Nvidia is everywhere
If today had a center of gravity, it was Nvidia. The chipmaker turned up in deal after deal, underscoring how thoroughly it now sits at the heart of the AI boom. CoreWeave, the AI infrastructure specialist, disclosed a staggering $99 billion backlog of future business, with commitments tied to some of the biggest names in tech—Nvidia, Meta, and Microsoft among them. That figure is a vivid snapshot of how much money is being locked in for AI compute years ahead of demand.
The partnerships kept coming. Australia's Firmus Technologies struck a strategic, eight-year deal with Nvidia to build what it calls a Southeast Asian 'AI factory'—its first data center, rising in Batam, Indonesia. And Finland's Nokia signed a major partnership with Nvidia to develop 'AI RAN' technology for next-generation 6G networks, pushing artificial intelligence deeper into the telecom stack. Taken together, the three deals show Nvidia's reach extending from cloud backlogs to physical infrastructure to the networks themselves.
The talent and access wars
The competitive knives were also out. Paul Meade, the Apple vice president who led the Vision Pro headset and the company's smart glasses efforts, is leaving to join OpenAI's hardware team, per TechCrunch. It's a notable defection that signals just how seriously OpenAI is pursuing its own devices—and how the fight for top hardware talent is intensifying.
Meanwhile, the rivalry between AI giants turned territorial. According to a Financial Times report relayed by Bloomberg, Google has placed limits on Meta's access to its Gemini AI. The two sit among the largest players racing to build and deploy advanced models, and the move is a reminder that even cloud-scale partnerships have hard edges when competitors are involved.
Robots get smarter—and more independent
Robotics produced the day's most varied news. The contrarian take came from Agibot, whose chief scientist argued that robots won't get their 'GPT moment' by copying the large-language-model playbook, per KrASIA—a pointed dissent from the prevailing wisdom. WIRobotics, by contrast, leaned in, kicking off its 'physical AI' push with the ALLEX simulation model, the first piece of what it calls a development ecosystem.
There were practical wins, too. A South Korean factory run by HD Hyundai Electric is now hitting near-perfect on-time delivery thanks to a blend of AI and robots. And the desktop Reachy Mini robot gained the ability to hold conversations using AI that runs entirely on-device—no cloud required, a meaningful step for privacy and offline reliability. From China, robotics firm WhalesBot took a different tack entirely, betting on 'off-screen' AI learning that deliberately keeps children away from screens.
Quantum's quiet buildout
Finally, quantum computing is gathering momentum on several fronts at once. As Ynetnews frames it, new tools, fresh funding, and deepening corporate strategy are reshaping a field that was once purely academic—positioning quantum as the next frontier the industry is eyeing even as the AI race runs hot.
The through-line across all of it: the AI infrastructure being built today is enormous, expensive, and increasingly fought over—whether the battleground is a $99 billion backlog, a star engineer, or access to a rival's model.