The Ironbone of AI
Investing in the $100 Billion Infrastructure Behind Artificial Intelligence
$61B
Data Center Transaction Volume (2025)
Why concrete, copper, and cooling systems are the new competitive moats
Comprehensive Analysis | 2025-2026 Investment Landscape
The Fundamental Inversion
Value creation has shifted from virtual to physical
Old Paradigm (2000-2020)
Software & Code: Scale with near-zero marginal cost
Cloud Platforms: Virtual infrastructure dominates
Network Effects: Winner-takes-all dynamics
Unit of Value: Users, engagement, data
New Reality (2025+)
Concrete & Steel: Physical infrastructure is the bottleneck
Amperes & Megawatts: Power is the primary constraint
Industrial Scale: Multi-billion dollar "AI factories"
Unit of Value: Energized, cooled capacity
92%
Of U.S. GDP Growth in H1 2025 from AI Equipment & Infrastructure
"Value has shifted from bits to atoms"
The Scale of Transformation
Data center expansion is reshaping the U.S. economy
Transaction Volume
$61B
Data center deals by late 2025, a historic peak for the sector
Construction Pipeline
8 GW
Under construction or planned, enough to power 6 million homes
Pre-Lease Rate
73%
Of pipeline sold before ground is broken—unprecedented visibility
Tech Cycle Comparison
Dot-com Era (1999-2001): Fiber optic deployment
Mobile Era (2007-2015): Cell towers and handsets
AI Era (2023+): Megawatts of energized capacity
"Zero vacancy risk, maximum execution risk"
The "Ironbone" Investment Framework
Three critical systems powering the AI revolution
Power System
The circulatory system: Delivering reliable electricity at industrial scale
Grid infrastructure & transmission
Backup generators & transformers
On-site power generation
Cooling System
The respiratory system: Managing intense thermal loads from AI chips
Liquid cooling infrastructure
HVAC & thermal management
Modular cooling units
Structural System
The skeletal system: Physical construction and site preparation
Civil engineering & site work
MEP (Mechanical/Electrical/Plumbing)
Materials & equipment rental
Investment Thesis: The AI revolution requires unprecedented physical infrastructure. Companies supplying these three systems have multi-year backlogs, structural demand drivers, and physical moats that protect against competition.
Ground Zero: The Four AI Infrastructure Hubs
Where the buildout is concentrated and why geography matters
Northern Virginia
"The Incumbent" — Processes ~70% of global internet traffic
Loudoun County saturation driving southern expansion
$50.1B Dominion Energy infrastructure plan
Intense NIMBY pressure and zoning battles
Dallas-Fort Worth
"The Scaling Challenger" — Abundant land, lower costs
$25B Vantage "Frontier" campus (1.4 GW)
Grid-independent facilities emerging (5 GW planned)
Proactive zoning reforms for data centers
Phoenix, Arizona
"The Desert Hub" — Low disaster risk, water constraints
$25B Hassayampa Ranch on 2,000 acres
2,000 MW natural gas plant with "subscription model"
Cooling water consumption controversy
Columbus, Ohio
"Silicon Heartland" — Intel fab-driven expansion
Google & Amazon following Intel investment
$36/month utility rate shock for residents by 2027
$2B+ Vantage campus expansion
Each hub faces unique constraints that favor specialized infrastructure providers
Northern Virginia: The Fortress Under Pressure
How "Data Center Alley" is managing saturation
70%
Of World Internet Traffic Flows Through This Region
The Expansion South
Loudoun County effectively full and power-constrained
Development bleeding into Caroline, Spotsylvania, Prince William counties
CleanArc's $3B, 900 MW Caroline County campus (Project VA1)
50 permanent jobs, hundreds of construction roles per project
Power & Regulatory Challenges
Dominion Energy's $50.1B infrastructure plan (2025-2029)
Rebuilding 500kV loop serving the region
Contentious zoning battles in "Data Center Overlay Districts"
Noise complaints from residents near cooling towers
KEY BENEFICIARIES
Dominion Energy (utility infrastructure) • Stack Infrastructure (developer) • Parsons Corporation (complex permitting/engineering) • HITT Contracting & Southland Industries (local construction specialists)
Scale vs. Scarcity: Texas & Arizona
Two distinct approaches to infrastructure constraints
Dallas-Fort Worth: Scale Play
