OpenAI to make own AI chips with Broadcom in 2026: report

OpenAI is reportedly partnering with Broadcom to begin mass production of its own custom AI chips by 2026, a strategic move aimed at reducing its reliance on Nvidia and bolstering its internal computational infrastructure. This initiative, intended exclusively for in-house use, aligns with Broadcom's recent disclosure of a $10 billion custom AI chip order from an unnamed client, and supports OpenAI CEO Sam Altman's goal of acquiring over one million GPUs by late 2025.

Analysis

OpenAI's reported partnership with Broadcom to mass-produce proprietary AI chips by 2026 marks a significant strategic pivot aimed at reducing its dependency on Nvidia and securing its computational infrastructure. This move is substantiated by Broadcom CEO Hock Tan's recent disclosure of a $10 billion order for custom AI chips from a new, unnamed client, which is now strongly implied to be OpenAI. For Broadcom (AVGO), this represents a material validation of its custom silicon strategy and a substantial, locked-in revenue stream from a leading AI entity. Conversely, for Nvidia (NVDA), this development is a notable long-term competitive threat, demonstrating that its largest customers are actively pursuing in-sourcing to control costs and performance, potentially signaling a trend that could erode market share and pricing power in the future. The plan for OpenAI's chip to be used exclusively in-house, coupled with its goal of surpassing one million GPUs by late 2025, underscores the immense scale of its computational needs and the strategic imperative behind this vertical integration.

Key Decisions for Investors

• Investors should view this development as a significant positive catalyst for Broadcom (AVGO), as the implied $10 billion order materially de-risks future revenue and solidifies its position as a key partner for custom AI silicon.
• For those holding Nvidia (NVDA), this news introduces a credible long-term risk factor; it may be prudent to monitor for similar in-sourcing announcements from other major AI and cloud computing firms, as this could signal a pattern of market share erosion.
• The trend toward custom silicon for AI workloads is now validated by a major player, suggesting investors should evaluate other semiconductor firms with strong custom design and manufacturing capabilities as potential beneficiaries of this industry shift.