South Korea, under President Lee Jae-Myung, is making a massive geopolitical and economic push to become one of the top-three global leaders in AI and semiconductors by 2030. The country’s sovereign AI push got a huge boost this week with Nvidia CEO Jensen Huang’s second visit to South Korea this year, and subsequent announcements yesterday to provide Blackwell and Vera Rubin HBM chips as an architectural foundation for not only South Korea’s sovereign AI plans, but also its emerging physical robotics market.

In announcing AI factories in South Korea and deals with not only SK Group, but also SK Hynix and Naver, Huang emphasized that “Telecom networks are becoming national AI infrastructure. They connect people, companies, devices and machines — and now they can become the backbone of new AI clouds.”

At the heart of his SK Group deal is the Nvidia DSX full-stack reference architecture, which will power a GW-scale AI Cloud for SK Telecom, and align with Chairman Chey Tae-won’s pronouncement that SK Group will be the “leading AI factory player shaping Asia’s AI ecosystem.” With GPU-based cloud computing, Tae-won wants to accelerate time to first production and dramatically improve per-MW token performance for sovereign, physical and agentic AI services.

With its first AI factory projected to come online by 2027, SK Group is targeting the burgeoning South Korean market of AI startups, robotics, and physical AI companies. Its partnership with Nvidia will focus on building scalable, resilient AI services across industry and enterprise, with both companies announcing plans to jointly research and codevelop next-gen AI infrastructure for full-stack AI factory optimization.

These announcements dovetail into last week’s unveiling during Computex in Taipei of Nvidia’s Vera Rubin AI computing platform and its Vera CPU. This week, South Korea represents a significant deployment phase for those Nvidia technologies and platforms.

As you’ll see below in “Top Stories,” South Korea isn’t the only country aggressively pursuing sovereignty, as France and the UK are among a myriad of European countries asserting more independence in AI infrastructure to further protect against the volatility of geopolitical changes worldwide.

€20.35bn SFR “sovereignty” deal: Pending regulatory approval, French telecoms have agreed among themselves to shrink from 4 to 3, with Bouygues Telecom, Orange, and Free-Iliad Group acquiring Altice France’s SFR operations. 

UK ‘AI hardware plan’: The UK government is pumping £1bn into AI infrastructure; AMD and Nebius have committed £4bn; Oriole, Cosine, and PhysicsX are pursuing sovereign frontier models, GPU photonics, and specialist AI digital twins.

Nebius CEO defines growth opps: Managed inference customers increasingly prefer consuming models and inference services rather than directly managing infrastructure, said CEO Roman Chernin when describing new opportunities.

Corning – Amazon optical fiber deal: Corning struck a multibillion-dollar deal to  supply the optical fiber, cable, and connectivity for Amazon’s data center infrastructure – a follow-up to Meta’s $6 billion deal with Corning earlier this year.

How AI has transformed CBRE: Fortune explores how the multibillion dollar AI data center boom transformed CBRE into a key behind-the-scenes player–something RCRTV explores in recent interview with Pat Lynch and Gordon Dolven.

Blackstone’s newest AI bet: Blackstone Digital Infrastructure Trust drew mixed reviews from Wall Street, with coverage initiated for the newly listed data center REIT, and RBC Capital Markets citing strong potential from AI-driven infrastructure.

Broadcom Private Cloud report: Broadcom’s Private Cloud Outlook 2026 reveals  56% of enterprises run or plan to run production AI inferencing on private cloud, with the data protection and privacy (37%) placing higher demands on IT.

Applied Digital funding: Applied Digital closed a $350 million revolving credit facility (with an additional $200 million accordion option) arranged by Goldman Sachs to accelerate its data center development.

It’s hard to keep up as chips go vertical: Advanced-node manufacturing is pushing semiconductor inspection beyond traditional top-down measurement and toward reconstructing buried 3D structures that conventional tools can’t easily see.