The $7 Trillion Race: AI Infrastructure as a Decade-Long Investment Cycle

McKinsey's landmark demand analysis, combined with KKR's structural framework, yields a clear investment conclusion: the AI data center build-out is unlike prior technology bubbles — it is constrained by physics, contracted before it is built, and compounding at a rate that will exceed even optimistic projections. This paper maps the full investment landscape and identifies where the real money is made.

Sources: McKinsey (Apr 2025), KKR (Nov 2025), JLL Research, BLS, Bloomberg
Key Insight

McKinsey projects $6.7 trillion in global data center capex by 2030, with 70% driven by AI workloads. Unlike the 1990s fiber overbuild, this cycle is contracts-first, power-constrained, and vacancy-tight at 2.3%. NVDA, VRT, EQIX, and CEG represent the four high-conviction positions across the Energizer, Technology Developer, and Operator archetypes.

Anton Ladnyi
Anton Ladnyi
Founder & Portfolio Architect — A.L. Capital Advisory
Ex-Goldman Sachs Equity Research · Ex-J.P. Morgan Wealth Management · CFA Level I & II Verified · CFA Level III Candidate

By 2030, data centers will require $6.7 trillion worldwide. That number — from McKinsey's most rigorous technology infrastructure analysis to date — represents roughly the combined GDP of Japan and Germany. The AI data center cycle is regularly compared to the 1990s fiber overbuild. It is a seductive analogy and a fundamentally misleading one. Understanding why is the difference between capturing a decade-long compounding trade and being burned by a narrative that looked good on paper.


01

Demand Landscape

Scale, Structure & the Demand Curve

McKinsey's research shows global demand for data center capacity could almost triple by 2030, with approximately 70% of that demand driven by AI workloads. Total projected capital expenditure: $6.7 trillion, of which $5.2 trillion is attributable to AI processing loads and $1.5 trillion to traditional IT applications.

The hyperscalers are leading the investment wave. Amazon, Google, Microsoft, and Meta are expected to spend over $350 billion on capex in 2025 alone — a year-over-year increase in the mid-30% range. In aggregate, AI-related infrastructure spend in 2025 is estimated at approximately $500 billion, and in H1 2025 it contributed more to US GDP growth than consumer spending. As a share of GDP, AI-related capex now sits at approximately 5% — a level comparable to the late-1990s technology boom.

Exhibit 1
Global Data Center Capacity Demand: AI vs. Non-AI Workloads, 2025–2030 (GW)
Base-case projection: 125 incremental GW added between 2025–2030 for AI workloads alone. Total demand nearly triples from ~82 GW (2025) to ~207 GW (2030). Source: McKinsey & Company, "The Cost of Compute," April 2025.
Year Total Capacity (GW) AI Workloads (70%) Non-AI Workloads (30%) YoY Growth
202582 GW~57 GW~25 GWBaseline
2026105 GW~74 GW~32 GW+28%
2027137 GW~96 GW~41 GW+30%
2028163 GW~114 GW~49 GW+19%
2029191 GW~134 GW~57 GW+17%
2030207 GW~145 GW~62 GW+8%

McKinsey constructed three scenarios ranging from constrained to accelerated demand, shaped by semiconductor supply constraints, enterprise AI adoption rates, efficiency improvements, and regulatory challenges. The base case — $5.2 trillion in AI data center capex — assumes continued growth without runaway acceleration or structural constraints.

Exhibit 2
Three AI Infrastructure Investment Scenarios, 2025–2030
Source: McKinsey & Company, proprietary data center demand model. April 2025.
Scenario Drivers Incremental GW AI Capex Total (AI + Non-AI)
Accelerated Transformative AI adoption; enterprise integration across all sectors; no supply constraints 205 GW $7.9T $9.4T est.
Base Case ★ BASE Continued growth; moderate enterprise adoption; some efficiency gains offset demand 125 GW $5.2T $6.7T
Constrained Supply chain bottlenecks; slower enterprise deployment; AI efficiency gains suppress demand 78 GW $3.7T $5.2T est.
★ Base case used throughout this paper. Range: $3.7T–$7.9T in AI-specific capex depending on adoption trajectory.
02

Structural Analysis

Why This Is Not the 1990s Fiber Overbuild

The analogy to the late-1990s telecommunications infrastructure bubble is compelling in one dimension — the scale of capital deployment — and misleading in every other. Fiber in the 1990s was built speculatively, with virtually unlimited capacity once laid and zero refresh requirement. Data centers are physically constrained, contractually committed before construction, and subject to accelerated depreciation cycles that naturally absorb any temporary excess. The evidence is visible in vacancy data: North American colocation vacancy has fallen from 9.8% in 2020 to 2.3% in H1 2025 (JLL Research), while the fiber glut post-2001 saw vacancy exceed 20%.

