Anthropic Revenue Tops $30 Billion as Google and Broadcom Sign Multi-GW TPU Deals

Surging demand for advanced AI models is reshaping cloud infrastructure, energy markets, and Bitcoin mining economics, with anthropic revenue emerging as a key signal of this shift.

Anthropic revenue growth and Claude adoption

Anthropic disclosed that its annualized revenue has climbed past $30 billion, a sharp acceleration from approximately $9 billion at the end of 2025. This jump in anthropic annual revenue reflects rapid uptake of its Claude AI models across enterprise clients and software developers. Moreover, the company is seeing strong demand from large organizations integrating Claude into workflows, products, and internal tooling.

The firm also reported that the number of business customers spending more than $1 million per year on Claude has doubled in under two months, rising from 500 to more than 1,000. That said, Anthropic has framed this as an early stage of broader enterprise AI adoption, suggesting significant room for further growth as organizations scale deployments.

Multi-gigawatt TPU capacity deals with Google and Broadcom

To sustain this trajectory, Anthropic announced long-term infrastructure agreements with Google and Broadcom for several gigawatts of next-generation TPU (Tensor Processing Unit) compute capacity. The new infrastructure is expected to begin coming online in 2027 and will be used to train and operate future versions of Claude. However, the scale of the deal also underscores how dedicated AI hardware is now central to performance and competitive advantage.

Anthropic revealed that it has secured roughly 3.5 gigawatts of next-generation Google TPU capacity through Broadcom starting in 2027. This commitment is in addition to about 1 gigawatt of Google compute that Anthropic is already slated to receive in 2026. Together, these agreements signal a multiyear build-out of specialized infrastructure to support increasingly capable models.

Across the sector, major AI developers are racing to lock in long-term access to training and inference capacity. Moreover, the combination of Google’s TPU ecosystem and Broadcom’s semiconductor design and manufacturing capabilities positions them as critical suppliers in this expanding market. These moves highlight broader ai hardware partnerships trends that are reshaping the cloud and chip landscape.

AI compute build-out and anthropic revenue implications

The surge in anthropic revenue is closely tied to the company’s ability to secure massive cloud compute deals and scale next-generation models. With multi-year TPU capacity in place, Anthropic is positioning itself to expand Claude’s capabilities while maintaining competitive performance. Furthermore, the agreements illustrate how compute availability is becoming a primary constraint on AI growth, rather than model design alone.

Market implications include rising enterprise demand for AI services, increasing capital intensity, and the growing strategic importance of hardware suppliers. As AI adoption spreads, access to low-cost, high-density compute is emerging as a key differentiator between leading AI labs and smaller competitors. This dynamic is likely to influence anthropic revenue growth and shape industry structure over the next several years.

Competition with Bitcoin mining for power and infrastructure

The rapid build-out of AI infrastructure is directly competing with Bitcoin mining for scarce physical resources such as grid connections, land, cooling capacity, and low-cost electricity. According to Cambridge tracking data, global Bitcoin mining continuously consumes an estimated 13 to 25 gigawatts of power. However, a single Anthropic deal delivering multiple gigawatts of demand shows that AI has become one of the largest new electricity users in the United States.

Several publicly listed Bitcoin mining companies are now pivoting toward ai hosting bitcoin miners strategies and high-performance computing to secure stable, contracted revenue. Examples include large conversions of mining facilities into AI data centers and long-term hosting agreements with Anthropic and other AI customers. Moreover, mining economics have come under pressure, with some operators facing loss-making conditions at current BTC prices, while AI hosting offers predictable cash flows backed by enterprise contracts.

Analysts estimate that a substantial share of mining companies‘ revenue could come from AI and high-performance computing by the end of the year. In aggregate, more than $70 billion in cumulative AI and HPC deals has been announced across the public mining sector. This capital reallocation underscores how AI demand is reshaping the business models of traditional mining players.

Data center power demand and grid constraints

The power grid is coming under increasing stress from concentrated data center demand, including large AI clusters. Grid operators in the United States project capacity shortfalls in coming years, while industry studies forecast U.S. data center electricity demand rising sharply through 2030. Moreover, single facilities reaching 1 gigawatt of load can rival the consumption of small cities, intensifying local constraints.

Many announced data center projects are already facing delays tied to power limitations and shortages of critical grid equipment. Anthropic’s multi-gigawatt commitment to new AI capacity enters this constrained environment, heightening competition for grid access, substations, and transmission upgrades. As a result, data center power demand is increasingly a central issue for regulators, utilities, and technology firms.

Bitcoin miners turn into AI infrastructure landlords

Over the past decade, Bitcoin miners have assembled portfolios of remote sites with favorable power purchase agreements, large grid connections, proximity to substations, and substantial cooling capacity and land. Now, these assets align closely with AI deployment needs. Consequently, many miners are converting mining facilities into data centers for AI customers and repositioning themselves as infrastructure landlords with long-duration leases and institutional tenants.

This strategic shift carries important implications for the Bitcoin network. Large miners are monetizing BTC holdings to finance AI conversions, adding sell pressure to spot markets. Moreover, as mining capacity is redirected toward AI workloads, Bitcoin hash rate and mining difficulty can decline, at least temporarily, affecting short-term network security metrics.

Over the longer term, the publicly listed mining sector may increasingly resemble diversified infrastructure operators. They could focus on leasing power, space, and uptime to AI companies while mining opportunistically when economics are favorable. That said, the pace and scale of this transition will depend on relative returns from AI hosting versus traditional block-reward mining.

Broadcom, Google, and the expanding AI hardware stack

Broadcom separately announced an extended partnership with Google to design and supply future generations of specialized AI processors and related technologies through 2031. Broadcom has long manufactured Google’s TPUs and confirmed that it is expanding deliveries. The company indicated it was already supplying approximately 1 gigawatt of computing power in 2026 and expects demand to exceed 3 gigawatts by 2027.

Analyst estimates suggest a significant AI-driven revenue opportunity for Broadcom tied to these long-term agreements. Moreover, Broadcom is also involved in custom processor design programs with other major AI developers, extending its reach across the gpu tpu trainium hardware stack. Google’s partnership with Broadcom on google broadcom tpu capacity reinforces the strategic importance of bespoke accelerators for leading cloud and AI providers.

Anthropic’s multi-cloud hardware strategy

Anthropic has emphasized that it trains and runs Claude across a range of hardware platforms, including AWS Trainium processors, Google TPUs, and Nvidia GPUs. This diversified approach aims to optimize performance, cost, and resilience while tapping different cloud ecosystems. Furthermore, the company has signaled plans for substantial investment in U.S. computing infrastructure as it scales next-generation models.

The combination of rapid revenue growth, large-scale TPU commitments, and a multi-vendor hardware strategy shows how compute capacity is becoming a core driver of growth and differentiation in the AI industry. In this environment, Anthropic’s agreements with Google and Broadcom, alongside its broader cloud relationships, position the company to compete aggressively for enterprise AI workloads over the coming decade.

In summary, Anthropic’s soaring revenue, multi-gigawatt compute deals, and the sector-wide shift toward AI hosting highlight how advanced models are reshaping infrastructure, energy demand, and even the economics of Bitcoin mining.