From Power Supply to Lithography Machines: These 14 Stocks Capture Every Bottleneck Layer of AI Expansion
TechFlow Selected TechFlow Selected
From Power Supply to Lithography Machines: These 14 Stocks Capture Every Bottleneck Layer of AI Expansion
While most investors are chasing AI, the real opportunity lies in owning what AI cannot do without.
Navigating Web3 tides with focused insightsAuthor: George Kikvadze
Compiled by: TechFlow
TechFlow Intro: George Kikvadze, Vice Chairman of Bitfury Group, proposes a contrarian thesis: the most profitable opportunities in AI lie not in the model layer, but in infrastructure bottlenecks—power, thermal management, memory, and networking. He identifies seven critical “choke points” across AI systems and discloses his 14-stock “bottleneck portfolio,” which has delivered approximately 60% returns to date. This “bottleneck investing” framework deserves careful study by every investor focused on AI.
To understand where money can actually be made in AI, ignore the headlines—look instead at where the system is straining.
The simplest analogy: today’s AI is like a factory with infinite orders—but its power supply, cabling, and cooling all lag behind.
This mismatch itself is the opportunity.
After detailed due diligence, we have positioned in the following “AI bottleneck” portfolio:
$CEG $GEV $VST $WMB $PWR $ETN $VRT $MU $ANET $ALAB $ASML $LRCX $CIFR $IREN
The Right Question to Ask
Most investors ask: “Who will win AI?” That question is wrong.
The right question is: Where will the system break—and who profits from fixing it?
In markets, dependencies are leverage.
AI dependencies are anything but abstract—they are physical:
- Megawatt-scale power
- Transformer delivery lead times
- Per-rack thermal capacity
- Memory bandwidth
Economic gravity is shifting toward these areas.
The Only Analytical Framework You Need
AI expansion → Infrastructure strain → Forced capital investment → Bottleneck → Pricing power → Earnings upgrades
When demand is inelastic and supply constrained, prices move first, earnings follow, and equity valuations re-rate last.
Why Now?
A few numbers tell the whole story:
Nearly 50% of U.S. data center projects are currently delayed—not due to lack of demand or funding, but because they cannot secure power. Transformer delivery lead times have stretched from 24 months pre-2020 to over five years today. Data center construction takes 18 months. The math simply does not add up.
Hyperscalers’ AI infrastructure spending alone is projected to reach $700 billion in 2026—nearly six times the 2022 level. Amazon: $200 billion; Google: $175–185 billion; Meta: $115–135 billion. None are slowing down.
Semiconductors now account for 42% of the S&P 500 Information Technology sector’s total market cap—more than double their weight at the 2022 bear-market bottom and over four times their 2013 weighting. Semiconductors also contribute 47% of the IT sector’s forward EPS—nearly triple the 2023 figure.
The market is flooding into the compute layer with unprecedented density.
But compute is no longer the bottleneck.
Capital is pouring into chips—while real constraints have shifted elsewhere.
This gap is the trade.
Bottleneck Map: Where Is the Pressure?
- Power: The Foundation
AI cannot scale without electricity. Period.
The U.S. needs to add, every two years, an amount of data center power capacity equivalent to its entire current base—to keep pace with AI demand projections through 2030. Nuclear power is the only baseload source capable of delivering the scale and reliability required by hyperscalers. Yet even the fastest nuclear restarts take years.
Names: $CEG $GEV $VST $WMB
These are not utility stocks—they are AI capacity providers. The market has not yet completed this reclassification. That mispricing is the opportunity.
Constellation Energy ($CEG) operates the largest fleet of nuclear power plants in the U.S. and is one of the few suppliers able to deliver large-scale, reliable, zero-carbon baseload power. Hyperscalers are accelerating long-term power purchase agreements (PPAs) with nuclear providers—and Constellation sits directly on that demand path.
