Q&A with Sam Altman on OpenAI's Stargate data center project in Saline, Michigan, coding models as the biggest AI demand driver, falling token prices, and more
Following is the unofficial transcript of a CNBC exclusive interview with OpenAI CEO Sam Altman on CNBC's “Power Lunch” (M-F, 2PM-3PM ET) today, Monday, June 1.
Context & Ripple Effects
OpenAI’s Stargate effort follows a long-running infrastructure push: earlier coverage described Altman discussing multibillion-dollar AI infrastructure spending and, more recently, a possible compute company majority-owned by OpenAI. The Saline project makes that strategy more concrete at the data-center level.
The interview also connects infrastructure to product demand. Previous coverage has emphasized OpenAI’s enterprise strategy, while Altman now identifies coding models as the largest demand driver, tying capacity planning to a specific high-value workload rather than general chatbot use.
First-order effects
- OpenAI gains a named physical expansion path for Stargate in Saline, while its compute strategy becomes more central to serving coding-model demand.
- Falling token prices can make OpenAI’s models more accessible to developers and enterprises, especially where coding usage is intensive.
Second-order effects
- Lower model costs and more available capacity raise pressure on rival AI providers to compete on coding performance, reliability, and unit economics rather than only general-purpose chat features.
- Enterprise software buyers can test broader coding-model deployments as inference becomes cheaper, increasing the importance of infrastructure access for vendors serving those customers.
Third-order effects
- If coding remains the leading source of demand, AI infrastructure investment is likely to concentrate around providers able to finance, secure, and operate large-scale compute capacity.
- The combination of falling token prices and expanding dedicated capacity points toward AI models becoming a more embedded input to software production, though the durability of that shift depends on sustained enterprise adoption and economics.
The trend: This is one data point in AI’s shift from consumer-chatbot competition toward infrastructure-backed, lower-cost coding automation for enterprise use.