OpenAI is expanding artificial intelligence deeper into specialised industries, including semiconductor design, life sciences and financial services, as competition intensifies over how companies turn generative AI into measurable business value.
Chief Financial Officer Sarah Friar said the company was increasingly focusing on AI systems designed for specific enterprise tasks rather than relying solely on general-purpose models.
Speaking at the Goldman Sachs Communacopia + Technology Conference, Friar said OpenAI was also experimenting with pricing models tied more closely to business outcomes as corporate customers demand clearer returns on increasingly large AI investments.
AI Helps Design OpenAI’s Own Chip
One of the strongest examples of the company’s strategy is its first custom inference chip, Jalapeño.
Friar said OpenAI used its own AI models to help develop the processor, moving from the initial design stage to tape-out — when a chip design is finalised and sent for manufacturing — in just nine months.
OpenAI has previously said AI accelerated implementation, measurement and verification work during the chip’s development and helped optimise arithmetic circuits.
The company developed Jalapeño with Broadcom as part of a broader effort to build specialised hardware optimised for large language model inference.
That strategy matters because computing infrastructure has become one of the biggest constraints facing AI companies.
Developing custom processors could reduce dependence on third-party accelerators while giving AI developers greater control over power consumption, performance and deployment costs.
OpenAI Targets Cheaper AI Deployment
Cost is emerging as another major battleground.
Friar said OpenAI recently cut the price of its lower-cost Luna model by 80 per cent, contributing to roughly a tenfold increase in usage.
She also said the company’s Codex coding platform had reached 25 million users. Enterprise revenue increased 32 per cent between June and July, compared with 20 per cent growth in overall annualised revenue during the same period.
According to Friar, enterprise and consumer revenue had reached roughly equal proportions by the middle of 2026, earlier than OpenAI had initially expected.
The company is also positioning its lower-cost models against Chinese open-weight competitors.
Friar argued that deploying Luna could in some circumstances be cheaper than operating models such as Z.ai’s GLM 5.3 through third-party cloud infrastructure.
AI Race Moves Beyond Chatbots
The development reflects a broader shift in the artificial intelligence industry.
Competition is moving beyond who can produce the most capable chatbot or foundation model. Companies are increasingly competing across chips, data centres, enterprise software, specialised industry applications and developer tools.
OpenAI’s strategy now spans those layers, from custom hardware to models and business applications.
If the company succeeds in lowering both development and deployment costs, it could strengthen its position against rivals such as Anthropic, Google and Chinese AI developers that have gained attention for cheaper open-weight alternatives.
The larger question is whether specialised AI can deliver productivity improvements significant enough to justify continued corporate spending.