$25B
Vantage "Frontier" Campus — 1,200 acres, 1.4 GW
Abundant land enables gigawatt-class campuses
Equinix's $836M Mockingbird facility with 20 generators
Grid independence trend: GridFree AI's 5 GW "South Dallas One" cluster
On-site gas generation bypassing utility queues
Beneficiaries: Comfort Systems USA, Oncor, CyrusOne, Stream Data Centers
Phoenix: Desert Innovation
$25B
Hassayampa Ranch — 2,000 acres approved by Maricopa County
Low natural disaster risk, but severe water constraints
Cooling consumes millions of gallons per facility
APS proposing 2,000 MW natural gas plant (Desert Sun)
"Subscription model" protects residential ratepayers
Beneficiaries: Arizona Public Service, Sterling Infrastructure, Vulcan Materials
Common theme: Localized solutions for systemic constraints create specialist opportunities
The Community Friction Factor
Why regulatory expertise is becoming a competitive advantage
NIMBY Pressure
Noise from cooling towers and generators
Visual impact of industrial facilities
Property value concerns from residents
Zoning overlay districts restricting development
Utility Rate Shock
Columbus: $36/month hikes by 2027 for residents
Political pressure on utility commissions
Demands for specific high-load customer tariffs
Data centers paying for dedicated infrastructure
Environmental Battles
Water consumption (Phoenix cooling crisis)
Discharge quality and water table concerns
Emissions from backup generators
Ohio EPA permit streamlining facing lawsuits
INVESTMENT IMPLICATION
Firms with deep expertise in complex permitting, environmental mitigation, and community relations have a structural advantage. This regulatory friction creates barriers to entry that protect established players like Parsons Corporation, AECOM, and large national contractors with government relationships.
Regional Investment Map
Geography dictates specialization—match investments to hub dynamics
Region Primary Need Key Constraints Beneficiary Companies
Northern Virginia Grid expansion, density management Power capacity, NIMBY zoning Dominion Energy, Parsons Corp, Stack Infrastructure
Dallas-Fort Worth Rapid scaling, on-site generation Construction velocity Comfort Systems USA, Oncor, Stream Data Centers
Phoenix Water management, backup power Cooling water, heat APS (Pinnacle West), Sterling Infrastructure, Vulcan Materials
Columbus Intel ecosystem support Utility rate pressure AEP Ohio, Vantage, Turner Construction
Atlanta (Emerging) 10,000 MW new capacity Generation build-out Georgia Power (Southern Company), QTS Realty
Portfolio Strategy: Balance high-growth pure-plays (Vertiv, Comfort Systems, Sterling) with diversified infrastructure exposure (MasTec, Quanta) and steady regulated utility plays (Dominion, Southern). Consider adding materials/rental for defensive exposure to secular buildout trend.
The Construction Value Chain
Where complexity drives margin expansion
PHASE 1
Civil / Site Work
~15% of cost
→
PHASE 2
Structural Build
~20% of cost
→
PHASE 3 ⭐
MEP Systems
~50%+ of cost
→
PHASE 4
Commissioning
~15% of cost
Traditional Data Centers
Air-cooled systems
10-20 kW per rack density
Standard electrical distribution
24-36 month construction timeline
Lower complexity = commodity pricing
AI Hyperscale Facilities
Liquid cooling loops required
100+ kW per rack (Nvidia Blackwell)
Complex power distribution & redundancy
12-18 month accelerated timelines
High complexity = premium pricing & margins
MEP specialists capturing 50%+ of project value with expanding margins
Case Study: EMCOR vs. Comfort Systems
Two paths to MEP dominance in the AI buildout
EMCOR Group
EME
$12.6B
Remaining Performance Obligations (Backlog)
+52.1% Electrical segment revenue growth YoY
9.4% Operating margins (expanding)
+29% RPO growth year-over-year
STRATEGY
On-site expertise and prefabricated electrical skids. Focuses on complex, mission-critical installations requiring deep integration with hyperscaler requirements.
Comfort Systems
FIX
$9.4B
Record Backlog (Q3 2025)
+65.1% Backlog growth (stunning acceleration)
42% of revenue from technology sector
$553M Operating cash flow in Q3 alone
STRATEGY
Modular manufacturing dominance. Expanding to 3M sq ft of factory space by 2026, building cooling & electrical rooms off-site for rapid deployment.
