Version 2

Anthropic has confirmed an annualized revenue run rate surpassing $30 billion as of April 2026 — up roughly 80x from its starting point — and is projecting its first operating profit in Q2 2026 [1][2]. To fuel that growth, the company has locked in compute at extraordinary scale: a 10-year, $100B+ commitment to AWS covering up to 5 GW of capacity [1], alongside a reported $200 billion deal with Google and Broadcom [7][8], while separately exploring Microsoft's Maia 200 chips as a third inference pathway [9]. The week of May 18–22 saw three additional moves: the acquisition of Stainless (SDK and MCP server generator, reportedly $300M+) [13][14], the arrival of Andrej Karpathy on the pre-training team [11][12], and the disclosure that Anthropic now serves 8 of the Fortune 10 and roughly 70% of the Fortune 100 [15].

The convergence of a $30B revenue run rate, imminent operating profitability, and compute commitments totaling hundreds of billions of dollars signals that Anthropic has moved from research-stage challenger to a company with structural leverage in enterprise AI infrastructure. The scale of multi-cloud commitments — far exceeding what any single provider can supply — reflects both enormous demand and a deliberate strategy to avoid single-vendor dependency, a posture that will shape how the industry prices and negotiates compute for years.

Anthropic's financial position has shifted more sharply in the first half of 2026 than almost any observer had predicted. In late April, the company confirmed in an official announcement that its annualized revenue run rate has surpassed $30 billion, up from approximately $9 billion at the end of 2025 [1] — a pace described by VentureBeat as approximately 80x growth [2]. That trajectory enabled the company to project its first operating profit in Q2 2026, roughly two years ahead of prior internal estimates [3][4]. Reporting across the Financial Times, TechCrunch, and the Wall Street Journal framed the inflection as evidence that frontier AI has crossed from discretionary experiment into core business infrastructure [3][5].

The revenue growth has triggered an equally dramatic build-out of compute supply. In April, Anthropic and Amazon formalized an expansion securing up to 5 gigawatts of capacity — with nearly 1 GW of Trainium2 and Trainium3 expected online by end of 2026 — under a 10-year commitment by Anthropic to spend more than $100 billion on AWS technologies [1]. Amazon is adding a further $5 billion investment in Anthropic with up to $20 billion more committed going forward, building on a prior $8 billion [1][6]. Simultaneously, Anthropic announced expanded partnerships with Google and Broadcom in an arrangement Reuters reported involves a $200 billion commitment to Google's cloud [7][8]. These arrangements make Claude the only frontier AI model available across all three of the world's largest cloud platforms: AWS Bedrock, Google Cloud Vertex AI, and Microsoft Azure Foundry [1]. Separately, The Information reported that Anthropic is in early-stage talks to lease Microsoft's custom Maia 200 chips for inference workloads, with Microsoft pitching the hardware as more cost-effective than Nvidia's at scale [9]. No deal has been finalized, but news of the talks moved Microsoft's stock approximately 2% on the day [10].

On the talent side, Andrej Karpathy — OpenAI co-founder, former Tesla AI director, and a prominent independent voice in the field — announced in May 2026 that he is joining Anthropic specifically to work on pre-training [11][12]. Karpathy framed the move in terms of historical significance: 'I think the next few years at the frontier of LLMs will be especially formative' [11]. Coverage noted the move as consistent with views he has stated publicly about the epistemic necessity of working inside a frontier lab. On the tooling side, Anthropic acquired Stainless, the company that had generated every official Anthropic SDK since the earliest days of the Claude API and had built MCP server generation capabilities used by hundreds of companies [13]. The deal was reportedly valued at over $300 million [14]. Anthropic's stated rationale was that agents are only as capable as the systems they can connect to, making SDK and MCP infrastructure as strategically important as the models themselves [13].

The enterprise penetration behind these figures has also come into clearer view: Anthropic reportedly serves 8 of the Fortune 10 and approximately 70% of the Fortune 100 [15]. The open question — flagged across financial commentary — is whether operating profitability can hold as the enormous compute commitments convert from signed agreements into real capital expenditure, and whether the multi-cloud strategy that enabled such broad enterprise reach can be sustained as hyperscaler interests and Anthropic's own infrastructure ambitions increasingly overlap.