In a recent episode of 20VC x SaaStr, recorded on the last day of June, Harry Stebbings, Jason Lemkin, and Rory O’Driscoll highlighted a growing disconnect in the AI industry. According to the discussion, firms have quintupled their token spend during the first half of the year, yet nearly none can point to a tangible revenue boost justifying that expense. This theme dominated the conversation, underscoring a potential crisis in AI investment.
The data, shared by the panel, indicates a dramatic surge in token consumption—a key metric for AI model usage—without proportional financial returns. This gap between spending and value creation suggests that many companies may be over-investing in AI capabilities without effective monetization strategies. The episode did not provide exact dollar figures but framed this as a systemic issue across the sector.
This token ROI crisis comes amid broader market turbulence. The same episode noted that Microsoft experienced its worst month since 2000, reflecting heightened investor scrutiny on tech giants' AI-related expenditures. Meanwhile, Anthropic has called for a ban on Chinese open-source AI models, signaling geopolitical tensions in the AI space. These developments together paint a picture of an industry at a crossroads.
For startups and enterprises alike, the implication is clear: the era of unchecked AI spending may be ending. The pressure is mounting for demonstrable returns from token-intensive deployments, which could shift focus toward more efficient models or targeted use cases. Investors are likely to demand clearer metrics linking token consumption to revenue generation, potentially reshaping AI funding and product strategies.
The counterargument, however, is that token spending could be a leading indicator of future value—much like early web investments that paid off later. Proponents argue that AI's transformative potential may take years to fully realize, and cutting spend now could sacrifice long-term competitive advantage. Without concrete data, this remains a speculative defense.