AI Sells Labor, Not Software — Legendary Investor Elad Gil
In a Nutshell
Elad Gil argues AI's biggest shift is enabling companies to sell labor and work product rather than software licenses, unlocking markets like law firms that previously resisted traditional SaaS. The current environment favors consensus over contrarian bets—scaling foundation models is the highest-probability path because AI has opened many markets and created unprecedented openness to experimentation. Market quality matters more than founder quality; strong teams fail in closed markets, while "why now" signals like regulatory changes or new technology capabilities determine which opportunities are viable.
These notes were generated by AI and may contain inaccuracies.
Investors often begin with a set of beliefs about what is interesting or viable. After investing or starting a company, the original thesis can prove much harder than expected, leading to failure. Years later another company succeeds in the same space. The key question is what changed: technology may have improved, regulations may have shifted, markets may have moved, or other external factors may have altered the landscape.
An example is Harvey AI in legal services. Selling to law firms has traditionally been viewed as a poor business. Harvey achieved strong enterprise adoption by shifting from selling tools or software seats to selling work product and units of labor. Generative AI enables companies to sell human labor equivalents—work hours or cognition—rather than traditional SaaS licenses. This fundamentally different product opened markets that were previously closed.
Two competing schools of thought exist on what drives big company creation. The Y Combinator view holds that insufficient founders are the bottleneck; ten times more founders would produce ten times more large companies. The alternative view emphasizes that only a limited number of markets are open to innovation at any moment. Without a market shift or openness to change, building large companies is difficult regardless of founder quality.
AI has opened numerous previously closed markets through new capabilities and because every CEO is actively seeking an AI strategy. This creates unprecedented openness to experimentation. AI companies that fail to achieve rapid growth likely have fundamental product or execution issues, because the current environment allows faster scaling than historically possible. OpenAI and Anthropic reached tens of billions in valuation at unprecedented speed.
When a strong team enters a poor market, execution becomes very difficult. Market assessment can draw on several signals. The "why now" framework examines what has recently changed: regulatory shifts, technology improvements, competitive events, or incumbents slowing down after acquisitions.
Regulatory examples include fleet management companies benefiting from new in-cab driver monitoring requirements. Technology shifts are exemplified by foundation models instantly accessing enterprise data, email, documents, and all white-collar work because language underpins nearly all such activity. Code itself is treated as a form of white-collar work that AI can transform.
Robotics differs because even advanced models lack immediate large commercial submarkets unless new hardware categories are created. Competitive shifts also matter: when HashiCorp was acquired by IBM, the resulting slowdown created openings for competitors such as In-Q-Tel in the security space. Market velocity can also surge suddenly, as seen with Coinbase and crypto token proliferation.
Founders sometimes present inflated market sizes by referencing broad categories they do not actually serve. A company building an optimization engine for SMB websites cannot legitimately claim the entire global e-commerce market. Accurate market definition is essential for realistic ambition and strategy.
Coca-Cola's historical pivot illustrates the power of redefinition. After decades of measuring share against Pepsi in soda, leadership reframed the market as all beverages. Market share dropped from roughly 50% to 0.5%, prompting acquisitions such as Dasani and expansion into new drink categories.
Certain past AI debates, such as whether capex spend on training will generate sufficient ROI, appear increasingly outdated. Timing matters: there are periods when contrarian bets are wise and periods when following consensus is the highest-probability approach. The current AI environment favors consensus—pursuing the obvious path of scaling AI capabilities—over elaborate contrarian strategies such as pivoting heavily into hardware.
Keep Tim Ferriss in your library
Save the videos and channels worth coming back to, and find them again in one place.