Different strategies, same outcome: Capturing premium margins from AI density requirements
The Foundation Builders
Civil contractors and grid infrastructure specialists
Sterling Infrastructure
STRL
+58%
E-Infrastructure revenue growth in Q3 2025
25% operating margins
60%+ of backlog is data centers
Dominates mega-site civil work
MasTec
MTZ
$16.8B
Record backlog across all segments
+22% revenue growth to $4.0B
Captures "behind-the-meter" trend
Solar + battery storage for data centers
Quanta Services
PWR
$39.2B
Backlog—largest in the sector
$7.6B quarterly revenue
Building dedicated substations
"Total Solutions" power generation
THE "DIRTY WORK" PREMIUM
Mega-sites like Hassayampa Ranch (2,000 acres) or Texas campuses require unprecedented earthmoving, concrete, and drainage. Sterling has cornered this market, cherry-picking high-margin projects. Meanwhile, MasTec and Quanta solve the "last mile" problem of getting power to the site—often more complex than the building itself.
The Labor Moat
Why skilled trades shortages protect large contractors
You Can't Scale
Expertise Overnight
The Barrier to Entry in Hyperscale Construction
The Shortage Crisis
Severe shortage of licensed electricians across all markets
Pipefitter/HVAC technician deficit limiting cooling installations
Competition with residential and industrial sectors
Union labor requirements in key markets (Virginia, Ohio)
Result: Wage inflation of 15-25% in data center trades since 2023, driving up project costs but protecting margins for firms that can secure talent.
Who Benefits
National footprint firms can mobilize workers across regions
Union relationships provide access to skilled labor pools
Training programs create proprietary talent pipelines
Modular approaches reduce on-site labor requirements
Beneficiaries: EMCOR, Comfort Systems, MasTec squeeze out smaller regional competitors who lack the scale to recruit and retain skilled workers.
Labor scarcity = structural competitive advantage for large, established contractors
Construction Sector Scorecard
Comparing growth, margins, and AI exposure across builders
Company (Ticker) Backlog Growth Operating Margin AI/Data Center Exposure Key Strength
EMCOR (EME) +29% YoY 9.4% (expanding) High Complex electrical systems
Comfort Systems (FIX) +65.1% YoY Industry-leading Very High (42%) Modular manufacturing
MasTec (MTZ) +21% YoY Stable, improving Medium-High Integrated power + fiber
Sterling (STRL) +58% YoY 25% (exceptional) Very High (60%+) Mega-site civil work
Quanta (PWR) Steady growth Consistent Medium (diversified) Grid infrastructure scale
Dycom (DY) +14.1% YoY Moderate Medium (growing) Power Solutions acquisition
HIGHEST CONVICTION
Comfort Systems (FIX) and Sterling (STRL) offer the purest play on AI data center construction with explosive backlog growth and premium margins.
DIVERSIFIED EXPOSURE
MasTec (MTZ) and Quanta (PWR) provide data center exposure plus infrastructure and clean energy diversification for risk management.
The Megawatt Economy
Why power is the primary unit of value in AI infrastructure
2010s Cloud Data Centers
50 MW
Typical facility power requirement
Served traditional workloads (web, storage, compute)
Air cooling sufficient for thermal management
Standard utility grid connections
12-24 month grid interconnection timelines
2025 "AI Factories"
1,000+ MW
Gigawatt-scale campuses emerging
GPU training clusters with extreme power density
Liquid cooling infrastructure mandatory
Dedicated substations & transmission lines required
36-48+ month grid interconnection timelines
4.5 GW
OpenAI/Oracle "Stargate" (Abilene, TX)
= Power for 3.4 million homes
1.4 GW
Vantage "Frontier" Campus (Shackelford County, TX)
$25 billion investment
10 GW
Georgia Power expansion request (2025)
80% for data centers
The megawatt is the new unit of competitive advantage
The Transformer Bottleneck
Supply chain scarcity creating pricing power
120-210 Weeks
Lead Time for High-Voltage Transformers (Up to 4 Years)
The Problem
Global manufacturing capacity insufficient for demand surge
30% supply deficit projected through 2025-2026
Custom transformers for data centers require specialized production
Can't substitute or workaround—transformers are non-negotiable
Projects stalled waiting for delivery despite having funding and permits
The Opportunity
Pricing power: Suppliers can command premium pricing
Visibility: Orders locked in years in advance
Margin expansion: Fixed cost base + higher prices = profitability surge
Barrier to entry: Building transformer capacity takes 3-5+ years
PRIMARY BENEFICIARY: EATON CORPORATION (ETN)
30.3%
Electrical Americas segment operating margins (Q3 2025)
+18%
Electrical sector backlog growth
17%
Revenue from data center segment
Eaton supplies mission-critical switchgear, transformers, and power distribution units. The supply shortage has given them unprecedented pricing power and multi-year backlog visibility.
