Anthropic Locked In Google & Broadcom While Its Revenue Tripled

Anthropic announced after Monday’s bell they signed an agreement with Google and Broadcom for “multiple gigawatts of next-generation TPU capacity” to continue to train Claude.

Both companies’ stocks were relatively flat on the news. For Anthropic, that was only half of the good news.

Anthropic says their run-rate has surpassed $30 billion, up from $9 billion at the end of 2025.

“This partnership gives us the compute to keep pace,” the company said.

For Anthropic, and competing models from OpenAI, Google, and xAI, it’s like water to plants. AI models cannot train or run without massive amounts of compute. In the last two years, the biggest bottleneck for every major AI company has not been talent, data, or algorithms but rather compute and energy, because they enable training bigger models faster and running more experiments which all lead to better performance and the ability to serve more customers.

And on that point, it’s useful to compare Anthropic’s revenue and cash burn with chief competitor OpenAI. Sam Altman’s firm is projected to spend $121 billion in compute by 2028, amounting to $85 billion in losses that year alone, reports The Wall Street Journal. And that’s after doubling their revenue. Anthropic’s numbers suggest a different story, however. They’re winning with both the general public and the more valuable enterprise customers. Nearly every dollar of revenue it books comes from paying business customers, a more efficient base than OpenAI’s largely free consumer user base. Their yearly training costs are also considerably lower and, as a share of their revenue, bottom out much sooner than OpenAI. And OpenAI’s free cash flow is deep in the red until 2030, when their projections, of course, show they’ll magically flip it all around. Anthropic’s free cash flow is modestly negative until 2028, telegraphing a more controlled ascent.

For Google and Broadcom, sorry, Nvidia, this is another vote of confidence, not that they needed one. It further solidifies their relationship with Anthropic and validates TPUs as a serious alternative to GPU-dominated infrastructure, delivering higher hardware utilization and ultimately, more value per compute dollar.

• Google (GOOGL) — This agreement secures Anthropic as a customer for its TPU capacity, generating revenue and validating its AI hardware strategy.

• Broadcom (AVGO) — As a supplier of TPU hardware, Broadcom benefits from increased demand and validation of its specialized AI chips.

• Anthropic — The agreement secures critical compute capacity, enabling continued training and scaling of its Claude AI model, supporting its rapid revenue growth.

• Microsoft (MSFT) — As a major cloud provider and investor in AI, increased demand for AI compute capacity generally benefits all major players in the cloud infrastructure market.

• Amazon (AMZN) — Amazon’s AWS division, a leading cloud provider, stands to benefit from the overall surge in demand for AI compute and data center services.

• Equinix (EQIX) — As a leading data center REIT, Equinix benefits from the increased demand for physical infrastructure to house AI compute resources.

• Artificial Intelligence — The agreement underscores the rapid growth and investment in AI model development, driving innovation and market expansion.

• Cloud Computing — The need for “multiple gigawatts of next-generation TPU capacity” signifies massive demand for cloud infrastructure services to power AI.

• Semiconductor Manufacturing — The validation of TPUs and the overall demand for AI compute hardware drives growth for specialized chip manufacturers.

• Utilities — The “multiple gigawatts” requirement indicates a significant increase in electricity demand for data centers, benefiting power generation and distribution companies.

Our analysts just identified a stock with the potential to be the next Nvidia. Tell us how you invest and we’ll show you why it’s our #1 pick. Tap here.

• Nvidia (NVDA) — While the validation of TPUs presents a competitive alternative to GPUs, the overall demand for AI compute remains extremely high, suggesting continued strong demand for Nvidia’s products.

• OpenAI — The article highlights OpenAI’s projected $121 billion compute spend by 2028 and $85 billion in losses, contrasting sharply with Anthropic’s more efficient model and potentially putting pressure on its financial sustainability.

• xAI — As a competitor to Anthropic and OpenAI, xAI faces similar intense compute demands and financial pressures in the race to develop advanced AI models.

• [Immediate] Increased Investment in AI Infrastructure — The “multiple gigawatts” deal signals an immediate and massive capital allocation towards AI compute infrastructure, driving demand for specialized hardware and data center capacity. Confidence: High.

• [Short-term] Validation of Alternative AI Hardware — The agreement validates Google’s and Broadcom’s TPU technology as a viable and competitive alternative to Nvidia’s GPUs, potentially diversifying the AI hardware market. Confidence: High.

• [Medium-term] Intensified Competition in AI Model Development — Anthropic’s reported efficiency and revenue growth, contrasted with OpenAI’s high burn rate, will intensify pressure on all AI model developers to secure compute efficiently and demonstrate viable business models. Confidence: High.

• [Long-term] Rising Energy Demand for Data Centers — The requirement for “multiple gigawatts” of power for AI training indicates a sustained and significant increase in electricity consumption, impacting energy markets and utility providers over the long term. Confidence: High.

• [Long-term] Shift Towards Enterprise AI Solutions — Anthropic’s success with “paying business customers” suggests a potential long-term trend where enterprise-focused AI models with clear monetization strategies gain an advantage over consumer-focused models with high free user bases. Confidence: Medium.

↑ [Data Center Construction Spending] — The need for massive compute capacity will drive increased investment in building and expanding data centers globally.

↑ [Semiconductor Equipment Orders] — Demand for specialized AI chips like TPUs and GPUs will lead to higher orders for semiconductor manufacturing equipment.

↑ [Electricity Prices (Regional)] — Localized electricity prices near major data center hubs could see upward pressure due to increased demand from AI compute.

→ [Cloud Services Spending] — While specific providers benefit, overall cloud spending will continue its upward trend, with AI being a major driver.

↓ [AI Startup Valuations (Inefficient Models)] — Startups with unsustainable compute burn rates and unclear monetization paths may face downward pressure on valuations as efficiency becomes paramount.

One stock. Nvidia-level potential. 30M+ investors trust Moby to find it first. Get the pick. Tap here.