz also addressed these concerns, pointing out that “In the last 12 months, public software companies will grow on average 17%, which will add $100b in revenue across these businesses.”

Vertical integration in AI. Ben Thompson at Stratechery took apart the varying approaches of key major players in the AI universe: Google, AWS, Microsoft, Meta, and Databricks. In his analysis, the key differences can be explain by which parts of the canonical AI stack (model, data, storage, cloud, chip) they chose to vertically integrate. Google, for example, has taken a vertically integrated approach designed to help them offer what they hope will be differentiated consumer services. AWS, by contrast, is betting on the commoditization of most of the stack. Meta is betting on open source. Nvidia and Databricks are focused on specific layers of the stack, and betting that customers (large and small) will build their own models on top. Here is Thompson’s critical conclusion: “LLMs are already incredible, and there is years of work to be done to fully productize the capabilities that exist today; are even better LLMs, though, capable of disrupting not just search but all of computing? To the extent that the answer is yes, the greater advantage I think that Google’s integrated approach will have, for the reasons [Professor Clayton] Christensen laid out: achieving something approaching AGI, whatever that means, will require maximizing every efficiency and optimization, which rewards the integrated approach. I am skeptical: I think that models will certainly differ, but not in a large enough way to not be treated as commodities; the most value will be derived from building platforms that treat models like processors, delivering performance improvements to developers who never need to know what is going on under the hood. This will mean the biggest benefits will accrue to horizontal reach — on the API layer, the model layer, and the GPU layer — as opposed to vertical integration; it is up to Google to prove me wrong.”

Agentic RAG. A deep dive on agentic RAG by Akash Takyar at LeewayHertz. The term is buzzy but seems largely on-point as it captures both the technical underpinning (RAG) of the current wave of AI applications as well as the key benefit of many of these applications: their ability to act as intelligent agents in service of a predefined goal based on a combination of deterministic and unstructured data. “While standard RAG excels at simple queries across a few documents, agentic RAG takes it a step further and emerges as a potent solution for question answering. It introduces a layer of intelligence by employing AI agents. These agents act as autonomous decision-makers, analyzing initial findings and strategically selecting the most effective tools for further data retrieval. This multi-step reasoning capability empowers agentic RAG to tackle intricate research tasks, like summarizing, comparing information across multiple documents and even formulating follow-up questions -all in an orchestrated and efficient manner. These newfound agents transform the LLM from a passive responder to an active investigator, capable of delving deep into complex information and delivering comprehensive, well-reasoned answers. Agentic RAG holds immense potential for such applications, empowering users to understand complex topics comprehensively, gain profound insights and make informed decisions.”

The real impact of LLMs on financial analysis. My first boss and unparalleled student of technology Michael Parekh took a deep look at how LLMs might impact his home turf: financial analysis, and concluded that they will be disruptive but - as with tech waves past - will probably unleash a wave of creativity and productivity. “Having been a professional sell-side analyst on Wall Street for a very long time, I can personally relate to the trepidation in the industry given these AI trends. But at the same time, I’m confident that these AI capabilities will meaningfully augment AND expand the financial analysis envelope far more than might currently be imagined. Just as spreadsheets in the eighties spurred the exponential expansion of the financial industries around investing in every asset class imaginable, I can’t help but think these nascent AI capabilities are likely poised to have the same effect.”

AI coding tools. Corinne Riley of Greylock put together an outstanding overview of AI-powered coding tools. First, she identifies three approaches taken by these companies: 1. AI copilots and chat interfaces to enhance engineering workflows; 2. AI agents that can replace engineering workflows; and 3. Code-specific foundation models. Next, she identifies and probes three critical questions, the largest of which is: “Does owning the model and model infra lead to a long-term differentiated product?” Here is part of her answer: “The key question here is whether the rate of improvement of the base models is larger than the performance increase from a code-specific model over time. I think it’s possible that most copilot companies will start taking frontier models and fine-tuning on their own data – for example, take a Llama3-8b and do RL from code execution feedback on top of that – this allows a company to benefit from the development in base models whilst biasing the model towards code performance in a very efficient way.”


