The race for enterprise AI dollars isn't about who has the smartest model — it's about who can get businesses to pay premium prices for intelligence they might not need.
The Summary
- OpenAI's GPT-5.6 Sol generated 25% more enterprise spend than Anthropic's premium Fable 5 model in July, according to Ramp's AI Index tracking 70,000+ US businesses
- Anthropic still leads overall adoption at 43.5% vs OpenAI's 39.7%, but weaker demand for its top-tier model signals pricing pressure
- The gap reveals the central tension in pre-IPO Anthropic's strategy: betting enterprises will pay premiums for state-of-the-art when "good enough" keeps getting cheaper
The Signal
Ramp's data exposes the fault line in enterprise AI economics. Anthropic holds the overall adoption lead, but its flagship Fable 5 model generated only 75% of the spend that OpenAI's GPT-5.6 Sol pulled in July. That's not a rounding error. That's businesses voting with budgets, and they're saying the premium tier isn't worth the premium price.
Fable 5 had a messy birth. Government regulators temporarily killed it before its July 1 resurrection, then Anthropic kept adjusting pricing throughout the month like a startup searching for product-market fit. Not the launch you want when your IPO pitch is "enterprises will pay top dollar for our superior intelligence."
"In the run-up to its IPO, Anthropic is staking its business on the idea that people will be willing to pay for its state-of-the-art tech rather than competitors' cheaper options."
The Ramp index tracks actual spending from bills and expenses, not surveys or self-reported adoption numbers. It skews toward smaller companies, which matters because those are exactly the businesses most sensitive to the price-performance tradeoff. They're not buying AI for the resume. They're buying it to replace headcount or accelerate revenue, and they'll take 90% of the capability at 60% of the cost every single time.
Meanwhile, the broader adoption numbers tell a different story:
- Anthropic: 43.5% of businesses (up from previous months)
- OpenAI: 39.7% (steady climb)
- xAI: 4% (new entrant making noise)
- Google: 6.2% (sliding from 6.4%, which is brutal for an incumbent)
Google's decline is the quiet catastrophe here. Anthropic and OpenAI are fighting for pole position while the tech giant that invented the transformer architecture bleeds share. That's what happens when you ship conservative products to protect search revenue while startups ship agents that actually do things.
The Implication
Anthropic faces a pricing paradox before its IPO. It needs to prove premium models command premium revenue, but enterprises are pattern-matching to SaaS economics: once the baseline gets good enough, the marginal value of "better" collapses. If GPT-5.6 Sol can handle 85% of business use cases at a lower price point, Fable 5 has to be 10x better at the remaining 15% to justify the gap.
Watch whether Anthropic doubles down on enterprise vertical specialization or follows OpenAI downmarket with cheaper tiers. The IPO roadshow will tell investors one story. July spending data is telling another. The smart money is on businesses that can monetize AI through agents and automation, not raw model access. The model is becoming the commodity. The orchestration layer is where margin lives.