NVIDIA and SK Group Strike Over $500 Billion AI Chip Alliance as Jensen Huang Predicts 10x Semiconductor Boom

The Scale and Scope of the NVIDIA–SK Alliance

NVIDIA and South Korea's SK Group have assembled what CNYES (news.cnyes.com) described as a "$500 billion collaboration map," built around AI chips and infrastructure (E2). Inside (inside.com.tw) reported the more specific breakdown: future cooperation between NVIDIA and SK Group in memory procurement, AI supercomputers, and data-center construction is projected to surpass $500 billion (E3). Both outlets converge on the same $500 billion figure, though CNYES frames it as the overall partnership "map" while Inside ties the number specifically to the memory, supercomputer, and data-center workstreams.

NVIDIA and SK Hynix (SK海力士): Multi-Year Memory Chip Commitments

Within that alliance, Inside reported that NVIDIA has committed to purchasing large volumes of memory chips from SK Hynix (SK海力士) for several consecutive years, while jointly advancing memory technology generations from HBM2 and HBM3 through HBM3E, and on to HBM4 and HBM4E (E4). This multi-year, multi-generation commitment is the memory-supply backbone of the broader $500 billion figure cited above (E3).

SK Telecom's AI Cloud Infrastructure Buildout

Inside also reported that SK Telecom separately plans to expand its AI cloud infrastructure to as much as 2 gigawatts, with NVIDIA supplying the corresponding supercomputer systems — a structure Inside characterized as a "buy-and-sell" two-way arrangement between the two companies (E5). This SK Telecom expansion sits alongside, and is distinct from, the SK Hynix memory-supply commitments described above (E4), showing the alliance spans both memory chips and cloud compute buildout.

Jensen Huang's Basis for a 10x Semiconductor Growth Forecast

CNYES's headline captured Huang's core claim directly: "AI will expand the semiconductor industry to 10 times its current scale" (E1). Inside reported the same forecast with more detail, stating that Huang expects this trend to drive the global semiconductor industry to expand to 10 times its current size within the next decade (E6). Huang's own $500 billion NVIDIA–SK partnership (E2, E3) — spanning memory, supercomputers, and data centers — is the concrete deal he points to as evidence of the AI-driven expansion he is forecasting.

Resource Bottlenecks Facing the AI Chip Industry

Despite that bullish 10x forecast, Huang acknowledged real constraints. Inside reported his admission that resources including high-bandwidth memory (HBM), land, power, and data-center construction are all currently strained, and that even with the industry pushing at full capacity, doubling production capacity annually is already difficult — further acceleration beyond that pace would be considerably harder (E7). This is Huang's own characterization of the bottleneck, not an independent measurement, but it comes from the same executive making the 10x growth call.

Global AI Talent: How Regions Compare

On the human-capital side of the AI buildout, Inside reported Huang's remark that China trains more AI research talent each year than the rest of the world combined, and that the United States must therefore keep learning and stay open to collaboration in order to maintain its edge in the competition (E8).

Key Figures at a Glance

What This Means

The same executive forecasting a 10x expansion of the global semiconductor industry within a decade (E1, E6) — backed by a concrete $500 billion NVIDIA–SK partnership spanning memory, supercomputers, and data centers (E2, E3) — also acknowledges that HBM, land, power, and data-center construction resources are already strained, with annual capacity doubling described as difficult to sustain, let alone accelerate (E7). Layered on top of that physical bottleneck is a human-capital dimension: Huang's own comparison of China's AI research talent output to the rest of the world combined (E8) suggests the constraints he sees are not limited to chips and power, but extend to the talent base needed to build them.