For much of the AI boom, investors have tracked the same four technology giants...
Amazon (AMZN), Alphabet (GOOGL), Meta Platforms (META), and Microsoft (MSFT) are known as the "hyperscalers."
They've poured money into sprawling data centers, which run AI models. And their combined capital-spending plans for 2026 exceed $700 billion.
The hyperscalers created the first wave of AI infrastructure. Now the companies supplying memory for that infrastructure are preparing a huge build-out of their own...
South Korea is coordinating at least 1,350 trillion won (equal to about $880 billion) of planned investment from some of its largest tech companies.
About 800 trillion won are earmarked for semiconductor manufacturing. Another 550 trillion won will support AI data-center capacity.
AI investments are clearly expanding beyond Silicon Valley. And investors need to prepare for a broader tech-infrastructure cycle.
Memory makers are joining the race...
Samsung Electronics (005930.KS) and SK Hynix (000660.KS) are the world's largest memory-chip manufacturers.
They plan to build two chipmaking plants apiece in southwestern South Korea. That's part of the country's broader goal to double its memory-production capacity within five years.
The chipmaking portion of that plan works out to roughly 400 trillion won ($260 billion) each for the two companies.
Samsung is already spending at a dizzying pace... It has announced more than $70 billion for production-capacity and research investment for 2026 alone.
This is a response to the massive memory shortage across the world today...
AI systems require huge amounts of advanced memory to quickly move data between processors. As companies deploy larger models and serve more users, they need more high-bandwidth memory across computing systems.
Supply has struggled to keep pace. JPMorgan Chase (JPM) estimates that the cost of some computer memory chips will soar by as much as 400% from 2024 through the end of this year. Device makers, including Apple (AAPL), have raised prices due to these memory constraints.
SK Group executives expect the shortage to persist for several more years. That gives Samsung and SK Hynix a strong reason to add chip capacity before competitors catch up. It also explains why South Korea views this build-out as a top priority.
Memory chips are essential for both data centers and consumer devices. And South Korea already holds a commanding position in the market. Expanding capacity would strengthen its role across the global AI supply chain.
Yet memory alone can't solve the bottleneck...
The 550 trillion won planned for data-center capacity shows just how far the AI cycle has spread.
Companies like Naver (035420.KS) are backing enough new capacity to consume 8.4 gigawatts of power by 2029. That's enough to keep 7 million homes running.
South Korea's government has pledged support for the electricity and water access needed to fuel the data-center expansion. (But officials haven't provided detailed policies or spending figures for water and electricity.)
That creates a much wider group of potential beneficiaries.
Semiconductor-equipment companies will supply the machinery used inside new factories. Data-center operators will manage the facilities. And utilities and engineering firms will have to deliver the power and cooling to help data centers run 24/7.
Samsung shares fell nearly 5% after South Korea's announcement, while SK Hynix dropped 2%.
Investors recognized that hundreds of billions of dollars in new capacity could eventually push supply ahead of demand.
Hyperscaler budgets were the first clear sign of a historic capital-spending cycle...
South Korea's investment shows that the AI cycle has entered a new stage. Memory producers are expanding capacity to address the current bottleneck.
That's a bullish signal for the broader AI economy...
A bigger memory supply may eventually create pricing pressure for chipmakers. It will also remove a constraint that has limited data-center expansion.
The bottom line is, AI spending is gaining momentum beyond the hyperscalers.
The focus on memory capacity will give the rest of the tech supply chain room to grow... And investors will have the opportunity to profit from the chipmakers.
Regards,
Joel Litman
August 5, 2026
P.S. The world's largest AI computing complex generates billions of dollars in annual revenue from the biggest names in AI.
And it sits inside Elon Musk's SpaceX – soon to be the world's first, and biggest, "AI landlord." Yet, no matter what you think of SpaceX's earnings from last night, the stocks best positioned to profit from it aren't SpaceX itself.
They're the specialized suppliers feeding Musk's $50 billion AI build-out. It's the same picks-and-shovels pattern that turned Vertiv into a 1,300% winner and Super Micro Computer into a 1,250% winner when the first AI boom took off.
Our Director of Research Rob Spivey recently divulged the best opportunities in this space. Watch his special presentation to get ahead of the market and profit from his stellar track record. Click here for all the details, including Rob's free stock pick.
