Nvidia and South Korea’s SK Group have announced a $500 billion-plus artificial intelligence initiative spanning large-scale AI factories and next-generation memory technology.
The partnership represents a potentially significant expansion of global AI computing capacity. SK Telecom plans to develop an AI cloud of up to 2 gigawatts using Nvidia’s Vera Rubin platform, while SK hynix will establish a long-term agreement to supply and jointly develop advanced memory products for Nvidia.
Despite the scale of the announcement, Nvidia and several other semiconductor stocks traded lower. Nvidia shares fell approximately 4.4%, Micron declined 4.7% and SanDisk lost 11% as the Philadelphia Semiconductor Index dropped around 4%.
The market reaction highlights a growing shift in investor sentiment. Demand for AI infrastructure remains strong, but investors are becoming more concerned about the enormous capital required to build it and the time needed to generate adequate returns.
What the Nvidia and SK Group Partnership Includes
The proposed initiative covers two closely connected parts of the artificial intelligence supply chain: computing infrastructure and advanced memory.
SK Telecom intends to build a large Nvidia-powered AI cloud in South Korea with capacity reaching as much as 2 gigawatts. The infrastructure will use Nvidia’s DSX architecture and Vera Rubin accelerated-computing systems supported by SK hynix HBM4 memory. The first AI factory is expected to begin operating in 2027.
An AI factory is a large computing facility designed to train models and operate artificial intelligence applications at scale. Unlike a conventional data center, it is optimized around accelerated processors, high-speed networking and specialized memory.
The partnership is intended to serve demand for enterprise AI, sovereign AI, autonomous agents and physical AI applications across South Korea and the wider Asia-Pacific region. Sovereign AI refers to models and computing infrastructure operated under the control and regulatory framework of a particular country.
Nvidia and SK Group have signed letters of intent, meaning some commercial, financing and deployment details may still need to be finalized.
SK Hynix Secures a Central Role
SK hynix will also enter a long-term AI memory partnership with Nvidia.
The companies plan to secure and jointly optimize future generations of high-bandwidth memory, including HBM products designed for large-language-model training, agentic AI and physical AI workloads.
HBM is essential to modern AI accelerators because it allows processors to retrieve large quantities of data rapidly. A powerful GPU can be limited when its memory system cannot supply data quickly enough.
This makes memory capacity an increasingly important part of the AI infrastructure market. As Nvidia introduces more powerful platforms, each system may require greater amounts of advanced HBM.
The agreement could provide Nvidia with improved visibility into future memory supply while giving SK hynix a substantial long-term customer opportunity. It may also help both companies coordinate product development so that future memory generations are optimized for Nvidia’s computing architecture.
For Micron and other memory suppliers, the partnership is more complicated. Continued expansion of AI infrastructure supports industry demand, but a deep Nvidia-SK hynix relationship could strengthen one competitor’s position in premium HBM products.
Why the Stock Fell on Positive News
The decline in Nvidia stock does not necessarily mean investors view the SK partnership as strategically weak.
The selloff occurred during a broader retreat in semiconductor shares. The Philadelphia Semiconductor Index fell approximately 4%, while the overall technology sector also declined as investors prepared for earnings from Microsoft, Amazon, Meta and Apple.
Markets are becoming increasingly sensitive to the financial burden associated with AI infrastructure. Major technology companies are spending unprecedented amounts on chips, data centers, networking equipment and electricity.
The central question is no longer whether companies want more computing capacity. It is whether the revenue and cash flow generated by that capacity will justify the investment.
The $500 billion-plus headline may therefore have intensified concerns about industrywide capital expenditure rather than providing an immediate catalyst for semiconductor stocks.
Nvidia benefits when customers build more AI infrastructure because those projects require its processors, networking systems and software. However, the company’s valuation also depends on customers remaining financially capable and willing to maintain their spending plans.
What the Deal Could Mean for Nvidia Revenue
The partnership could support demand for several parts of Nvidia’s product portfolio.
SK Telecom’s planned AI factories will use the Vera Rubin platform, which follows Nvidia’s Blackwell generation. The deployment may include GPUs, CPUs, networking products, rack-scale systems and Nvidia’s software stack.
