OpenAI Partners with TSMC, Broadcom to Develop In-House AI Chips, Aiming to Rival NVIDIA's Market Dominance
October 30, 2024
OpenAI is collaborating with Broadcom and Taiwan Semiconductor Manufacturing Company (TSMC) to create its first in-house chip, aimed at bolstering its artificial intelligence infrastructure.
This strategic move comes as major clients like Microsoft and Meta seek alternatives to NVIDIA's GPUs, which dominate the market but are hindered by high costs and supply shortages.
OpenAI's decision to develop custom chips follows a thorough evaluation of its options to diversify its chip supply and manage costs, ultimately abandoning plans for in-house foundries due to their prohibitive expenses and lengthy timelines.
Despite a projected revenue of $3.7 billion in 2024, OpenAI anticipates incurring a staggering $5 billion loss this year, primarily driven by high compute costs associated with training AI models like ChatGPT.
The demand for significant computing power is crucial for generative AI applications, which require extensive resources for training and real-time inference.
Following a recent funding round of $6.6 billion led by Thrive Capital, OpenAI's valuation has soared to approximately $157 billion, solidifying its position in the AI chip market.
In addition to chip development, OpenAI is actively pursuing investments for data centers that will support the new AI chips, engaging with global investors to facilitate this expansion.
CEO Sam Altman has highlighted the urgent need to boost global chip supply, previously suggesting a monumental $7 trillion investment to enhance chip production capabilities.
This shift towards custom chip infrastructure reflects a broader trend in the tech industry, where companies are investing in self-sufficiency to meet rising computational demands.
The announcement of OpenAI's partnership has positively influenced Broadcom's stock, which rose over 4.5%, as investors view this collaboration as a promising venture amid forecasts of significant AI silicon sales.
Analysts are advising investors to keep an eye on companies supplying AI hardware, as the current market dynamics suggest potential growth opportunities.
The urgency for efficient AI processing and the complexity of AI applications are driving this strategic shift in OpenAI's approach to chip development.
Summary based on 51 sources
Get a daily email with more Tech stories
Sources

The Verge • Oct 29, 2024
OpenAI will start using AMD chips and could make its own AI hardware in 2026
TechCrunch • Oct 29, 2024
OpenAI reportedly planning to build its first AI chip in 2026 | TechCrunch
Insider • Oct 31, 2024
OpenAI is taking a page out of Big Tech's playbook by reportedly building its own chips
Yahoo Finance • Oct 29, 2024
Exclusive-OpenAI builds first chip with Broadcom and TSMC, scales back foundry ambition