Nvidia Boosts Buyback: $150 Billion Power Move!

# Weekly Nvidia News Brief
September 24–30, 2026

Nvidia’s key moves this week centered on financial strength and AI agent security, building on themes from recent coverage. These steps reinforce the company’s full-stack approach to scalable, reliable AI infrastructure.[1]

Record Share Buyback Authorization

Nvidia’s board approved an additional $150 billion for its share repurchase program, bringing the total remaining authorization to $235 billion—the largest increase in company history. The program runs through fiscal 2028 and reflects robust cash flow from AI demand.[2]

Nvidia’s bigger vision: This demonstrates confidence in sustained growth, allowing continued heavy investment in next-generation chips, software, and AI factories while returning capital to shareholders.[3]

Impact on AI overall: A financially robust leader like Nvidia supports ongoing industry-wide R&D in hardware, networking, and platforms that underpin model training, inference, and real-world AI deployment.

Open Agent Safety Platform Launch

Nvidia introduced the Open Agent Safety Platform, featuring open-source software (including OpenShell) and a reference hardware design (Sentry) for monitoring and controlling autonomous AI agents across the full stack—from testing to deployment. It enables quick isolation of problematic agents and has drawn partners from tech, finance, and security.[4]

Nvidia’s bigger vision: Safety features are integrated into Nvidia’s hardware and software layers, supporting trusted, scalable “AI factories” and agent-based applications in data centers, robotics, and beyond.

Impact on AI overall: As agents grow more autonomous, open standards and shared tools reduce risks of unintended actions, fostering safer, more responsible adoption across the ecosystem.[5]

Additional Notes from September 30

Applications opened for the 2027–2028 NVIDIA Graduate Fellowship Program (up to $60,000 per award, with mandatory summer internship), continuing support for PhD research in accelerated computing and AI.[6] CoreWeave announced production availability of NVIDIA Vera Rubin NVL72 systems (with Cognition as the first customer) and expanded offerings including the Vera CPU for agent workloads.[7]

These updates highlight ongoing ecosystem momentum but largely extend prior topics. No other major new developments emerged in the past seven days.

Sources: NVIDIA Newsroom (releases dated September 28 and 30, 2026), related coverage from Motley Fool, Bloomberg, and other outlets.