Powering Reliability
Three solutions to the grid capacity crisis
1. Backup Generators
100%
Uptime requirement driving generator demand
Generac Holdings (GNRC)
C&I sales up 9% in Q3 2025
Doubling backlog to $300M+
$12B annual generator market
2. On-Site Generation
+33%
Caterpillar power gen sales growth
Caterpillar (CAT)
$725M capacity expansion (Lafayette, IN)
Reciprocating engines fastest-growing segment
Production slots booked years ahead
3. Utility Partnerships
$50.1B
Dominion Energy infrastructure plan
Multiple Utilities
Georgia Power: 10,000 MW expansion
APS: 2,000 MW "subscription" model
Dedicated infrastructure for hyperscalers
THE "GRID INDEPENDENCE" TREND
With 36-48 month utility interconnection timelines, data centers are increasingly building "islanded" facilities with on-site power generation. Example: GridFree AI's 5 GW South Dallas cluster powered entirely by natural gas generators, bypassing the grid entirely. This trend massively benefits generator manufacturers and on-site power specialists.
Power Sector Investment Framework
Matching time horizons to opportunity
Short-Term (2-3 Years)
Solving Today's Problems
Generators & Backup Power — Immediate deployment to bypass grid queues
Transformers & Switchgear — Supply-constrained equipment with pricing power
Grid Contractors — Building the connections today
TOP PICKS
Eaton (ETN), Generac (GNRC), Caterpillar (CAT), Quanta (PWR)
Long-Term (3-5+ Years)
Building Tomorrow's Infrastructure
Grid Modernization — Transmission capacity expansion for sustained growth
Renewable Integration — Solar, wind, battery storage for sustainability goals
Nuclear Renaissance — Amazon's nuclear-adjacent PA campuses
TOP PICKS
Dominion (D), Southern Company (SO), MasTec (MTZ) Clean Energy segment
Key Question: "Is the company solving today's problem or tomorrow's?"
Portfolio should balance immediate needs (generators, transformers) with structural buildout (grid, renewables)
The Thermal Tipping Point
Why AI chips are forcing a cooling revolution
Traditional Air Cooling
10-20 kW
Per rack thermal design power
CRAC/CRAH units sufficient
Hot aisle / cold aisle containment
Standard raised floor infrastructure
Lower capital expenditure
Liquid Cooling (AI Racks)
100+ kW
Nvidia Blackwell GPU density
Direct-to-chip liquid cooling loops
Coolant distribution units (CDUs)
Rear-door heat exchangers
Premium pricing & specialized installation
VERTIV HOLDINGS (VRT): THE PURE-PLAY WINNER
+60%
Organic order growth YoY (Q3 2025)
+43%
Organic Americas sales growth
27%
Full-year organic sales growth guidance (2025)
Vertiv co-engineers cooling solutions with Nvidia for Blackwell deployments. As data centers retrofit for AI, they must rip out air conditioning and install Vertiv's high-density thermal systems. This is a multi-year replacement cycle with structural demand drivers.
SECONDARY PLAY: MODINE MANUFACTURING (MOD)
Transformed from auto parts to data center cooling via Airedale brand. Provides chillers and air handling units for liquid cooling loops. Value entry point into the cooling theme.
The Fiber Nervous System
How AI clusters are rewiring connectivity demand
Beads on a String
Fiber Demand Grows Non-Linearly with GPU Cluster Size
The Connectivity Challenge
AI training requires synchronization across thousands of GPUs
Data must move between geographically dispersed availability zones
Training weight updates demand massive bandwidth
Latency sensitivity requires high-capacity fiber interconnects
Back-end networks between data centers as critical as internal networks
Corning Inc. (GLW)
+58%
Enterprise/AI product sales growth YoY
New MMC connector standard packing more fibers into smaller spaces
Dense faceplates for AI server racks
Management signals "significantly increased fiber demand" into 2026
S&P predicts 39% optical segment growth in 2025
DYCOM INDUSTRIES (DY)
Traditional fiber installer acquiring electrical capabilities. $1.95B Power Solutions acquisition brings $1B in data center electrical revenue. Now a hybrid data center services firm capturing both fiber and power infrastructure.
