OpenAI to launch its first AI chip in 2026 with Broadcom, FT reports

OpenAI is reportedly set to launch its first in-house AI chip in 2026, developed in partnership with Broadcom, for internal use to reduce its reliance on external chip suppliers like Nvidia and diversify its computing power infrastructure. This strategic move aligns with efforts by other tech giants like Google, Amazon, and Meta to develop custom AI silicon to manage surging AI workloads and costs. Broadcom's recent forecast of significant AI revenue growth in fiscal 2026, stemming from a new, unnamed customer, further corroborates this development.

Analysis

OpenAI is advancing its strategy to reduce dependency on external chip suppliers by partnering with Broadcom (AVGO) to produce its first in-house AI chip, slated for a 2026 launch. This chip is intended for internal use, aiming to curb the substantial costs associated with training and running its AI systems, a move that directly challenges Nvidia's (NVDA) market dominance. The development is strongly corroborated by Broadcom CEO Hock Tan's recent disclosure of securing over $10 billion in AI infrastructure orders from a new, unnamed customer, and his forecast for 'significant' AI revenue growth in fiscal 2026. This strategic vertical integration follows a well-established trend among tech giants like Google, Amazon, and Meta, who are also developing custom silicon to manage surging AI workloads. For Broadcom, this partnership marks a significant win, positioning it as a key enabler in the custom AI chip space. Conversely, for Nvidia, it represents the materialization of a key risk: a major customer actively and successfully diversifying its supply chain.

Key Decisions for Investors

• Investors should view this as a significant positive catalyst for Broadcom (AVGO), as the implied multi-billion dollar order from OpenAI validates its custom silicon strategy and provides strong visibility into its fiscal 2026 AI revenue growth.
• For Nvidia (NVDA) shareholders, this news serves as a concrete data point on the long-term competitive risk from hyperscale customers developing in-house silicon, which could gradually erode market share and pricing power.
• The trend of major AI players developing custom chips is accelerating, suggesting that portfolio allocation within the semiconductor sector should consider not only dominant platform players but also key design and manufacturing partners like Broadcom and TSMC.