Broadcom Signs Expanded Deals with Google and Anthropic, Giving the AI Startup Access to 3.5 Gigawatts of Compute Capacity

Key Takeaways

• Anthropic gains 3.5 GW TPU capacity, U.S.-focused
• Broadcom to produce next‑gen Google AI chips
• Anthropic revenue surpasses $30 B, over 1,000 million‑dollar clients
• Mizuho forecasts $21 B AI revenue for Broadcom in 2026
• Deal could double Broadcom’s AI revenue by 2027

Pulse Analysis

Broadcom’s deepening ties with Google signal a strategic push into the high‑performance AI chip arena. By taking on the production of Google’s next‑generation tensor processors, Broadcom leverages its semiconductor expertise to meet escalating demand from cloud providers and enterprises seeking faster, more efficient inference workloads. This move not only diversifies Broadcom’s product portfolio beyond its traditional networking and storage segments but also places it directly in competition with industry heavyweights like Nvidia and AMD, whose GPUs dominate current AI training pipelines.

Anthropic’s expanded access to 3.5 GW of TPU capacity marks a significant scaling milestone for the AI startup. The additional compute, primarily located in the United States, enables Anthropic to broaden its model offerings, reduce latency for enterprise customers, and compete more aggressively with OpenAI and other large‑scale model providers. The company’s revenue surge to over $30 billion—driven by a growing base of more than 1,000 clients each spending at least $1 million annually—demonstrates the commercial viability of subscription‑based AI services and underscores the importance of robust, dedicated hardware infrastructure.

Financial analysts see the Broadcom‑Anthropic alliance as a catalyst for massive earnings growth. Mizuho’s forecast of $21 billion in AI revenue for Broadcom in 2026, potentially doubling to $42 billion in 2027, reflects expectations that AI‑related hardware sales will become a core profit driver. As AI adoption accelerates across sectors—from finance to healthcare—Broadcom’s ability to supply both custom chips for Google and dedicated compute for Anthropic could secure a dominant foothold in the emerging AI infrastructure ecosystem, reshaping competitive dynamics and investor outlooks.