In the first half of 2026, companies increased their AI token spending fivefold, yet few can demonstrate corresponding revenue growth, according to a discussion on saastr.com. This surge in expenditure contrasts with the lack of clear financial returns, raising concerns about the efficiency of AI investments across the B2B sector.
The trend was highlighted by Brian Armstrong's post on Coinbase, which reported a 50% reduction in AI spending this quarter while usage rose, driven by a shift to open-source models and away from frontier AI applications. Despite the cut, the spending level returned only to November 2025 figures, reflecting a broader pattern where companies aggressively ramped up AI costs late last year before implementing cost controls in early 2026.
This pattern underscores a growing challenge in the AI industry: high operational costs without evident revenue uplift. The initial spike in spending coincided with the rise of agentic coding technologies, but the subsequent cost discipline suggests companies are recalibrating their AI budgets. The situation highlights the difficulty for even well-established B2B firms to justify large AI expenditures without clear financial benefits.
The discussion on saastr.com took place on June 30, 2026, featuring insights from Harry Stebbings, Jason Lemkin, and Rory O’Driscoll, who emphasized the need for measurable returns on AI investments amid rising costs and usage.