AI spending has stalled even as usage rises, according to new data from Ramp. Token spend grew just 1.5% across August and September, against 42% month-on-month growth in July. Ramp's lead economist, Ara Kharazian, blames price cuts and more thrifty customers rather than weak demand. OpenAI halved API prices for GPT-6 Sol and GPT-6 Luna in September. Anthropic also cut prices, which it said in a filing reduced revenue growth by about $600 million. The top 1% of firms, spending close to $10,000 per employee, have eased off slightly. Median firms spend $12-$15 per employee per month. Bloomberg reported Anthropic's potential valuation fell from $2 trillion to $1.2-$1.3 trillion as it nears an IPO. OpenAI cut its proposed valuation to $1.25 trillion from $1.4 trillion.

OpenAI has pledged to pay for the electricity and grid infrastructure at data center campuses it owns and operates. It announced the move on November 5, 2026, two days after voters favoured restrictions or rejected pro-development proposals in 40 of 41 local data center measures across 10 states. OpenAI will also post financial guarantees before major spending begins, to protect other utility customers if a project is delayed, reduced, or cancelled. The pledges are voluntary, legally unenforceable, and name no dollar figure. The pledge may not cover leased sites, such as those supplied by Oracle. OpenAI will reportedly pay Oracle about $30 billion a year under a 4.5-gigawatt deal, while Oracle reported free cash flow of negative $5.4 billion in its fiscal first quarter.

OpenAI on Thursday released GPT-6 Cyber, a cybersecurity-tuned model it says is better at finding and exploiting vulnerabilities while staying within researchers' rules of engagement. It scored 48.5% on the ExploitGym evaluation, against 32.1% for GPT-5.6 Cyber. In a separate test, the model avoided exploiting out-of-scope services over 99% of the time. OpenAI declined to detail the 0.07% of cases that went beyond scope. Researchers used it to find 81 previously unknown high-severity vulnerabilities in Firefox. It also helped develop working exploits for 948 Linux kernel vulnerabilities, 588 of them previously unknown. OpenAI rates the model "Critical" under its Preparedness Framework, its highest cyber risk tier. Access is limited to approved Daybreak Red customers, with wider access for that group over the coming weeks.