IDC says only 19% use Claude extensively as Anthropic pushes enterprise via usage pricing

Enterprises may have watched Claude spend months chasing mass appeal, but IDC wants them to look again at Anthropic’s enterprise push. In IDC’s tracking over the past six months, only 19 percent of organizations use Claude “extensively,” while 25 percent are actively evaluating it, per IDC’s FERS Survey for March 2026. IDC frames the moment as an inflection point: the frontier-model era is still young, yet Anthropic is moving like it plans to be “enterprise AI provider” material before the rest of the pack catches up.

The urgency is real because IDC’s baseline assumption is blunt. “Currently, no frontier model company is mature enough to be evaluated as an enterprise AI provider on its own,” IDC said in a recent report titled “The Transformation of Anthropic (and What to Do About It).” IDC also argues Anthropic is running “at full speed” to reach that maturity bar ahead of competitors. The reason enterprises should care is not just adoption numbers. It is whether an enterprise actually has an evaluable, supportable, supply-backed system for real-world deployments, and whether the vendor can earn enough trust to become a durable part of an organization’s AI stack.

That stack is where adoption data gets interesting. IDC says many enterprise buyers still lean toward a vendor-neutral, multi-LLM strategy, which typically means they do not want their AI future locked into a single model provider too early. And yet, IDC’s report suggests Anthropic’s job is not just to win pilots, but to become visible and credible enough that Claude starts showing up in internal IT conversations alongside other options. In IDC’s view, Anthropic is doing that by shifting how it sells and by increasing the volume and breadth of enterprise-relevant activity.

The report points to a pricing pivot as part of the enterprise acceleration. In January, only two months after Anthropic began shifting enterprises away from seat-based pricing toward usage-based pricing, IDC notes that Anthropic’s enterprise work started stacking up. That shift matters in plain terms: seat-based pricing can feel like a tax on headcount, while usage-based pricing maps more closely to how organizations actually consume compute and model calls. It can make budgeting and procurement easier to justify for teams that want measurable usage controls. If enterprise buyers are already wary of “frontier model” immaturity, aligning commercial terms with consumption patterns is one way to reduce friction.

IDC also marshals a mountain of public movement to support the maturity narrative. From January through May 2026, the report says Anthropic produced well over 100 public interactions, including official announcements, release notes, blog posts, X posts, partner announcements, hiring news, policy moves, and press-covered transactions. The specific examples IDC highlights include the launch of the Claude Partner Network. IDC argues these initiatives expand distribution, bolster brand perception, facilitate future growth, enhance “stickiness” (aka lock-in), strengthen enterprise support, address needs of specific industries, demonstrate innovation, and shore up compute supply necessary to deliver services at scale. In other words, Anthropic is trying to solve the same enterprise problems buyers usually cite when they say they like the model but do not yet trust the provider.

Revenue signals are the other half of the equation, and IDC supplies a comparison that procurement teams will inevitably ask about. According to The Information, about 86 percent of Anthropic’s 2025 revenue was projected to come from enterprise sales. Meanwhile, IDC says OpenAI derives just 40 percent of its revenue from business sales, although the report notes that OpenAI’s business sales figure at the time was $5.2 billion, higher than Anthropic’s business revenue of $3.9 billion. Put differently, Anthropic may be pushing hard toward enterprise readiness, but it is also doing so while already looking like a company whose growth depends heavily on enterprise budgets. That is a different incentive profile than a supplier who can wait for enterprises to “catch up.”

IDC closes by urging a reassessment, not a reflexive dismissal. “Anthropic’s transformation has just started, but the direction is clear enough for CIOs and CISOs to pay attention and reassess where Claude fits in a multi-LLM or an agentic AI Strategy,” IDC said. For leaders who govern AI risk, data handling, vendor support, and operational continuity, this is not a vibe check. It is a call to revisit LLM and agent evaluations with an eye toward whether Anthropic can function as a reliable technology provider, not just a demo-worthy model. In this phase of AI adoption, the models change quickly. What actually decides winners is whether the enterprise machinery can keep up.