OpenAI has acquired a set of patents from chip startup Rain AI after negotiations to purchase the company outright fell through. The announcement did not disclose the purchase price or the number of patents involved.

Rain AI previously received personal investment from Sam Altman, CEO of OpenAI. The patent acquisition comes as OpenAI simultaneously defends itself in court against Apple's allegations of misappropriating hardware trade secrets, filing a motion this week to dismiss the suit.

Failed Acquisition, Valuable IP

OpenAI and Rain AI were unable to reach an agreement on a full acquisition, but OpenAI walked away with a portion of Rain AI's patent portfolio. The outcome is bleak for Rain AI: the startup has lost nearly all of its employees and operations have largely ground to a halt. No employee transfer from Rain AI to OpenAI has been announced, according to The Information.

OpenAI has not publicly disclosed what it intends to do with the acquired patents. Acquiring intellectual property without the originating team is unusual in semiconductor deals, where the value often lies in the engineers who designed the technology. Without the inventors, patents can still serve as defensive assets against infringement claims or as building blocks for future in-house designs.

The Altman Connection

Founded in 2018, Rain AI specialized in neuromorphic processing units — chips designed to mimic features of the human brain in order to reduce energy consumption and lower the cost of training AI models. Neuromorphic computing has attracted interest from researchers and national labs as a potential path to dramatically more efficient AI hardware, though commercialization has proven difficult. Rain AI is not the only company to struggle in this space; Intel's own neuromorphic research program, Loihi, has remained largely experimental for years.

The relationship between OpenAI and Rain AI dates back several years. In 2019, OpenAI signed a letter of intent (LOI) to purchase $51 million worth of chips once Rain AI began production. Those chips were never delivered.

Rain AI eventually exhausted its funding and lost its customer base. Around 2025, the company began seeking a buyer. A planned $150 million Series B fundraising round failed to materialize.

Apple Lawsuit Continues

Apple's lawsuit against OpenAI alleges that former Apple engineers brought confidential hardware information with them to OpenAI. Apple claims that approximately 400 of its former engineers are now employed at the AI company, according to TechCrunch.

OpenAI filed a motion to dismiss, arguing that Apple's own practices created the situation. OpenAI contends that Apple permitted employees to work through personal iCloud accounts and failed to revoke access after they departed, a condition OpenAI describes as "residual access." In its official response, OpenAI pushed back against the allegations.

Apple has not yet responded to the motion to dismiss, and the court has not issued a ruling. The case highlights the competitive tensions that arise as AI companies recruit aggressively from hardware incumbents.

Both the Apple litigation and the Rain AI patent acquisition underscore OpenAI's broader push into hardware development, particularly custom silicon for AI-powered devices.

OpenAI's Custom Chip Strategy

OpenAI has been steadily advancing its custom chip program. In June, the company introduced Jalapeno, a purpose-built inference chip developed in collaboration with Broadcom, claiming the design would cut inference costs by half.

The strategy aligns with a wider trend among major AI labs in the United States seeking to reduce reliance on Nvidia GPUs, which currently dominate AI training and inference. Nvidia's GPUs have become one of the most expensive line items for AI companies, with demand consistently outstripping supply. Anthropic and Meta have also recently announced initiatives to develop their own silicon. Google has operated its own Tensor Processing Units for years, illustrating both the potential cost savings and the multi-year commitment required to bring custom AI silicon into production.