The key structural difference McKinsey identifies is the cost of carrying excess capacity. Fiber, once laid, is nearly free to maintain. Data centers are the opposite: power, cooling, and maintenance are ongoing high costs regardless of utilization. But crucially, AI accelerators have 3–4 year refresh cycles — meaning any overcapacity is rapidly converted into obsolescence, and new workloads pull spare capacity well before it becomes stranded.

1800s
Railroads
Speculative overbuilding across UK and US. Bankruptcies, fraud, market crashes. Networks connected ports & cities — the backbone of industrial commerce for 100 years.
1920s
Electrification
228% kWh capacity growth 1920–30. Overleverage met the Depression's demand shock. Interconnected regional grids; factory redesign around electric motors unlocked decades of productivity.
Late 1990s
Fiber 1.0
Comms capex $62B (1996) to $135B (2000). NASDAQ –78%. Telecom bankruptcies. $500B fiber overbuild became the backbone of the modern internet. Capacity endured.
2020s — Now
AI Infrastructure
$6.7T projected capex. Contracts before construction. Power as ultimate constraint. 2.3% vacancy. KKR thesis: "AI isn't a bubble. It's the backbone of the next industrial revolution."
03

Investment Architecture

Five Archetypes: Where the $5.2 Trillion Flows

McKinsey's analysis maps the $5.2 trillion AI capex envelope across five distinct investor archetypes. Understanding this architecture is essential: the investment case, risk profile, and return dynamics differ fundamentally across archetypes.

Archetype 01 · 15% of AI Capex
Builders
15% $0.8T
Real estate developers, design firms, and construction companies that expand and upgrade data center facilities. Key investments: land acquisition, materials, skilled labour, site development.
Examples: Turner Construction · AECOM · Bechtel
Archetype 02 · 25% of AI Capex
Energizers
25% $1.3T
Utilities, energy providers, cooling & electrical equipment manufacturers. Key investments: power generation (nuclear, gas, renewables), direct-to-chip liquid cooling, transformers, network connectivity.
Examples: Duke Energy · Vertiv · Schneider Electric · Constellation Energy
Archetype 03 · 60% of AI Capex
Technology Developers & Designers
60% $3.1T
Semiconductor companies and computing hardware suppliers. The largest single share — because every watt of AI compute ultimately flows through a chip.
Examples: NVIDIA · AMD · Intel · TSMC · Samsung · SK Hynix
Archetype 04 · Unquantified
Operators
Not modelled
Hyperscalers, colocation providers, GPU-as-a-service platforms. Own and run large-scale facilities. Capex overlaps with broader cloud & infrastructure spending — not isolated in McKinsey's model.
Examples: AWS · Google Cloud · Microsoft Azure · Equinix · Digital Realty
04

Signal vs. Noise

What the Bears Get Right — and Wrong
Structural Bull Case
  • Vacancy at 2.3% in N. America H1 2025 — no speculative overbuild visible (JLL)
  • Contracts-first builds: hyperscalers require offtake agreements before construction begins
  • Power is the ultimate physical constraint on overbuild — grid queues, transformer lead times, permits
  • 3–4 year accelerator refresh cycles naturally absorb any temporary excess capacity
  • AI is a horizontal productivity layer across all industries, not a niche connectivity play
  • Lower unit costs drive accelerated adoption (Jevons Paradox — efficiency creates more demand)
  • Both inference and training workloads growing; inference to dominate by 2030
Risks & Bear Case
  • AI use-case failure: enterprises building but not deploying at scale — ROI visibility remains limited
  • Efficiency disruption: DeepSeek V3's 18× training cost reduction could suppress GPU demand
  • Concentration risk: NVIDIA at ~8% of S&P 500 — single-stock exposure in any AI basket
  • Geopolitical: US–China semiconductor export controls create supply chain and demand uncertainty
  • Rising power costs squeeze operators without long-term power contracts
  • Some business models (GPU rental, thin-margin operators, non-core markets) will not survive

"The stakes are high. Overinvesting in data center infrastructure risks stranding assets, while underinvesting means falling behind. The winners of the AI-driven computing era will be the companies that anticipate compute power demand and invest accordingly."

— McKinsey & Company, "The Cost of Compute," April 2025
05

Investor Framework

Winners, Losers & the Asset Playbook

The $5.2–$6.7 trillion capex envelope flows through a defined set of public equities. But raw exposure to the AI theme is not sufficient — the archetype, moat, and balance sheet quality of each company determine whether they capture compounding returns or get crushed in the shake-out.