Selling complete systems can generate more revenue per installation than selling individual processors. It also allows Nvidia to position itself as an infrastructure-platform provider rather than simply a chip designer.
A 2-gigawatt cloud would represent a substantial computing deployment. However, the total initiative value should not be treated as immediate Nvidia revenue.
The announced figure covers a broad, multiyear collaboration involving infrastructure construction, energy, memory, systems and other investments. Revenue will be recognized gradually as equipment is ordered, delivered and accepted.
Investors should therefore focus on confirmed purchase commitments, deployment schedules and the timing of the first operational capacity rather than assuming that the full headline value will flow directly to Nvidia.
The Capex Debate Is Becoming More Important
Capital expenditure, or capex, refers to spending on long-term assets such as servers, processors, data centers and power infrastructure.
AI projects often require large cash outlays years before they reach full utilization. Even when customer demand is strong, construction delays, power constraints and hardware installation schedules can postpone revenue.
Investors are also concerned about the economic life of expensive AI systems. New processor generations can deliver better performance and energy efficiency, potentially making older infrastructure less competitive before it has generated the expected return.
These concerns have become particularly important after reports of weaker free cash flow at major technology companies and growing scrutiny of debt-funded corporate spending.
The positive case is that AI demand remains constrained by insufficient computing capacity. Under that scenario, new infrastructure could achieve high utilization and strong pricing.
The negative case is that companies are building capacity faster than profitable applications can develop. That could eventually result in lower utilization, weaker pricing and pressure on investment returns.
Which Semiconductor Stocks Could Be Affected?
Nvidia is the most direct potential beneficiary because the planned AI factories will use its accelerated-computing platform.
SK hynix could benefit from long-term HBM demand and deeper product collaboration with Nvidia. The agreement may strengthen its visibility into future memory requirements and support additional manufacturing investment.
Micron remains exposed to the broader expansion of HBM demand, but investors may question whether the SK hynix partnership changes competitive dynamics.
Other AI hardware companies, including AMD, server manufacturers and networking suppliers, could face mixed implications. A larger AI market may expand total demand, but Nvidia’s increasingly integrated platform could make it harder for competitors to win parts of each deployment.
The market reaction shows that investors are currently evaluating the initiative through the wider capex debate rather than treating it as a simple positive for every chip stock.
What Nvidia Investors Should Watch Next
The most important development will be the conversion of the letters of intent into detailed commercial agreements.
Investors should monitor the financing structure, confirmed capacity timeline and the amount of Nvidia hardware expected in each stage of the SK Telecom deployment.
The planned 2027 launch of the first AI factory will be a key execution milestone. Delays involving construction, electricity, networking or chip supply could postpone revenue.
Nvidia investors should also examine whether the company can maintain strong margins as it provides increasingly complete systems. Larger deployments can increase revenue, but they may involve a different product mix from standalone GPU sales.
The SK Group agreement reinforces Nvidia’s dominant role in the AI infrastructure ecosystem. The stock-market decline, however, shows that investors now require more than ambitious investment announcements. They want clearer evidence that the next wave of data-center spending will produce sustainable cash returns.
FAQ
How large is the Nvidia and SK Group partnership?
The companies announced a planned initiative valued at more than $500 billion across AI factories and next-generation memory infrastructure.
What is SK Telecom planning to build?
SK Telecom plans to develop an Nvidia-powered AI cloud with capacity of up to 2 gigawatts. The first AI factory is expected to begin operating in 2027.
What role will SK hynix play?
SK hynix will enter a long-term partnership with Nvidia covering the supply and joint optimization of next-generation AI memory, including HBM.
Why did Nvidia stock fall after the announcement?
Nvidia declined during a broader semiconductor selloff as investors worried about rising AI infrastructure spending and the returns companies may generate from that investment.
Does the $500 billion figure represent Nvidia revenue?
No. The figure covers a broad, multiyear initiative involving data centers, memory and supporting infrastructure. Nvidia revenue would be recognized gradually as products and services are delivered.
Disclaimer: This article is for informational purposes only and does not constitute financial advice. Always conduct your own research before making any investment decisions.
