Broadcom AI Chip Revenue Triples as Applied Digital Secures $36 Billion in Data Center Capacity Deals
The AI infrastructure market sees significant capital flow into custom silicon and long-term data center capacity, with new financial instruments emerging to price compute.
The story
The AI hardware landscape continues to expand rapidly, with Broadcom reporting that its AI semiconductor revenue more than tripled year-over-year last quarter. This surge indicates robust demand for custom AI chips, a sector where Broadcom is deepening its partnerships with major AI players like Alphabet, Meta Platforms, and OpenAI.
Concurrently, data center operator Applied Digital announced securing $36 billion in contracted capacity deals with an investment-grade hyperscaler for its Polaris Forge 1 campus in North Dakota. This substantial commitment provides Applied Digital with significant revenue visibility and underscores the ongoing, large-scale investment in physical AI infrastructure. The developments highlight a critical shift in the AI build-out, where specialized silicon and dedicated data center capacity are attracting immense capital, signaling that the hardware layer is becoming a key battleground for investment.
Silicon
| Chip | Maker | What | For Whom |
|---|---|---|---|
| BR20X | Biren Technology | Next-generation GPU with broader low-precision support for AI acceleration. | China's domestic AI accelerator roadmap. |
The build-out
| Project | Who | Scale | Where |
|---|---|---|---|
| Contracted AI data center capacity | Applied Digital | $36 billion in contracted deals | Polaris Forge 1 campus, North Dakota |
Supply & policy signals
Broadcom's AI semiconductor revenue tripled year-over-year last quarter, with management expecting further acceleration.
Implication: Indicates strong and growing demand for custom AI chips, driving increased investment in specialized silicon.
Applied Digital secured $36 billion in contracted capacity deals with an investment-grade hyperscaler.
Implication: Provides significant revenue visibility for data center operators and signals continued hyperscaler investment in AI infrastructure.
Anthropic expects to spend $518 billion on AI infrastructure over the next decade.
Implication: Highlights the immense, long-term capital requirements for building out AI compute at scale.
CME Group to launch the first compute futures contract on October 5, 2026, tracking Nvidia H100 and B200 GPU rental rates.
Implication: Establishes a public reference price for AI compute capacity, allowing for hedging and potentially attracting new capital to the underlying hardware.
What we'll be watching
- CME Group's launch of the first compute futures contract on October 5, 2026, pending regulatory approval.
Reporting + analyst voices: grounded via Google Search at publish time.