NVDA
High Conviction
NVIDIA Corporation
The dominant AI accelerator: $3.1T of the AI capex envelope flows through Technology Developers, and NVIDIA captures the largest single share. The CUDA ecosystem creates a software lock-in that AMD and Intel have spent years trying to break without success. H100/H200/B200 backlog extends well into 2026. Risk: export controls on H20 chips to China, and NVIDIA's weight at ~8% of the S&P 500 creates index-level concentration. The moat is real; the valuation demands discipline on position sizing.
Archetype Tech Developer
Capex Pool $3.1T
Key Moat CUDA Ecosystem
VRT
High Conviction
Vertiv Holdings
Critical power and thermal management infrastructure for data centers. AI chips run at 10–15× the power density of CPUs, making liquid cooling a necessity rather than a luxury. Vertiv is the global leader in direct-to-chip and immersion cooling systems — technologies McKinsey identifies as essential for the $1.3T Energizer archetype. Long-term hyperscaler contracts provide revenue visibility. This is the "overlooked play" in AI infrastructure: less glamorous than NVIDIA, structurally more defensible.
Archetype Energizer
Capex Pool $1.3T
Key Moat Thermal IP
EQIX
High Conviction
Equinix (REIT)
The gold-standard Operator: 260+ data centers across 70 metros, with interconnect moats that hyperscalers cannot replicate. KKR specifically identifies "entitled land and expansion permits in super-core markets" and "operational hyperscaler relationships" as the hardest competitive barriers to build. Equinix controls both. REIT structure provides dividend yield alongside secular growth. London, Singapore, and Northern Virginia assets command premium EV/MW multiples that will only widen as vacancy tightens further.
Archetype Operator
Key Moat Interconnect + Land
Markets 70 metros
CEG
High Conviction
Constellation Energy
Nuclear baseload as the clean power solution to AI's energy problem. McKinsey identifies nuclear as a key solution for Energizers facing "clean-energy transition requirements." Hyperscalers need carbon-free, uninterruptible power — a specification only nuclear can meet at scale. Microsoft's Three Mile Island PPA agreement is the template. Constellation holds ~5% of US electricity generation capacity in nuclear. With data center power demand growing ~20% pa, 20-year PPAs at premium rates represent a structural earnings uplift that current consensus does not fully price.
Archetype Energizer
Capex Pool $1.3T
Contract Type 20-yr PPAs
MSFT / GOOGL
Selective
Microsoft / Alphabet
Both are simultaneously the largest customers and investors in AI infrastructure. Bull case: they own the cloud margin moat and customer relationships that determine where AI revenue accrues. Bear case: competitive dynamics force defensive capex without ROI discipline. Watch capex/revenue ratios in 2026 earnings closely — this is the key leading indicator.
Archetype Operator
Combined Capex '25 ~$200B
Watch ROI Discipline
AMD
Selective
Advanced Micro Devices
The credible challenger to NVIDIA's GPU monopoly. MI300X competitive benchmarks are genuine, and the ROCm software ecosystem is maturing. The investment case is asymmetric: NVIDIA share loss of even 5–10 percentage points would be transformative for AMD.
Archetype Tech Developer
Thesis Challenger Moat
Risk CUDA Stickiness
GPU Rental
Avoid
GPU Rental Platforms / Thin-Margin Operators
KKR explicitly warns against assets with "single-tenant concentration, short-term leases, thin power margins, and secondary market exposure." GPU rental platforms that arbitrage compute at thin spreads have no structural moat: when hyperscalers build their own capacity (as they are actively doing), demand for rented GPUs collapses.
Risk No Moat
Pattern 1990s ISP
View Avoid
06

