Companies Are Throttling Employees’ AI Use Because It’s Too Expensive

The story frames this as a cost problem, but the more precise read is a governance problem that cost is forcing. Enterprises never built procurement discipline around token consumption because vendors priced early access to encourage volume, and now the bill has arrived before the ROI case was closed.

This connects directly to the 'Tokenpocalypse' podcast coverage from 404 Media on July 1st, which flagged token economics as an unsustainable pressure point for heavy API consumers. That piece treated it as an infrastructure challenge; this story confirms it has already crossed into HR and access policy. Pair that with the Platformer piece from the same day on the AI backlash, which identified a structural lag between deployment pace and the industry's ability to manage downstream costs, and you have a coherent pattern: the enterprise adoption curve is bending not because of capability disappointment but because the pricing architecture was never designed for sustained, org-wide use. Meta's move to sell surplus compute (covered July 1st via The Decoder) adds an ironic counterpoint: hyperscalers are building new revenue lines off the same infrastructure costs that are now forcing enterprise customers to ration access.

Watch whether any of the named vendors (Adobe, Atlassian) publicly revise their per-seat or consumption pricing tiers before Q3 earnings calls. If they do, that confirms enterprise pushback has reached the product and pricing teams, not just sales.