OpenAI is pushing its artificial intelligence technology beyond software after revealing that it has been using its own AI models to help design specialized computer chips. The company says its internally developed Jalapeno chip reached the “taped out” stage within nine months, marking a significant step in its effort to apply AI directly to semiconductor engineering.
OpenAI Moves Deeper Into Semiconductor Development
The disclosure, made by OpenAI CFO Sarah Friar at the Goldman Sachs Communacopia + Technology Conference, illustrates how AI companies are increasingly looking beyond conventional software applications.
Chip design is traditionally a highly specialized process involving large engineering teams and lengthy development cycles. OpenAI’s use of its own models in that process suggests that AI could increasingly become a tool for accelerating the development of the hardware required to run future AI systems.
Reaching the tape-out stage means the design has progressed far enough for manufacturing preparation. While this does not mean a finished commercial processor is already available, it represents an important milestone in the chip-development process.
AI Could Change How Chips Are Designed
OpenAI’s approach reflects a larger shift in semiconductor engineering. As chip complexity increases, companies are exploring AI-assisted methods to optimize architectures, identify design problems and accelerate portions of the engineering workflow.
For OpenAI, there is an additional strategic incentive. The company’s AI models require enormous amounts of computing power, meaning improvements in chip efficiency could eventually have a direct impact on the economics of running its services.
The company is also expanding AI applications into areas including life sciences and financial services, suggesting that it sees specialized, outcome-focused AI systems as a major part of its enterprise strategy.
Enterprise AI Becomes a Bigger Focus
OpenAI’s comments also point to the company’s growing emphasis on enterprise customers.
Friar said enterprise revenue grew 32% from June to July, outpacing the company’s overall annualized revenue growth of 20%. OpenAI’s coding tool Codex has also reached 25 million users, highlighting the growing role of AI-assisted software development.
The company has been pushing lower-cost AI models as well. According to Friar, OpenAI reduced the price of its Luna model by 80%, which subsequently drove a tenfold increase in usage.
The strategy reflects a broader reality in the AI industry: as models become more capable, companies need to make them cheaper and easier to deploy if they want businesses to use them at scale.
Hardware and Software Are Converging
OpenAI’s chip-design efforts could eventually give the company greater control over part of its technology stack.
The AI industry has traditionally depended heavily on specialized hardware suppliers, particularly Nvidia. Developing or influencing custom silicon could give large AI companies another way to optimize computing costs and performance.
That does not necessarily mean OpenAI is attempting to replace Nvidia. Instead, the Jalapeno project highlights an emerging model in which AI companies use AI to develop the hardware that powers AI itself.
As competition intensifies among OpenAI, Anthropic and developers of open-weight models, controlling more layers of the computing infrastructure could become increasingly important. OpenAI’s move into chip design suggests that the next phase of the AI race may extend well beyond models and applications into the underlying semiconductor technology.
This website uses cookies.