Are you playing to win? David Kellogg, an EIR with Balderton and a thoughtful writer on entrepreneurship, highlighted the stark contrast between “playing to win” in a market and “playing to make plan.” “While there certainly remains a large winner-take-all world (e.g., AI) that dominates most Silicon Valley thinking, there is now also a parallel, more mundane world. The danger is when strategic thinking designed for one gets applied to the other.” Kellogg defines both modalities in detail and then highlights some risks (that playing to make plan can either lead a company to miss key changes in its market or become dependent on near-perfect execution as baseline growth rates shrink). Then it gets really interesting: “One interesting case is when an entire market goes into “make plan” mode. I’d argue that my old category, enterprise performance management (EPM), is largely there. All the major independent players are owned by PE firms and presumably more focused on making plan than on winning in the market. Thus, innovation has slowed. The level of competitiveness has diminished (also helped greatly by Workday’s acquisition of Adaptive Insights, a well-funded, aggressive, price-slashing competitor back in the day). It strikes me today as a sleepy category with a bunch of PE-owned firms all grinding out their “make plan” goals, hoping to get sold for 3x+ their invested capital in the coming years. What happens then? Alas capitalism works. There is now a crop of VC-backed startups like Cube trying to fill the Adaptive void or Mosaic in financial analytics. France’s Pigment is the most aggressive grower and capital raiser in the space, but more focused on mid-market and the Anaplan void in enterprise. But there is a short answer to the question, what happens when the whole category ends up PE-owned and focused largely on “making plan?” A crop of new startups enter to seize on that opportunity. (And see my disclaimers as I’m working with several of them in different ways.)”

A throwback to ICQ. With the final shutdown of ICQ looming after 30 years, Calcalist looks back at the early days of the company which - in many ways - helped give rise to Israel’s current startup ecosystem. “In June 1998, a contract was signed according to which [ICQ developer[ Mirabilis was sold to AOL for $287 million in cash and another $120 million were contingent on meeting the business goals. Mirabilis, which was then only two years old and had some 70 employees, most of whom worked there less than a year, landed a $407 million deal. That was the first Israeli exit and it even turned the very term “exit” - thus far spoken only in elegant conference rooms of Wall Street investment banks or open spaces in Silicon Valley - into a legitimate Hebrew word. The pictures of Goldfinger, Vigiser, Amir, and Vardi in the traditional geek look - long hair, glasses, and t-shirts - adorned newspapers and ignited the imagination of tens of thousands computer aficionados. That was the revenge of the geeks. It was hard to believe that those young men, all in their early 20’s, were suddenly worth more than $100 million each.”


The great reset. Frank Rotman, CIO of Fintech VC QED, wrote that we are in the midst of a great reset of the venture capital industry, with huge implications for VCs, LPs, and founders. “The net result of this “Grand Reset” is that there’s no longer incentive for anyone to maintain the illusion they can shepherd mediocre companies towards billion-dollar IPOs that aren’t going to happen. Instead, the focus has shifted towards "landing the plane" for the 90% of companies that aren’t ever going to achieve escape velocity….This shift can be brutal for Founders who envisioned a triumphant IPO. But for many, it's a wake-up call. The pressure to "grow at all costs" has receded, replaced by a need to focus on building sustainable businesses with real revenue models and clear paths to profitability.”

The best are doubling down on Israel. Both Greylock and Sequoia recently announced that they were re-opening their Israel offices after a long hiatus. (Congrats Mor and Dean!) Both firms closed their Israel offices many years ago, believing they could cover the market from their US offices. While many firms do successfully cover Israel and Europe from their US HQs, the vitality and significance of the Israeli tech scene are definitely great enough to sustain a local office - and we are excited that such great firms have rejoined the local ecosystem with a team on the ground.


Tensorleap will be hosting a meetup on “Exploring AI Frontiers: Explainability in Multimodality and Unlabeled Data” at MIT Innovation Headquarters in Cambridge. Register here to join.

LightSolver CEO’s Ruti Ben-Shlomi recently discussed in Scientific Computing World how laser-based computing can solve some of the most challenging computational puzzles, such as image processing, credit scoring and feature selection in biological sciences.


Join the conversation

or to participate.