Any company signing a strategic agreement with OpenAI is witnessing an instant reaction on the stock market. The artificial intelligence sector is entering a dramatically more competitive and dynamic phase — and Elon Musk, who once co-founded OpenAI but failed to secure a controlling stake, may now have reason to regret it.
In October, OpenAI announced a strategic partnership with Broadcom to co-develop and deploy its first internally designed AI chips. The goal is to meet the company’s rapidly growing computing demand, driven by large generative models such as ChatGPT.
The new custom accelerators are expected to launch in the second half of 2026, delivering 10 gigawatts of computing capacity — over five times the output of the Hoover Dam, enough to power more than 8 million U.S. households.
Following the announcement, Broadcom shares jumped nearly 8% at market open, marking one of the most commercially and strategically significant contracts in the company’s history. Traditionally associated with networking and telecom chips, Broadcom is now emerging as a major contender in core AI infrastructure — with some investors calling it “the next Nvidia.”
“Investors increasingly view Broadcom as the next Nvidia — its AI products could become the next breakout phenomenon. Competition in AI infrastructure is accelerating rapidly,”
said Mikołaj Sobierajski, equity analyst at XTB, in an interview with MarketNews24.
Under the agreement, OpenAI will design the GPU architecture, while Broadcom will handle the technical development, manufacturing and data center systems integration. The chips will power both OpenAI’s own facilities and third-party cloud providers, using Broadcom’s Ethernet networking, seen as a potential alternative to Nvidia’s proprietary interconnects.
OpenAI aggressively diversifies beyond Nvidia
This latest deal aligns with OpenAI’s broader strategy to reduce reliance on Nvidia, whose limited supply and high pricing are major constraints on AI scaling.
The company has already:
- Committed to 6 gigawatts of compute from AMD
- Maintains a 10-gigawatt infrastructure build with Nvidia
- Now adds another 10 gigawatts via Broadcom
In total, OpenAI has secured 26 gigawatts of contracted compute — more than the entire summer electricity consumption of New York City.
Also in October, AMD and OpenAI finalized a deal to deploy 6 gigawatts of AMD GPUs over the coming years. As part of the agreement, AMD granted OpenAI warrants to purchase up to 160 million AMD shares, exercisable in stages upon milestone delivery.
“AMD is Nvidia’s main GPU competitor — its chips are technologically strong but far cheaper. After the OpenAI deal, AMD stock surged 25% in a single trading session,”
noted Sobierajski. “The upcoming OpenAI IPO may well become the turning point of the current tech bull market.”
Microsoft, Oracle, Nvidia — and now Broadcom & AMD
In parallel, Microsoft and OpenAI signed an agreement in October that paves the way for OpenAI to convert into a public-benefit corporation, valuing the company at $500 billion and giving it greater operating independence.
In September, Nvidia announced a $100 billion infrastructure project to build at least 10 gigawatts of AI compute capacity dedicated to OpenAI — again matching New York City’s peak power demand.
“This is now the fourth major deal involving OpenAI in just a few weeks,”
adds Sobierajski.
“It is becoming clear that AMD, Broadcom, Nvidia and Oracle will likely lead the tech bull market — while OpenAI is becoming increasingly independent from Nvidia.”
Meanwhile, Elon Musk missed the moment
OpenAI, founded in 2015 as a nonprofit AI research lab, once had Elon Musk on its board. Musk attempted to acquire control of the company — and failed. OpenAI has since become the uncontested market leader in Generative AI infrastructure and platforms.
The company is now also working on an AI-driven music creation tool, capable of composing fully original musical tracks based on text prompts and style descriptors.
