OpenAI offers AI for chip design, touts cost advantage over open-source, CFO says

By Krystal Hu

Sept 8 (Reuters) San Francisco - OpenAI is pushing its AI into specialized industries and undercutting open-source rivals on cost, Chief Financial Officer Sarah Friar said on Monday, as the ChatGPT maker sees accelerating enterprise growth.

Speaking at Goldman Sachs' Communacopia + Technology Conference in San Francisco, Friar said OpenAI is focusing on sectors including chip design, life sciences and financial services, as businesses increasingly seek AI systems tailored to specific tasks. It also is experimenting with pricing based on business outcomes rather than use, as more enterprises demand return on investment in AI. 

The push into industry-specific AI and aggressive pricing reflects the growing competition OpenAI faces from Chinese open-weight models and rivals such as Anthropic, as enterprise customers demand more measurable returns from AI spending.

Friar said OpenAI had a positive experience of its own, using its own models in developing its Jalapeno chip, which was "taped out", a stage when a chip design is finalized and sent to a factory ​for production, within nine months.

Open-source and open-weight models are widely seen as a cheaper alternative to frontier models from OpenAI and Anthropic. Friar said OpenAI recently cut the price of its lower-cost Luna model by 80%, helping drive a roughly 10-fold increase in usage. OpenAI's Codex, its coding tool, is attracting customers and has 25 million users, she said.

Friar said enterprise revenue increased 32% from June to July, compared with 20% growth in overall annualized revenue during the period. Enterprise and consumer businesses had reached roughly an even split by the middle of the year, ahead of OpenAI's target of reaching that balance by year-end, she said.

Deploying OpenAI's lower-cost model could be cheaper than running Chinese open-source alternatives through cloud providers, she added.

"If you're deploying Luna and compare that to (Z.ai's) GLM 5.3, for example, on a cloud layer, we are cheaper," Friar said.

(Reporting by Krystal Hu in San Francisco; editing by Peter Henderson and Lincoln Feast.)