Projections & Outlook

What to Expect: A 5-Year Asset Impact Roadmap
Exhibit 3
AI Infrastructure Cycle: Asset Impact Projections by Phase
A.L. Capital Advisory analysis. Arrows: ↑ Positive, ► Neutral/Transitioning, ↓ Negative.
Asset / Sector Phase 1: Build (2024–26) Phase 2: Deploy (2026–28) Phase 3: Compound (2028–30) A.L.C. View
AI Semiconductors
NVDA, AMD
↑ Accelerating. Backlog extends 12–18 months. Pricing power at peak. ► Elevated but normalising. Efficiency gains may compress unit economics. ↑ Next-gen inference demand drives new cycle. Moat compounds. High Conviction Long
Power & Cooling
VRT, CEG
↑ Rapid growth as rack density escalates. Power PPAs being locked in now. ↑ Continued deployment of liquid cooling. Nuclear PPAs extending. ↑ Structural beneficiary of all three phases. Most durable earnings quality. High Conviction Long
Data Center REITs
EQIX, DLR
↑ Vacancy tightening. Premium pricing in core markets. Land value accruing. ↑ Expansion of AI-optimised facilities. Interconnect moats widen. ↑ Long-term lease revenue compounds. REIT dividend yield supported. High Conviction Long
Hyperscalers
MSFT, GOOGL, AMZN
↓ Capex absorbs free cash flow. Market questions ROI discipline. ► Cloud revenue inflection as AI workloads monetise. Watch margins. ↑ AI-driven cloud revenue compounds. CapEx declining as % of revenue. Selective. Monitor capex
Construction / Builders ↑ Labour and materials in high demand. Early-cycle beneficiary. ► Growth but margins compress as capacity builds. ↓ Cycle matures. Commodity dynamics. No moat. Tactical only. Not core.
GPU Rental / Thin-Margin Ops ► Works during scarcity. Business model intact for now. ↓ Hyperscalers self-build eliminates demand for rented compute. ↓ Model collapses. Structural shake-out. Avoid. Avoid

Portfolio Construction Framework — Five principles for building the AI infrastructure position without getting burned:

1
Own the moats, not the narrative.
KKR's core tenet. Power access, entitled land, interconnects, CUDA lock-in, and hyperscaler relationships are durable. GPU rental and thin-margin operators are not. The shake-out will concentrate in business models that work only during scarcity.
2
The overlooked play is power.
Of the $5.2T AI capex, $1.3T flows to Energizers — the segment most under-owned relative to its capex share. CEG, VRT, and utility-scale operators in data center proximity markets represent this allocation. Less crowded than semiconductors, more durable in the long run.
3
Size for the volatility, not just the conviction.
Even highest-conviction names will experience 30–40% drawdowns as the cycle matures. Position sizing should reflect that the structural thesis is sound but the path is non-linear. NVIDIA at 8% of S&P 500 demands position sizing discipline.
4
Phase your exposure.
Build vs. Deploy vs. Compound phases favour different archetypes. Semiconductors and power dominate Phase 1. Software and cloud infrastructure dominate Phase 2. Productivity beneficiaries compound in Phase 3. A static allocation to "AI" misses this rotation.
5
Watch the McKinsey indicators.
The three key signal variables: (1) North American vacancy rate — below 3% is healthy; above 6% is a warning; (2) Hyperscaler capex-to-revenue ratios — rising is a bear signal for operators; (3) Enterprise AI deployment rate — the key leading indicator for whether the demand curve achieves base case or slips to constrained.

References

  1. 1. McKinsey & Company. "The cost of compute: A $7 trillion race to scale data centers." Jesse Noffsinger, Mark Patel, Pankaj Sachdeva. TMT Practice, April 2025.
  2. 2. JLL Research. North America Colocation Vacancy, H1 2025. Published June 2025.
  3. 3. KKR Global Infrastructure. "Beyond the Bubble: Why We Think AI Infrastructure Will Compound Long after the Hype." November 2025.
  4. 4. U.S. Bureau of Labor Statistics. GDP and capex share data. Bloomberg terminal data as of June 30, 2025 (cited via KKR GMAA).
  5. 5. DeepSeek V3 efficiency claims: TechCrunch January 27, 2025; Artificial Analysis January 27, 2025.
  6. 6. All stock-specific analysis and projections represent independent views of A.L. Capital Advisory. Not investment advice.
Investment Disclaimer
This report is published by A.L. Capital Advisory for informational and educational purposes only. It does not constitute investment advice, a solicitation to buy or sell any security, or a recommendation to take any specific investment action. All analysis, projections, and opinions expressed are those of the author and are subject to change without notice. Past performance is not indicative of future results. Investing involves risk, including the possible loss of principal. Readers should conduct their own due diligence and consult with a qualified financial advisor before making any investment decisions. References to specific securities (NVDA, VRT, EQIX, CEG, MSFT, GOOGL, AMD) are for illustrative purposes and do not constitute a recommendation to buy or sell those securities.
Apply This Analysis

Translate research into portfolio decisions

The Strategic Session is where we take research like this and build concrete allocation decisions — position sizing, archetype exposure, phase timing — tailored to your risk profile.

Book a Strategic Session →
More Research
Asset Class Deep-Dive · Alternatives
Private Equity in 2026: Navigating the New Vintage
How rising rates, compressed multiples, and a backlog of exits are reshaping private equity returns — and where the structural opportunity lies for patient capital.
Coming Soon · Macro
The Dollar Regime: Reserve Currency Risk in a Multipolar World
A structural analysis of US dollar dominance, de-dollarisation trends, and the portfolio implications of a gradual reserve currency transition.