저자: TomLee · 외국인
요약 보기
AI 인프라 투자 규모(약 30조달러)는 1997년 인터넷·모바일 인프라 투자(약 35조달러)와 유사해 현재를 버블 말기가 아닌 초기 단계로 본다. 연간 약 7조달러의 신규 유동성이 유입되며 대규모 데이터센터 투자가 이를 다 흡수하지 못한다. 이 흐름에 따라 AMD(서버 CPU), Vertiv(액체냉각·전력), NVIDIA(GPU, 저평가 PEG 0.2~0.5)를 유망 인프라 종목으로 제시한다.
다듬은 글
First of all, the market is huge. I mean, it's at least $200 trillion, at the very least, but it's also growing at about $7 trillion a year. Of course? That's what I mean , just think about it for a minute. Every year, another $7 trillion in liquid assets emerge, looking for a place to invest. If you all think it's a bubble, then you're probably not going to blow it up because you all think it's a bubble.
ChatGPT was launched in 2023. So, we have been developing innovations coming from artificial or superintelligence for 3 years now. Total costs for developing the Internet, mobile communications, and gadgets in 2026 prices amounted to about 35 trillion. AI infrastructure spending is around 30 trillion.
So it costs about the same . That's the thing. Cisco's peak came only in 2000. Of course? And between 1993 and 2000, Cisco shares fell by 50% twice. So, you were on the verge of total collapse twice. And I believe that today we are actually in 1997 . I think we are far from a bubble. The dominant narrative on Wall Street claims that we are inside a dangerously overheated AI bubble that will inevitably burst.
However, this panicked consensus ignores the fundamental historical context and dynamics of capital. The average fund manager today has a planning horizon of barely 10 years, making him completely blind to structural technological revolutions. When you examine the real balance sheets of the global economy, the numbers tell a very different story.
Total liquid assets in stock and bond markets now exceed $200 trillion, growing by a staggering $7 trillion in new capital each year. Annual capital spending on hyperscale facilities of $1 trillion does not exhaust the financial system. They absorb less than 15% of new global liquidity seeking yield.
Comparing current infrastructure development with past technology cycles confirms that we are not at the end of the hype, but rather in 1997. During the rollout of the Internet and mobile networks in the late 1990s, global infrastructure spending reached approximately $35 trillion in inflation-adjusted prices before the market fully matured years later.
Current aggregate investment in AI infrastructure is approaching $30 trillion—an almost identical physical structure, but backed by an economy that generates massive free cash flow, not speculative debt. Capital expenditures at the world's largest technology companies are projected to grow by more than 25% annually through 2030, driven by real business adoption rather than inflated valuations.
To benefit from this multi-trillion dollar wave of liquidity, investors must look past the short-term volatility and focus on the irreplaceable bottlenecks fueling this expansion. The largest capital investments over the next decade will go directly into three critical pillars. A primary and secondary computing giant that dominates the high- end server processor market.
Liquid cooling and power distribution infrastructure is critical, without which next-generation data centers cannot operate. And the major semiconductor monopoly that underpins the entire global transformation of artificial intelligence. Here are three high-value infrastructure leaders ready to absorb this huge annual capital inflow today.
Let's start with our first pick: Advanced Micro Devices, stock ticker AMD. While Wall Street has historically ranked AMD as only a secondary player behind the major GPU vendors, the structural mechanics of AI infrastructure have rapidly changed. As agent computing, spatial intelligence, and multi-step reasoning models permeate enterprise technologies, the workload bottleneck is shifting back to high-performance server CPUs.
This structural transition directly aligns with AMD's core operational strengths. The company has officially become the market leader in premium server processors , with current-generation inventory effectively sold out for future cycles as global technology hyperscalers scale their hardware platforms.
From an operational and financial perspective, the sheer speed of this expansion is outpacing Wall Street's outdated baseline models . Thanks to massive infrastructure deployments by leading AI developers and social media platforms, institutional forecasts now predict that AMD server processor sales could exceed $50 billion in just the next few years.
Overall corporate revenue estimates are approaching $125 billion, driven by a year-over-year doubling of the data center segment . Gross margins continue to grow to levels of 55–60%, creating tremendous operating leverage. As net profits increase, long-term projected EPS could realistically jump from single-digit baselines to $30–$40 per share over the next three to four years.
