Key Takeaways:
Anthropic is betting a 75 percent cut to cache-read pricing and a looser data-residency policy will lock in the enterprise revenue that must carry a roughly $965 billion valuation into its fall IPO.
The company on Tuesday released Fable 5.1, its flagship model update, slashing the price of cache reads — the charge for re-reading previously processed content — by 75 percent. Product management lead Dianne Penn said that step can consume half or more of token usage in complex long-context tasks such as software development and code review, and was the most common complaint from customers.
"Cache reads are where the cost concentrates in the workloads our enterprise clients actually run," Penn said. "This is the area we wanted to improve, and it is a direct response to what they told us."
The pricing move targets the fastest-growing slice of Anthropic's business. The company projects annualized revenue above $65 billion, more than seven times the pace at the end of 2025, with about 80 percent coming from enterprise customers including eight of the Fortune 10. It counts more than 1,000 clients paying over $1 million a year, up from 500 in February. Claude Code, its agentic coding product, reached a $2.5 billion annualized run rate in nine months.
Anthropic also reversed a long-standing data policy, now letting enterprise customers keep data on their own cloud infrastructure while still subjecting it to Anthropic's security review. The prior rule, designed to limit cyber risk, had pushed away clients with strict data-sovereignty requirements. The company separately opened an invitation-only preview of its higher-tier Mythos 5.1 model.
The moves come as Anthropic and OpenAI race to convert private-market valuations into public-market revenue. Anthropic filed a confidential IPO registration with the SEC on June 1 and could list as early as October, after a Series H round in May that valued it at about $965 billion — above OpenAI's roughly $852 billion. Anthropic's enterprise AI market share reached 34.4 percent in April, overtaking OpenAI's 32.3 percent for the first time.
Pricing pressure reshapes the enterprise AI market
The cache-read cut reflects a broader squeeze across the US AI model market, where enterprise buyers are demanding more predictable cost structures as they scale deployments. Anthropic kept new-input token pricing unchanged from Fable 5, concentrating the discount on the reuse-heavy workloads that dominate coding and research tasks. The model was also optimized to complete the same work with fewer tokens, compounding the savings.
Fable 5.1 strengthens performance in two professional areas: long-horizon coding across entire codebases, including systematic code review, and scientific work such as experiment design, process simulation, and reading dense chart data. Anthropic said an improved safety classifier cut false-positive rates, reducing erroneous blocks for users in cybersecurity and biology.
The update follows a regulatory flashpoint. The US government imposed access restrictions on the Fable and Mythos models in June over security concerns; the limits on Fable 5 were lifted after Anthropic added a classifier that identifies and blocks more cyber tasks, and Mythos 5 access was later restored on a narrower basis. The episode underscores the oversight frontier-model vendors face even as they court government and defense clients.
What the pricing strategy means for the IPO
The discount is a direct challenge to OpenAI, which is also preparing to go public and is expected to file within weeks. Anthropic's path to profitability is a key differentiator: it reported a non-GAAP operating profit in the second quarter of 2026 and targets $70 billion in revenue with $17 billion in free cash flow by 2028, while OpenAI is not expected to turn a profit until 2030.
At roughly 22 times annualized sales, Anthropic's valuation already prices in sustained hypergrowth. The cache-read cut and data-sovereignty shift are designed to keep enterprise share expanding — and to give institutional investors a revenue story that justifies the multiple when the company opens its books. The risk is that aggressive pricing compresses the gross margins that must eventually support that valuation, a tension Anthropic will have to resolve in its public-market debut.
This article is for informational purposes only and does not constitute investment advice.
