Key Points
- Curious about Micron Technology, Inc.? Here are five stocks that we prefer more.
- Keith Kaplan from TradeSmith highlights five physical "bottlenecks" in the development of AI—high bandwidth memory, photonics, thermal management, power generation, and grid transmission—where the demand is already surpassing the global capacity to meet it.
- The top four U.S. hyperscalers are expected to invest over $700 billion in AI infrastructure this year, while global data center power consumption is forecasted to more than double by 2026.
- Although there have been notable recent improvements, each of the five highlighted stocks still possesses demand fundamentals that have not yet been fully reflected in their prices, as indicated by multi-year backlogs and established capacity, suggesting a long-term growth potential.
The most significant growth developments in the current market are not taking place among the chip manufacturers. Instead, they are occurring one level below them.
Keith Kaplan, the chief executive officer of TradeSmith, has been recently outlining what he refers to as the"bottleneck areas" of the AI expansion- the physical constraints where a trillion-dollar demand is encountering a supply system that can't keep up quickly enough. His argument: the biggest fortunes in this AI era won't be captured by the obvious players. Instead, they'll go to the companies that those prominent players cannot operate without.
The $700 Billion Problem
By the conclusion of 2025, the four biggest U.S. hyperscalers-Microsoft (NASDAQ: MSFT), Alphabet (NASDAQ: GOOGL), Amazon (NASDAQ: AMZN), and Meta Platforms (NASDAQ: META)-are anticipated to invest over $700 billion in AI infrastructure. This figure is expected to rise closer to $1 trillion by 2027, according to forecasts provided byNVIDIA (NASDAQ: NVDA) during its most recent earnings call, could amount to between $3 and $4 trillion in overall economic effect within three to five years.
That funds are not being directed toward software. Instead, they are being invested in massive structures that span the size of football fields, power systems capable of generating gigawatts, and transmission lines that have not been built yet. Imagine it as demand being forced through a very small opening: regardless of how much water is available upstream, the flow is limited by what the narrowest point can handle.
Kaplan outlines five of these critical points, along with a corresponding stock for each.
Memory: The Microchip Within the Microchip
High-bandwidth memory (HBM) is essential for a GPU to operate effectively.
Without it, artificial intelligence chips cannot function.Micron Technology (NASDAQ: MU)It is the sole American manufacturer of HBM, and its production for 2025 was entirely sold before the year started. The capacity for 2026 is already mostly reserved.
As NVIDIA shifts its Rubin chips from HBM3E to HBM4, Micron's role in the supply chain continues to grow stronger. Annual revenue is expected to reach $58 billion, while net income has already hit $24 billion. Kaplan views this as a three-to-five-year period of stability, with the upcoming earnings report on June 24 serving as an early indicator of future demand trends.
Photonics: The Velocity of Light Between Integrated Circuits
As data exits a single chip, it needs to interact with thousands of other components, and copper wiring is unable to handle the demand. The answer is silicon photonics: using light through fiber to achieve speeds that copper cannot match.
Coherent Corp. (NYSE: COHR)is the top provider of optical transceivers that enable this process, transforming electrical signals into light pulses and vice versa at each connection within an AI cluster. NVIDIA invested $2 billion in Coherent earlier this year—a move that Kaplan says highlights the company's crucial role in the upcoming stage of AI development.
Thermal Control: Preventing It from Melting
High-end AI chips currently consume as much as 1,200 watts each.
Place 72 of them in a rack, and you'll achieve the heat output equivalent to a small apartment within a space similar to a refrigerator. Air cooling is inadequate for this. Direct-to-chip liquid cooling is 3,500 times more effective at removing heat than air at the same flow rate.
Vertiv Holdings (NYSE: VRT)currently provides the majority of the large hyperscaler build-outs with its cooling distribution units. With a market capitalization exceeding $100 billion and annual revenue surpassing $10 billion, it is not a speculative stock.recent decline of over 10%the day before Memorial Day is what Kaplan would consider a favorable opportunity to enter the market ahead of its earnings on July 29.
Power Production: Nuclear Energy Resumes Operations
A single AI data center now uses as much power as a medium-sized city. Meta's Hyperion facility is currently running at 2 gigawatts and is projected to expand to 5. Meeting this continuous energy demand around the clock requires natural gas and nuclear power. Both are experiencing significant growth.
Constellation Energy (NASDAQ: CEG)is the leading nuclear energy provider in the United States, operating 21 reactors and having entered into a 20-year agreement with Microsoft in September 2024 to restart Three Mile Island. Constellation is allocating $1.6 billion to restore the facility and provide 835 megawatts of exclusive power generation.
The change in who is purchasing nuclear power reveals the story: a decade ago, the main buyers were utility companies. Now, they are software firms. Amazon's 17-year, $18 billion power purchase agreement withTalen Energy (NASDAQ: TLN)emphasizes how fiercely major technology companies are striving to secure base-load supply.
The the stock has remained unchanged for the last three months, a consolidation that Kaplan sees as a preparation before the early August earnings, not an indication that the thesis has failed.
The Grid: The Final Stretch That Proves the Most Challenging
Even once the power plant is constructed, delivering electricity to the data center may still require several years. The lead time for transformers has increased from 12 months to as long as two and a half years. Large gas turbines are currently scheduled up to seven years in advance. In certain utility areas of northern Virginia, grid connections for projects submitted after 2024 might not be accessible until 2028 at the earliest.
Eaton Corporation (NYSE: ETN)has been addressing this issue for over a century. Transformers, circuit breakers, and power distribution— the systems that link each data center to the main electrical grid.
Data centers have become Eaton's fastest-expanding sector, with the company experiencing long-term order backlogs, a $148 billion market valuation, and almost $30 billion in yearly revenue. Thestock has remained mostly unchangedOver the last three months, Kaplan interprets this as a period of quiet growth. Next earnings report: August 4.
A 5-Year Major Trend Still in Its Initial Stages
The portion of hyperscaler budgets allocated to memory has increased from 8% to 30% within two years. U.S. data centers may represent 17% of the country's total electricity usage by 2030, compared to 4% currently. These figures are not just predictions being discussed in hushed tones—they are numbers actively influencing how capital is being directed throughout the economy.
The primary question investors are currently pondering is whether they have overlooked something. According to Kaplan, that's far from the case. Although stock prices have risen, the necessary foundation has yet to be established. This discrepancy is precisely where the potential lies.
Where to Allocate $1,000 at This Moment?
Prior to executing your next trade, you should listen to this.
monitors the leading and most successful research analysts on Wall Street and the stocks they suggest to their clients each day.
Our group has recognized thefive stocksThat top analysts are subtly advising their clients to purchase now before the general market becomes aware... and none of the major stocks were included on the list.
They think these five stocks represent the top five companies for investors to purchase at this time...
Recommended Stories
- The Leading Five Stocks to Hold in June
- 5 Stocks Leading the AI Race as Everyone Focuses on NVIDIA
- Berkshire Disposes of Visa, Domino's, and Pool Corp: Should You Mimic the Move?
- Why Walmart's Surge Might Require a Reset
- Cloud-inflation: Dell's In-House AI Strategy Proves Successful
- Alphabet Has Revealed Its Most Bold AI Portfolio So Far
No comments:
Post a Comment