Snowflake

The Crumbling Terminal Value of Traditional SaaS For decades, software as a service (SaaS) stood as the ultimate business model. Investors treated these companies like high-yield annuities—reliable, recurring revenue streams with impenetrable profit pools. The market assumed these cash cows would churn indefinitely. That certainty has evaporated. We are witnessing a fundamental breakdown in the public-private boundary because the AI wave forces us to question the terminal value of existing software. When coding models from Anthropic and OpenAI can replicate complex workflows or automate the maintenance of legacy code, the 'insurance company' stability of SaaS disappears. This shift isn't just theoretical. It is hitting public market caps with brutal force. Investors are walking away from the sector because they cannot distinguish between the winners and the victims. If a design tool can be replaced by a prompt in ChatGPT, why hold the stock? The leading indicators we once relied on—sequential revenue growth and net new ARR—are now lagging indicators. They tell us what happened three months ago, but in a world where technology cycles move faster than an earnings call, the past is a poor predictor of survival. The market is effectively clearing the decks, exiting SaaS positions to find refuge in consumer internet or semiconductors while the dust settles on terminal value. The New Math of Platform Companies and Mega-Funds A decade ago, today's private giants would already be public. Companies like Revolute, SpaceX, and Open Evidence are staying private longer, choosing to scale within the venture ecosystem rather than facing the quarterly scrutiny of public analysts. This has birthed the 'Platform Company'—entities with multiple product lines, massive scale, and growth rates that exceed 30% even at billion-dollar revenues. For those of us in venture, this is the greatest gift. It allows us to capture the bulk of a company's value creation before it ever hits the New York Stock Exchange. This transition has also fundamentally changed the math for mega-funds. A $5 billion growth fund can only generate venture-like returns if it remains concentrated. The 'spray and pray' approach is a death sentence at this scale. You must identify the four or five companies that generate 65% of the entire market’s enterprise value. If you can deploy $1 billion into a single round and see a 10x return, you’ve doubled your fund. The outcomes in the AI era are potentially much larger than the SaaS era because we are moving from augmenting human labor to replacing it with tokens. When you address the labor market directly, the TAM isn't just a software budget; it’s the global GDP of human effort. Market Pull and the Founder’s S-Curve I often get asked what matters more: the founder or the market. It’s a trick question, but if forced to choose, market size wins every time. A phenomenal founder in a small, rigid niche will build a good business, but they won't build a $100 billion empire. You need a market that is actively yanking the product out of your hands. We look for 'Market Pull'—a revenue curve that doesn't just grow but screams. This is the difference between an act-one success and an enduring institution. However, the founder is the one who navigates the S-curves. Look at Ali Ghodsi at Databricks. He didn't just build a data transformation layer; he reinvented the company multiple times to stay at the center of the enterprise data stack. Most founders get comfortable after their first win. The truly elite founders have a 'talent density' and a restless vision that allows them to hop from one technology wave to the next. In our world, valuation is the last question we ask. If a company is growing 50x year-on-year, any entry price looks cheap in twelve months. The real risk isn't overpaying; it's missing the horse that has the stamina to run for a decade. Rethinking Margin and the Cost of Innovation There is a lot of noise about margins in AI. The purists argue that if it isn't 80% gross margin, it isn't software. They are missing the forest for the trees. Margin matters at scale, but early on, it is a misleading indicator. During an architecture shift, the best businesses often have horrific margins. Snowflake and the hyperscalers were low-margin early because they were building the infrastructure of the future. In AI, the cost of inference is plummeting. Today’s negative margin is tomorrow’s profit pool as token costs descend. We are substituting lower gross margins for significantly lower operating expenses. A lean engineering team using AI tools can replace a massive legacy workforce. Your terminal operating margin—the real bottom line—may actually be higher in this generation than the last. If customer behavior is sticky and retention is high, you can afford to be fragile on margins in the early days. The fragility only becomes fatal if you lack product-market fit. The Fallacy of Kingmaking The concept of 'Kingmaking'—the idea that a pile of capital from Coatue or Sequoia Capital guarantees victory—is a myth. Capital is an advantage, but it can also be a sedative. Too much money without product-market fit breeds complacency and waste. It makes companies focus on vanity metrics rather than the hard work of product iteration. Real power comes from optionality. Look at how Anthropic architected their business to be cloud-agnostic and chip-agnostic. They positioned themselves so that everyone wants them to win. They can take capacity on Google Cloud or Amazon Web Services while others are locked into single-provider bottlenecks. That isn't kingmaking; that is strategic brilliance. In a capacity-constrained world, the ability to deploy compute where others cannot is the ultimate competitive moat. Lessons from the Masters: Data as a Prerequisite Reflecting on my time with Mary Meeker and Mamoon Hamid, one lesson stands out: data is a prerequisite, not the answer. You must be able to express a complex company in a few lines of Excel, but you cannot live in the spreadsheet. Mamoon Hamid is a master at identifying the 'kink' in the curve—the moment a company shifts from linear to exponential growth. He saw it with Figma when they had only $500k in ARR because the usage curves at companies like Square and Google were undeniable. If you want to survive as an investor or a founder, you have to get off the linear path. The safe route is an illusion. The real returns come from the calculated risks—the 'unknown unknowns' that others are too afraid to back. Whether it's OpenAI moving into consumer hardware or Harvey disrupting the legal profession, the winners will be those who embrace the chaos of this transition and build for the $100 billion outcome.