Trending:

OpenAI is designing its own in-house chip, will tie up with Broadcom, TSMC as it moves away from NVIDIA

FP Staff October 30, 2024, 14:05:33 IST

While NVIDIA’s GPUs have long dominated the AI space, their limited availability has left many tech giants, like OpenAI, scrambling to secure supplies. Even companies like Amazon, Google, Meta, and Microsoft have found it challenging to meet their growing computational needs

Advertisement
OpenAI has already diversified by incorporating AMD chips alongside NVIDIA’s offerings, but designing its own chips marks a significant step towards reducing dependency on external suppliers. Image Credit: Reuters
OpenAI has already diversified by incorporating AMD chips alongside NVIDIA’s offerings, but designing its own chips marks a significant step towards reducing dependency on external suppliers. Image Credit: Reuters

OpenAI is charting a new course in its AI journey by designing its own in-house chip. The company, known for its groundbreaking AI models, is teaming up with Broadcom and Taiwan Semiconductor Manufacturing Company (TSMC) to manage costs and secure a more reliable supply chain, according to reports from Reuters.

With the demand for advanced computational power growing rapidly, this move reflects OpenAI’s strategy to stay ahead of the curve.

Moving beyond NVIDIA
While NVIDIA’s GPUs have long dominated the AI space, the limited availability of these powerful chips has left many tech giants, including OpenAI, scrambling to secure supplies. Even companies like Amazon, Google, Meta, and Microsoft have found it challenging to meet their growing computational needs.

STORY CONTINUES BELOW THIS AD

OpenAI has already diversified by incorporating AMD chips alongside NVIDIA’s offerings, but designing its own chips marks a significant step towards reducing dependency on external suppliers.

Although the company initially explored the idea of building its own chip manufacturing plants, the steep costs and time-consuming nature of such a project led OpenAI to focus instead on in-house chip design. To drive this initiative, OpenAI has put together a specialised team of around 20 chip engineers, including veterans with experience in developing Google’s Tensor Processing Units (TPUs). They aim to create the company’s first custom chip by 2026, though the timeline remains flexible as development progresses.

A strategic partnership
The collaboration with Broadcom and TSMC comes at a crucial time, as companies across the tech industry wrestle with the challenges of sourcing high-performance chips. OpenAI’s shift to custom design is part of a broader trend where companies are taking more control over their hardware to reduce bottlenecks and cut infrastructure costs. This strategy echoes efforts from other tech heavyweights who have explored similar paths to keep their AI ambitions on track.

The announcement had an immediate impact on the stock market, with Broadcom’s shares jumping 4 per cent and TSMC’s US-listed shares gaining over 1 per cent. This positive reaction reflects investor confidence in OpenAI’s ability to navigate the complexities of chip design while managing key industry partnerships.

As competition in the AI world heats up, OpenAI’s decision to dive into custom chip development signals its determination to stay competitive. While NVIDIA remains a significant player, OpenAI’s latest move hints at a future where companies rely less on external suppliers and build more of their own infrastructure to power the next generation of AI technologies.

STORY CONTINUES BELOW THIS AD
Home Video Shorts Live TV