Broadcom Expands AI Chip Partnerships with Google and Anthropic
Broadcom has announced a long‑term agreement with Google to develop and supply custom Tensor Processing Units for future generations of Google’s AI systems, deepening a partnership that has become central to one of the industry’s most significant chip programs. In parallel, the company disclosed a supply assurance arrangement covering networking and other components for Google’s next‑generation AI racks through 2031.
Moreover, Broadcom, Google, and Anthropic have expanded their three‑way collaboration, giving Anthropic access to approximately 3.5 gigawatts of next‑generation TPU‑based AI compute capacity beginning in 2027. According to the filing, Anthropic’s consumption of this expanded capacity depends on its continued commercial success, and the parties are also discussing plans with operational and financial partners to support the rollout.
A Long‑Term AI Partnership
The new agreements formalize and extend a relationship that has evolved over more than a decade. Broadcom has co‑designed every generation of Google’s TPU since the program began, and these chips now underpin much of Google’s AI strategy, including the training and serving of its Gemini model family. In late 2025, Google introduced its seventh‑generation TPU, Ironwood, and is already preparing an eighth‑generation chip built on a 3nm process node for future mass production.
At the same time, Anthropic’s participation marks a major expansion of TPU usage beyond Google’s own data centers. The AI firm, known for its Claude family of models, first announced a landmark deal with Google in October 2025 to access up to one million TPUs, a multi‑billion‑dollar arrangement. Later, Broadcom’s CEO confirmed that Anthropic was behind a $21 billion order for Ironwood‑based TPU systems and reported that the company had already shipped one gigawatt of TPU capacity to Anthropic, with plans to scale beyond three gigawatts by 2027.
Broadcom’s Growing Role in AI
These agreements also highlight Broadcom’s push to position itself as the leading partner for hyperscale custom silicon. In its fiscal first quarter ended February 2026, AI‑related revenue jumped 106 percent year over year to 8.4 billion dollars, reflecting the rapid growth of demand for specialized AI chips. Company leadership has said it has “line of sight” to more than 100 billion dollars in AI chip sales by 2027, pointing to planned gigawatt‑scale deployments at several major AI firms.
Following the news, Broadcom’s shares rose in after‑hours trading, signaling strong investor confidence in the firm’s expanding role at the heart of large‑scale AI infrastructure.
