OpenAI maintains its bet on data centers despite local resistance
OpenAI says it continues to invest in data centers despite increased local resistance in the United States. The company highlights demand, computing needs, and stricter commitments on water, electricity, and noise.
OpenAI is maintaining its investment strategy in data centers despite a surge of local opposition in the United States, explained its CFO Sarah Friar on November 6, 2026, at the Citi TMT Leadership Summit in Pebble Beach, California. The issue is straightforward: the company needs computing capacity to meet demand, continue training its models, and serve its customers. At the same time, it must deal with elected officials and residents who are more attentive to the tangible effects of a new campus, from electricity bills to noise. The matter goes beyond OpenAI; it now concerns how AI is established in local communities.
The demand continues, so do the servers
Sarah Friar emphasized one point: demand justifies construction. According to her remarks, OpenAI's business grew by 70% quarter over quarter in Q3 2026, enterprise activity doubled, and 1.2 billion people use its products weekly. She also indicated that revenue is split evenly between the general public and businesses. In this context, the company believes it must secure more computing power, as models and products do not run on goodwill alone.
Friar also confirmed that a funding round is underway. Her reasoning is straightforward: capital remains available for companies deemed the strongest, and OpenAI intends to raise money as long as demand continues. This financial logic intertwines with industrial deployment, maintaining relationships with major cloud infrastructure providers, the hyperscalers—that is, giants capable of supplying servers and storage at a global scale.
Communautés locales : le nerf de la guerre
The issue is no longer just technical. Following local elections on November 5, 2026, 41 ballot measures across 10 U.S. states saw voters support restrictions or reject development projects in all but one case, according to CNBC. Candidates favoring limitations on data centers also gained support. This sequence weighs on OpenAI’s expansion, as each new site must now convince far beyond investors.
Friar says she hears three recurring objections: the cost of electricity, water usage, and noise. OpenAI promises not to increase local bills and claims it can work with utilities to invest in the grid. The company also detailed examples cited in Georgia: a 300-foot-high noise barrier, free school meals, a park, and an AI summit for small businesses. The message is clear: without local acceptance, servers may be silent on paper, but they always end up making noise in local politics.
Des engagements plus stricts sur l’eau et l’électricité
On November 5, 2026, OpenAI announced new commitments for its future owned campuses. The company promises local agreements, annual public reporting, full payment of electricity consumption and infrastructure costs, as well as site-by-site disclosures on water and electricity. It also says it wants to limit the use of confidentiality agreements with public officials. These measures aim to reduce mistrust around projects often perceived as opaque.
Regarding water, Friar described a large computing site as consuming about two Olympic swimming pools of water, with circulation and reuse on site. According to her, a large part of the consumption also comes from people working on site. However, for communities, the issue is not theoretical: it concerns networks, aquifers, and urban planning decisions, immediately.
Une trajectoire publique inchangée, malgré la pression
Friar said OpenAI is not abandoning its plan for an IPO in 2027. She also specified that the midterm elections do not change the company’s position. The company is therefore advancing on two tracks simultaneously: financing its expansion and building campuses that do not trigger lasting conflict with residents and elected officials. It is less spectacular than a product launch but much harder to manage over time.
In this context, available computing power also conditions the product roadmap. Friar explained that OpenAI launched its professional virtual assistant first in enterprises because business uses show their value faster, security is more demanding there, and computing constraints require prioritization. The consumer version is to follow later. For OpenAI, the real question is no longer just what AI can do, but where it can be installed without triggering an outcry.
Key points
- OpenAI maintains its investments in data centers.
- Sarah Friar spoke on November 6, 2026, in Pebble Beach.
- Business grew by 70% in Q3 2026.
- 1.2 billion people use the products weekly.
- 41 local measures have tightened restrictions in 10 states.
- OpenAI still aims for a 2027 IPO.
By the numbers
- 70% — OpenAI’s quarterly business growth in Q3 2026, according to Sarah Friar.
- 1.2 billion — weekly users of OpenAI products, according to the company.
- 41 — local measures related to data centers in 10 U.S. states, on November 5, 2026.
- 300 feet — height of the noise barrier mentioned for a site in Georgia.