BELDEN INC. (BDC)
"Steady Eddie" play providing copper and fiber cabling inside buildings. Automation Solutions segment up 10% organically. Essential infrastructure growing at moderate but reliable pace.
The Enablers: Materials & Equipment
Even the picks and shovels need their own picks and shovels
United Rentals
URI
$3.7B
Record quarterly rental revenue (Q3 2025)
"Mega projects" (>$500M) driving demand
Specialty rental segment +11% (pumps, power, HVAC)
Fleet mobility to hot spots (VA, TX)
Vulcan Materials
VMC
+13%
Cash gross profit per ton increase (Q3 2025)
1M metric tons cement needed by 2028
Local quarry monopolies = pricing power
Strategic positions near Dallas, Atlanta
Martin Marietta
MLM
$2.32B
Raised 2025 EBITDA guidance
Data center development key driver
Aggregates heavy and expensive to ship
Regional dominance protects margins
THE MATERIALS REALITY
A 100 MW data center is essentially a concrete fortress. The American Cement Association predicts the U.S. will need 1 million metric tons of cement for AI data centers by 2028. These facilities require millions of tons of aggregates for foundations, roads, and structural shells. Because aggregates are expensive to ship, companies with quarries near key hubs (Dallas, Atlanta, Phoenix) enjoy local monopolies and immense pricing power as construction density increases.
Materials and rental companies provide essential infrastructure for the infrastructure builders
Comprehensive Investment Matrix
Mapping the entire value chain
Value Chain Stage Business Model Company Examples Growth Rate Key Characteristic
POWER Equipment Manufacturing Eaton, Generac, Caterpillar Very High Supply-constrained, pricing power
Infrastructure Construction Quanta, MasTec High Massive backlog visibility
COOLING Thermal Systems Vertiv, Modine Very High Pure-play AI beneficiaries
MEP Contractors Comfort Systems, EMCOR Very High Premium margins from complexity
STRUCTURE Civil & Site Work Sterling Infrastructure Very High Mega-site specialist, 25% margins
General Contracting MasTec, Parsons High Diversified infrastructure exposure
CONNECTIVITY Optical Equipment Corning Very High +58% Enterprise AI sales growth
Installation Services Dycom, Belden Medium-High Essential infrastructure services
MATERIALS Aggregates & Cement Vulcan, Martin Marietta Medium Local monopolies, pricing power
Equipment Rental United Rentals Medium Mega-project driven demand
UTILITIES Regulated Power Providers Dominion, Southern Co, PNW Medium Rate base growth, steady returns
PORTFOLIO CONSTRUCTION GUIDANCE
Core Holdings (60%): High-conviction pure plays with explosive growth (Vertiv, Comfort Systems, EMCOR, Sterling) — maximum AI exposure

Diversified Infrastructure (25%): Large diversified contractors with data center exposure plus other infrastructure (Quanta, MasTec) — balanced risk/reward

Defensive Anchors (15%): Materials, utilities, rental equipment for stable exposure to secular buildout trend with lower volatility
Risk Assessment & Mitigation
Four categories of risk with structural mitigants
1. Regulatory & Community
NIMBY zoning battles, utility rate shock backlash, environmental permit challenges. Growing political pressure as residents bear infrastructure costs.
✓ MITIGATION: Large contractors with government relationships and permitting expertise (Parsons, AECOM) navigate complexity. Pre-leasing reduces speculative risk.
2. Supply Chain Constraints
Transformer lead times of 4+ years, skilled labor shortages, material cost inflation. Physical bottlenecks can stall projects despite funding availability.
✓ MITIGATION: Long-term equipment orders locked in (Eaton, Caterpillar), modular/prefab approaches reduce on-site labor (Comfort Systems), vertical integration (MasTec).
3. Execution & Project Delays
Complex mega-projects prone to cost overruns, timeline slippage, quality issues. Hyperscale requirements exceed traditional construction expertise.
✓ MITIGATION: 73% pre-leased capacity creates revenue visibility. Specialized contractors with proven track records (EMCOR, Sterling) command premium pricing that absorbs delays.
