Jensen Huang’s AI World Tour: What Nvidia’s New Deals With SK Hynix & Others Mean for Investors

Nvidia stock jumoed more than 1% today as tech stocks attempted a recovery after brutal bloodfath on Friday. At 92% data center revenue, Nvidia is the most concentrated trillion-dollar company in history. And since January, Jensen Huang has done something that looks paradoxical: signed deals with more than 15 companies — memory makers in Seoul, optical fiber producers in Texas, AI software companies in California, and assembly plants in Arizona. If the business is going perfectly, why the urgency?

Let's break down the real logic behind Nvidia's partnership surge, why the geography reads like a geopolitical map, and what it means for investors watching from India.

Source: Nvidia Q1 FY2027 Earnings Report (NASDAQ: NVDA), May 20, 2026

Why Customer Concentration Is Nvidia’s Biggest Strategic Risk

The data center segment grew 92% year over year, faster than Nvidia's total revenue. That is the good news. The harder news: two unnamed customers represented 36% of Nvidia's full-year FY26 revenue. If either one slowed AI capex, even briefly, the financial impact would show up immediately.

Huang is not signing 15 partnerships because business is struggling. He is signing them because when 92% of your revenue comes from one segment, and within that segment two customers are nearly 40% of your business, you build walls around the entire structure.

Every deal Nvidia has announced in 2026 maps to one of three objectives, and understanding those three objectives is the only way to make sense of why Jensen Huang was throwing a baseball pitch at a stadium in Seoul this past weekend.

Nvidia’s Partnerships Are About Advantage Building, Not Diversification

The easy comparison is to Google's diversification playbook, Sundar Pichai expanding search's flywheel into Maps, YouTube, Android, and Waymo to grow beyond one revenue stream before the core peaked. That framing is wrong. Nvidia's data center revenue grew 92% last quarter. There is no peak in sight.

Think of what Maruti Suzuki did when it built India's auto components ecosystem in the 1980s. Maruti did not just manufacture cars. It co-financed key suppliers, invited them to build factories next to its Gurgaon plant, and shared Suzuki's engineering specifications so components were designed specifically around Maruti's models.

Nvidia's version: invest billions in the companies that supply the components your chips depend on, engineer those components so deeply into your architecture that competitors cannot access the same supply chain at the same priority, and build a software ecosystem that runs best, sometimes only runs, on Nvidia hardware. 

This is , not diversification. The 15 partnerships are not a strategic pivot. They are a systematic attempt to make the 92% unassailable.

Nvidia’s AI Infrastructure Strategy

Layer 1: Supply chain security: Nvidia’s $2 billion investments in Lumentum and Coherent are about locking in critical optics supply. Both companies make laser components for co-packaged optics, the technology that connects thousands of GPUs inside Blackwell clusters. Once suppliers build capacity around Nvidia’s specs, serving rival GPU platforms becomes harder. The same applies to SK Hynix, which is engineering HBM4 memory around Blackwell’s interconnect architecture.

Layer 2: Platform lock-in: Marvell’s $2 billion deal ties its networking silicon to Nvidia’s NVLink Fusion, making large AI clusters work better inside Nvidia’s ecosystem. Microsoft extends that lock-in to PCs through RTX Spark, bringing CUDA, Nemotron and OpenShell natively to high-end Windows machines. ServiceNow and Adobe push the same stack into enterprise and creative workflows.

Layer 3: Market creation: IREN, SK Telecom and Naver are not just buying Nvidia chips. They are South Korean giants are building AI cloud businesses on Nvidia’s full technology stack. SK Telecom’s gigawatt-scale AI cloud in South Korea, with its first factory expected in 2027, turns future demand into committed infrastructure today.

Sources: Nvidia Newsroom, Reuters

Why Nvidia’s US, Taiwan and South Korea Partnerships Matter

8 US companies, 2 Taiwanese companies and 6 South Korean companies. This is not accidental. The split matters.

In the US, Nvidia’s optical investments in Corning, Lumentum and Coherent serve two goals.

• They secure critical components for next-generation AI clusters, where copper connections are being replaced by fiber and lasers. Without this, Nvidia’s own Blackwell roadmap could face physical bottlenecks. 
• They strengthen Nvidia’s US manufacturing story at a time when AI infrastructure is becoming a national security issue.

The US angle is clear: Corning is building three new factories in North Carolina and Texas. Nvidia plans to produce up to $500 billion worth of AI infrastructure in the US by the end of the decade.

Taiwan remains the manufacturing backbone. TSMC is making Blackwell chips in Arizona, while Foxconn is assembling GB200 supercomputer clusters for hyperscale data centers. The relationship is also deepening beyond manufacturing. TSMC is now using Nvidia’s CUDA-X software and AI models inside its own fabs to improve yield, simulation and process control.

South Korea is the bigger strategic shift. Its chaebol system means one relationship can open multiple markets at once: memory, telecom, cars, robots and internet platforms. Jensen Huang’s June 5–8 Seoul visit reflected that. He met the chairs of LG, SK, Hyundai and Naver, appeared with Doosan Group’s chairman, and left with multiple signed agreements.

This is different from a normal supply chain deal. SK Telecom is not just supplying Nvidia. It is building a gigawatt-scale AI cloud on Nvidia’s full infrastructure stack, with the first factory expected in 2027. Naver is building AI data centers on Nvidia technology. LG is using Nvidia Isaac for its CLOi humanoid robot.

South Korea is moving from being a supplier to Nvidia to becoming a customer of Nvidia’s full AI platform. This is market creation, not just supply security.

What Nvidia’s Partnerships Mean for Investors

When Jensen Huang publicly called Marvell "the next trillion-dollar company" at Computex on June 2, Marvell's stock surged 32% within hours. Nvidia had backed the endorsement with a $2 billion equity investment in March, tied to Marvell's role in building NVLink-compatible networking and connectivity silicon for Nvidia's distributed computing architecture. 

When the world's most valuable company puts capital behind a public endorsement, markets treat it as a directional signal. This raises a question every investor in this space is now asking: do Nvidia's partnership announcements function as forward guidance on which companies will matter most in AI infrastructure?

The answer is yes but the signal quality depends entirely on the type of partnership. Here is the framework.

• Tier 1: Strongest signals. Marvell, Lumentum, Coherent and Corning are being built directly into Nvidia’s next-gen AI stack.
• Tier 2: Ecosystem partners. ServiceNow, Adobe and Microsoft extend Nvidia into enterprise software, creative AI and AI PCs. These are meaningful integrations, but the revenue impact depends on how quickly agentic AI and edge AI adoption scales.
• Tier 3: Future demand plays. SK Telecom, Naver, LG and Hyundai expand Nvidia into Korean AI clouds, robotics and physical AI. The opportunity is real, but many deployments begin from 2027, so execution matters.

For investors, Nvidia remains the core AI infrastructure bet with $81.6 billion quarterly revenue and 85% YoY growth. Tier 1 partners offer the clearest second-order opportunity. But Marvell’s 32% move worked because it already had $6.1 billion in FY2026 data center revenue, 18 cloud-provider design wins and a custom silicon run rate of $1.5 billion. Not every Nvidia partnership has that foundation.