Broadcom shares surge 9.4% following OpenAI partnership news- a forward-thinking move into the highly profitable and growing AI hardware market.
The competition outweighs the need. The market has witnessed an unimaginable demand for AI systems. But training and running them requires immense computing power. The leading players have realized that manufacturing their own AI chips could be a way to cure the insatiable demand and take the lead.
And with the same realization, there’s a new player in the chip manufacturing arena-
In recent developments, OpenAI has partnered with Broadcom to launch internal AI chips by 2026.
The one significant difference is that it’ll be using the chips in-house, rather than selling them. This is quite a strategic move given how the tech giants keep eyeing each other’s next move in the game. For now, OpenAI wishes to reduce its reliance on NVIDIA and develop its first-gen AI silicon.
AI models such as ChatGPT are the end products, and that’s what users ultimately perceive. But the engineers behind the development of these models have much more to address. From computing power to operate on and AI chips to fuel the systems, the demand is currently outweighing the supply.
This is a conundrum for tech companies that want to lead the AI race. Market dominance comes later; the first step is to get hands on as many raw materials as one can. It’s why investors are investing billions in data centers and other AI hardware.
The same applies to this alliance between OpenAI and Broadcom.
The AI powerhouse has committed to over $10 billion worth of orders with its co-creator, says Broadcom’s chief executive. The crux of the deal is developing customer AI chips to remedy the shortage of processing power.
In line with this, businesses are torn between manufacturing chips in-house or outsourcing them. The ones belonging to the major leagues, from Microsoft to Apple, have been manufacturing and leveraging their own in-house chips for their AI models. OpenAI is planning to ride the same coattails.
Meanwhile, some in the market choose to outsource their hardware from. And this is an arena where companies like NVIDIA and TSMC come into the picture.
While NVIDIA continues to dominate the AI hardware space, its numbers have significantly slowed down. And Sam Altman seems to be aiming to catch up, especially after GPT-5, which required extensive compute power.
The eyes aren’t on the end product for now, but the medium to deliver the same- that’s where the capabilities of AI models fuel from.