OpenAI's Business Grew 70% in Q3 as Data Center Backlash Builds

By Olivia Hart

OpenAI's business grew 70% quarter-over-quarter in the third quarter of 2026, and its enterprise business doubled, Chief Financial Officer Sarah Friar told CNBC in an interview at Pebble Beach, California. She framed the numbers as proof that demand for artificial intelligence keeps accelerating — even as midterm election results showed voters turning against the data centers that make that growth possible.

The timing is awkward. During the recent U.S. midterms, candidates campaigning on proposals to pause data center construction won several races, including contests previously held by Republicans. The results raise a concrete question: can the industry's expansion plans survive local opposition?

What do the usage numbers actually show?

Friar said approximately 1.2 billion people worldwide use OpenAI's services every week — roughly one in seven people on the planet. Of those, about 300 million engage with the platform for health and wellness activities, and 200 million use it for personal finance tasks. Those figures stretch well beyond chatbots answering search queries.

Enterprise adoption is also broadening. JPMorgan Chase joined OpenAI as a customer in the third quarter, with particular interest in cybersecurity applications. Friar said organizations from small businesses to major corporations are deploying the technology across business functions.

The takeaway: usage is growing fast, but OpenAI disclosed no revenue figures, profitability data or cash-burn numbers alongside the growth rate. A 70% quarterly growth claim without a revenue base tells investors little on its own.

How does OpenAI plan to fund the buildout?

Friar said OpenAI is building a durable business capable of supporting long-term investments in AI infrastructure. The company expects to continue raising capital and deepening relationships with major cloud computing providers, which supply the computing capacity required to train and operate advanced AI models.

She did not specify how much capital OpenAI intends to raise, on what terms, or from whom. The funding question matters because data centers, computing equipment and electricity infrastructure carry costs that scale with usage.

Can community engagement defuse the opposition?

Friar acknowledged local concerns about electricity prices, water consumption and noise, and said OpenAI must engage directly with communities before developing major infrastructure projects. Her approach: listen to residents, establish plans addressing local concerns, and remain accountable for results.

On electricity, she said OpenAI has committed to ensuring its infrastructure projects do not increase local residents' electricity bills. She also suggested partnerships with utilities could fund improvements to local electrical grids.

On water, Friar described a cooling approach involving water equivalent to roughly two Olympic-sized swimming pools, transported to a facility and circulated through systems resembling automotive radiators. She argued much of the water remains within the cooling system rather than being continuously consumed. That is a description of one design, not an independent audit of water use.

On noise, Friar described visiting a community in Georgia to discuss a proposed data center, including measures such as a substantial surrounding berm.

OpenAI is also spending on community programs. In Georgia, Friar highlighted free school breakfasts and lunches, development of a public park, and an AI summit for small businesses. She said participation in school meal programs rose approximately 70% after the company's involvement.

CNBC pressed Friar on whether this engagement strategy arrived too late, given the election results. Friar responded that OpenAI had maintained a consistent approach to community involvement and argued that direct engagement often produces a more nuanced understanding of local concerns than election outcomes alone.

The caveat is structural: school lunches and parks come after siting decisions, not before. Voters have already signaled resistance in several districts, and goodwill programs have not yet been tested against a contested project.

Does AI actually destroy jobs?

Friar pointed to economic indicators — improving productivity, rising wages and lower unemployment — as evidence that AI adoption does not necessarily translate into immediate job displacement. She argued AI expands employees' capabilities beyond their traditional responsibilities.

OpenAI's internal research found that approximately 40% of work-related ChatGPT usage involves tasks extending beyond an individual's formal job description. Friar cited her own finance organization, where employees now perform software development, marketing and other functions that would not traditionally fall within finance roles.

She also named Cisco, which she said achieved a 20% improvement in the speed of developing its product roadmap after deploying OpenAI's technology. Internal usage data and customer anecdotes are not labor-market evidence, and Friar did not address projections of longer-term displacement.

What comes next

Friar's comments sketch a strategy: keep raising capital, keep engaging communities, and keep arguing that AI creates measurable economic value. The midterm results suggest the political environment will make each data center harder to permit — and that 70% quarterly growth will be tested against electricity grids, water tables and voters who have not yet been persuaded.