In addition to manufacturing its own hardware, AMD is taking aggressive strategic steps to strengthen its entire software and architectural stack . The firm's recent $8.2 billion acquisition of AI powerhouse World Labs brings world-class spatial intelligence and robotics capabilities to its ecosystem .
This strategic move allows the company to develop a full stack of hardware and software specifically tailored for autonomous systems and physical AI, challenging established software monopolies before the next-generation robotics markets mature . Backed by a strong balance sheet with over $8.8 billion in net cash and $10 billion in annual operating cash flow, AMD has both the financial resources and architectural momentum to capture a significant share of the $ 7 trillion annual capital wave .
Moving on to our second pick: Vertiv Holdings, trading under the ticker symbol VRT. You simply cannot invest tens of trillions of dollars in high- density computing clusters without solving the extreme engineering challenges of heat dissipation and power distribution. While market corrections sometimes hit physical infrastructure providers, creating short-term valuation divergences, Vertiv's operating fundamentals remain exceptionally strong.
The company occupies a dominant position in an industry oligopoly, deriving approximately 85% of its revenues directly from data centers. As compute platforms transition from legacy air cooling to high-efficiency liquid cooling and advanced 800V DC architectures , Vertiv's revenue share per megawatt of power continues to grow rapidly.
From a financial perspective, Vertiv demonstrates impressive operating leverage and cash generation. The annual corporate sales forecast indicates $14 billion, reflecting over 31% organic revenue growth year over year. Adjusted operating income and earnings per share are growing nearly 60% annually.
Even more impressive are the company's cash flow figures: quarterly operating cash flow increased 241% year-over-year to $1.1 billion . This cash growth is driven by aggressive customer prepayments as large data center operators rush to pre-order critical power, cooling distribution, and thermal management equipment.
Innovative modular solutions, such as One Core and Smart Row, contribute to the industrialization of data center construction, eliminating complex on- site setup steps and accelerating the delivery of ready-made factory solutions. Looking ahead, the industry's capital spending on data center construction is projected to reach trillions of dollars by the end of the decade, creating huge multi-year demand.
Co-developing next-generation cooling and DC power architectures with semiconductor industry leaders directly connects Vertiv to future high-density chip implementations. Valuation metrics reflect this compelling growth profile, with forward PEG ratios sitting at an extremely attractive level of around 0.6.
With a growing high-margin service pipeline and unmatched engineering scale , Vertiv stands out as a key player poised to absorb the huge annual capital inflows driving the global expansion of AI. Moving on to our third and final stock, we come to the epicenter of the global AI computing boom— NVIDIA Corporation, ticker NVDA.
As major tech giants direct their share of the annual $ 7 trillion capital wave to massive data center construction, NVIDIA remains a major beneficiary in the hardware space. Despite its dominant market position in AI accelerators and enterprise networking solutions, the stock's valuation structure appears significantly undervalued relative to its growth profile.
Based on forward earnings estimates for future years, the stock is trading at around 14-15 times earnings, a discount to broader market indices, while non-GAAP PEG ratios are at historic lows of 0.2-0.5. The structural demand filling Nvidia's balance sheet continues to break historical precedents. The world's leading technology giants are increasing infrastructure spending by more than 54%, bringing total capital commitments to $1.2 trillion over the next 2 years.
Analysts predict corporate revenues will exceed $692 billion, weathering the cycles, supported by impeccable gross margins exceeding 74% and industry-leading net profit margins. Management's unrivaled internal confidence was recently underscored by the expansion of its share buyback program by a whopping $150 billion, bringing the total capital return to $235 billion by fiscal year 2028.
In addition to producing its own GPUs, Nvidia aggressively surrounds its ecosystem with protective barriers in the form of software and distribution networks. High-profile acquisitions like Hugging Face ensure the company controls the core distribution levels of open- source models, keeping millions of software developers in the integrated environment of its computing architecture.
While supply chain bottlenecks and data center power infrastructure expansion remain pressing factors, the cloud services backlog of over $2 trillion confirms that demand for computing still far exceeds global supply. With an extraordinary discount to growth rates and generating over $134 billion in operating cash flow, Nvidia is the main structural foundation of the entire $200 trillion AI transformation.
Wall Street’s obsession with short-term sentiment has created a rare gap in the valuation of the infrastructure foundation of the digital economy. As $7 trillion in new capital continues to flow into global markets each year, early investment in these three infrastructure giants offers the clearest path to capital growth during this multi-year transition.
Comments 0