OpenAI is making its own AI chips with Broadcom's help

OpenAI has partnered with Broadcom to develop custom AI chips and systems, securing an additional 10 gigawatts of compute power in a multi-billion dollar deal, with deployment expected from H2 2026. This agreement is part of OpenAI's aggressive strategy to scale its AI infrastructure, following recent significant deals with NVIDIA, AMD, and Oracle, which collectively add substantial compute capacity. CEO Sam Altman's ambitious goal of reaching 250 gigawatts of compute power within eight years, estimated to cost $10 trillion, highlights the company's massive capital expenditure requirements and the need for innovative financing, far exceeding its current revenue projections.

Analysis

OpenAI has significantly expanded its compute infrastructure strategy, partnering with Broadcom for custom AI chips and systems, securing an additional 10 gigawatts (GW) of compute power in a multi-billion dollar deal. This follows recent substantial agreements with NVIDIA (10GW, $100B investment), AMD (6GW, tens of billions, potential 10% stake), and Oracle (4.5GW), indicating an aggressive push to scale AI capabilities. Deployment for these deals is largely slated for the second half of 2026, with Broadcom's rollout completing by 2029.

CEO Sam Altman's ambitious goal to achieve 250 GW of compute power over the next eight years, a significant leap from the projected 2 GW by end-2024, underscores OpenAI's long-term vision but highlights immense capital requirements. This target is estimated to cost around $10 trillion, contrasting sharply with OpenAI's expected 2024 revenue of $13 billion and current deal commitments totaling hundreds of billions.

The substantial demand from OpenAI represents a significant revenue opportunity for its chip and infrastructure partners, including Broadcom (AVGO), NVIDIA (NVDA), and AMD, as reflected by their positive per-ticker sentiment. However, OpenAI's stated need for "new financing tools" to bridge the $10 trillion funding gap introduces considerable financial uncertainty for its long-term growth trajectory and the sustainability of such massive infrastructure build-outs.

Key Decisions for Investors

• Investors should monitor the execution timelines and financial implications of OpenAI's massive compute infrastructure build-out, particularly the development of "new financing tools" given the $10 trillion funding gap.
• Consider the long-term revenue potential for key suppliers like Broadcom (AVGO), NVIDIA (NVDA), and AMD, as OpenAI's aggressive expansion directly translates to significant orders and market share for these companies.
• Evaluate the broader market impact on AI infrastructure demand and the potential for increased competition or consolidation among chipmakers and cloud providers as AI compute requirements continue to escalate.