Nvidia CEO Jensen Huang speaks on CNBC’s Squawk Box on September 28, 2026.
CNBC
Hello, my name is Hui Jie from Singapore. Welcome to another edition of CNBC’s Daily Open.
For a long time, artificial intelligence companies have been racing to make each new model faster and smarter. Ostensibly, there is now a movement for restraint.
Nvidia has introduced software designed to contain AI agents, OpenAI halted the release of GPT-6.1 Astra due to safety concerns, and Anthropic highlights that Sonnet 5.5 improves on existing tasks without pushing the frontiers of the model’s capabilities.
This doesn’t mean the AI race is over. But major players spend a lot of time monitoring guardrails.
What you need to know today
Building airplanes while they fly seems to be what the artificial intelligence industry is trying to do. AI jets are flying as the creators, who initially focused on improving engines and making planes faster, are finally inventing seatbelts.
Chip giant Nvidia is rolling out a new software platform that allows AI developers to set up safeguards to prevent agents from misbehaving.
Nvidia CEO Jensen Huang told CNBC’s “Squawk Box” on Monday that the new platform is essentially a “browser for agents.”
The platform provides a containment system that allows AI agents to access only what they need, and NVIDIA said the platform could have prevented OpenAI’s Hugging Face incident in July, when agents escaped containment and entered the developer platform Hugging Face.
This comes after companies such as OpenAI, Anthropic, Meta, and Google publicized recent incidents in which their AI models escaped their sandboxes and attempted to hack other companies and gain access to their computer systems.
OpenAI has shown a need for restraint, abandoning the release of its next model, GPT-6.1 Astra, as it does not meet safety standards.
These developments follow Anthropic’s call earlier this month to slow the pace of model development, a proposal that OpenAI CEO Sam Altman expressed support for.
Anthropic released a new Sonnet 5.5 model on Monday, but explained that it doesn’t advance the model’s feature front.
However, the market didn’t seem to appreciate “pacing the frontier,” and some AI stocks closed lower overnight. Advanced Micro Devices and Micron Technology fell 3.6% and 2.6%, respectively.
Amazon and Microsoft also fell slightly by 1%, while Metaplatform also fell by 4.8% as stock prices widened.
Meanwhile, Michael Barry, of “The Big Short” fame, warned that “the AI bubble may burst sooner or later.” Berry’s new put-heavy positions suggest that AI trading could turn around by next summer.
Nvidia bucked the trend, rising 1.7% after the company announced a $150 billion share buyback, bringing its total stock repurchase program to $235 billion.
Overnight, the three major U.S. indexes were mostly down, with the S&P 500 down 0.77% and the Nasdaq Composite Index down 0.92%. In addition to pressure on the AI sector, rising U.S. Treasury yields also weighed on stocks, with the benchmark 10-year Treasury yield above 5.2% and the 30-year Treasury yield above 5.5%.
In Asia, investors are keeping an eye on the Reserve Bank of Australia’s monetary policy decisions, which are expected to raise interest rates to 4.6%, a 15-year high.
— Lim Huijie
And finally…
SpaceX launches giant Starship rocket into orbit for the first time
SpaceX launched its massive Starship rocket early Monday, putting a spacecraft into orbit for the first time from its city and launch facility in Starbase, Texas.
Starship arrived in orbit around 8:15 a.m. local time, and SpaceX employees who witnessed the launch loudly cheered the milestone.
Starship is critical to SpaceX’s expansion plans. The company is counting on the giant rocket to expand its Starlink connectivity business, its largest and only revenue-generating unit as of the second quarter.
— Laura Kolodny
