OpenAI CEO Discusses Compute Cost Headwinds and Model Efficiency Gains
In an interview with CNBC on 9 July 2026, OpenAI chief executive Sam Altman acknowledged that the soaring costs of compute and memory are "definitely a headwind" for the company’s rapid expansion. He highlighted that the financial strain from massive infrastructure spending is a key challenge as OpenAI strives to stay at the frontier of generative AI.
Altman also announced a significant technical breakthrough: the newest OpenAI model delivers a 54% improvement in token efficiency specifically for agentic coding applications, representing a major gain in performance per compute unit.
Addressing partnership rumors, Altman confirmed that Microsoft (NASDAQ:MSFT) remains one of OpenAI’s largest and most critical customers, dispelling speculation of any deterioration in the relationship. When questioned about reports of a potential U.S. government equity stake, he rejected the claim, stating there are "a lot of inaccuracies" and explicitly denying a rumored 5% government ownership.
Altman noted that OpenAI has made "many changes" after extensive discussions with Washington officials and expressed optimism that future interactions with the government will be smoother and more collaborative.
Regarding a possible initial public offering, Altman was non‑committal when asked about a 2026 IPO, replying simply, "I don’t know."
On the geopolitical front, he warned that Chinese open‑source AI models are becoming "very good," underscoring rising global competition in the AI space.
The article, authored by Louis Juricic for Investing.com and sourced from Reuters, was published on 09‑07‑2026 at 08:12 pm under the Stock Market category.