4. Economic & Demand Cyclicality
Potential AI hype correction, hyperscaler CapEx cuts, recession reducing enterprise cloud spending, overbuilding in certain markets.
✓ MITIGATION: Multi-year backlogs ($12B+ EMCOR, $39B Quanta) buffer near-term volatility. Structural AI adoption trend independent of short-term sentiment. 0% vacancy rate provides floor.
OVERALL RISK ASSESSMENT: MODERATE-LOW
Unlike pure-play AI software companies facing winner-take-all dynamics, infrastructure providers have tangible backlogs, multi-year contracts, and structural demand drivers largely agnostic to which AI model wins. The 73% pre-leasing rate and near-0% vacancy create unprecedented downside protection. Primary risk is execution (can they build fast enough?) rather than demand (will customers show up?).
Top 5 Conviction Picks
Highest-quality exposure to the AI infrastructure buildout
1. Vertiv Holdings
VRT
The Pure-Play Thermal Leader. Co-engineering liquid cooling solutions with Nvidia for Blackwell deployments. As data centers retrofit for AI, they must replace entire cooling infrastructure with Vertiv's high-density systems. This is a multi-year replacement cycle with 100% attach rate to AI deployments.
Key Metric: 60% organic order growth | 43% Americas sales growth | 27% FY2025 guidance
2. EMCOR Group
EME
The Electrical Complexity Moat. MEP specialist capturing 50%+ of project value through prefabricated electrical skids and complex mission-critical installations. Margin expansion story as AI density drives up revenue per square foot. Deep relationships with hyperscalers create recurring revenue streams.
Key Metric: $12.6B backlog (+29% YoY) | 52% electrical revenue growth | 9.4% expanding margins
3. Quanta Services
PWR
The Grid Bottleneck Solver. Uniquely positioned to build the transmission infrastructure, substations, and increasingly "Total Solutions" on-site power generation that hyperscalers need to energize gigawatt-scale campuses. Scale and union labor force create barriers to competition.
Key Metric: $39.2B record backlog | $7.6B quarterly revenue | Utility + tech convergence driver
4. Eaton Corporation
ETN
The Transformer Scarcity Winner. Supplies mission-critical switchgear, transformers, and power distribution units with 4-year lead times creating massive pricing power. The 30% transformer supply deficit protects margins through 2027+. Data centers account for 17% of revenue with 30%+ segment margins.
Key Metric: 30.3% Electrical Americas margins | 18% backlog growth | Supply-driven pricing power
5. Comfort Systems USA
FIX
The Modular Advantage. Dominates modular manufacturing for data center cooling with 3M sq ft of factory capacity by 2026. Technology sector now 42% of revenue (up from 32% a year ago). Factory-built approach solves skilled labor shortage and accelerates deployment timelines, making them indispensable to hyperscalers racing to deploy capacity.
Key Metric: $9.4B backlog (+65% YoY!) | $553M Q3 operating cash flow | 42% tech revenue exposure
The Long-Term Investment Thesis
Why these infrastructure plays win through the decade
📅
Duration
Backlog visibility extends through 2027-2028 minimum. The 73% pre-leased pipeline creates multi-year revenue certainty. Even if new orders slow, existing commitments sustain growth.
🛡️
Defensibility
Physical and regulatory moats protect incumbents. Transformer shortages, skilled labor scarcity, and permitting complexity create barriers that can't be disrupted overnight.
📊
Diversification
Infrastructure plays reduce pure-AI volatility. These companies win regardless of which AI model dominates, providing exposure to the secular trend with downside protection.
"When everyone's digging for AI gold,
sell the picks, shovels, and blue jeans"
The AI revolution is as much a construction and energy story as it is a software story. The companies identified in this analysis are the structural winners of a capital expenditure cycle that shows no signs of abating through the latter half of the decade. The digitization of the economy requires the industrialization of the cloud, and business is booming for the industrialists.
FINAL TAKEAWAY
The "Ironbone" of AI—the physical infrastructure powering artificial intelligence—represents a multi-year investment opportunity with unprecedented visibility, structural demand drivers, and tangible moats. While AI software valuations remain volatile and speculative, the infrastructure providers offer lower-risk exposure to the same secular trend with the added benefit of massive backlogs and regulated returns (in the case of utilities).
1 / 25