GE Vernova ($GEV) is building the generation backbone of the next energy cycle—including gas turbines, renewables, and grid solutions. As AI-driven power demand accelerates, the ability to deploy electricity quickly and at scale becomes critical—placing GE Vernova’s gas turbine and electrification capabilities squarely at the center.
Vistra Corp ($VST) maintains a diversified generation portfolio—including nuclear, gas, and retail power—enabling it to meet both baseload and peak demand. AI workloads generate highly volatile power demand, making this flexibility exceptionally valuable.
Williams Companies ($WMB) operates one of the largest natural gas pipeline networks in the U.S., supplying fuel to bridge the gap between current demand and future nuclear capacity. Natural gas is the fastest path to incremental power for AI infrastructure expansion. Williams is, in effect, the energy raw-material supplier to AI growth.
Grid & Electrification: The Constraint Behind Power
Generating power is one thing—delivering it is harder.
The U.S. interconnection queue now extends beyond 2030. Over the next decade, more than $50 billion in transmission investment will be needed just to fulfill existing commitments—before adding a single new AI data center.
Names: $PWR $ETN
Schedules slip here—and margins expand here. Companies solving the “last-mile” delivery problem hold durable, long-cycle pricing power.
Quanta Services ($PWR) is the leading contractor for building and upgrading transmission infrastructure—the physical link between generation and consumption. When grid congestion becomes the primary bottleneck to AI expansion, Quanta sits directly in the path of multi-year, non-discretionary capital expenditures. Its backlog is a forward-looking indicator of grid stress.
Eaton Corporation ($ETN) provides distribution systems, switchgear, and power management technologies that enable safe, efficient, large-scale power delivery. As data centers push toward higher power densities and more complex energy flows, Eaton’s components shift from standardized hardware to mission-critical infrastructure.
Thermal Management: The Silent Ceiling
Heat kills performance. Thermodynamics has no software patch.
Next-generation AI facilities target 250 kW per rack—versus 10–15 kW in standard enterprise data centers a decade ago. Liquid cooling is no longer optional—it is mandatory infrastructure. Every GPU sold requires matching thermal capacity, and that ratio is fixed.
Name: $VRT
Vertiv holds near-monopoly status in thermal management for hyperscale data centers. This is one of the most underappreciated links in the entire AI stack—because nobody thinks about cooling until clusters go down.
Vertiv Holdings ($VRT) designs and deploys thermal management systems that keep high-density AI clusters running under extreme power loads. As racks shift from air to liquid cooling, Vertiv sits at the very center of this structural upgrade cycle—expanding in lockstep with AI compute deployment. This is not discretionary spend—it is the prerequisite for operation.
Memory: The Next Bottleneck
AI is shifting from compute-constrained to memory-constrained.
As models grow larger and inference volumes explode, memory bandwidth and capacity—not raw processing power—have become the limiting factors. HBM (High Bandwidth Memory) supply is already tight. The world’s top three AI memory suppliers control over 90% of global HBM output. Micron is the primary Western beneficiary.
Core name: $MU
This is the next wave of earnings upgrades. Most portfolios are not yet positioned for it. They will be—once the market catches up.
Micron Technology ($MU) is one of the few global manufacturers capable of mass-producing advanced HBM—a critical component for AI training and inference workloads. When memory becomes the system’s performance constraint, Micron transitions from a historically cyclical supplier to a structural beneficiary of AI demand. This shift remains underappreciated in valuation—leaving room for sustained earnings upgrades and multiple expansion.
Networking: The Throughput Layer
An AI cluster is only as fast as its slowest connection.
A single network bottleneck can stall an entire multi-thousand-GPU cluster—wasting hundreds of millions of dollars in capital per facility. As cluster sizes scale toward 100,000-GPU configurations, interconnect challenges compound exponentially. One choke point—and the entire line stops.
Names: $ANET $ALAB
Quiet. Critical. Underowned. Nobody talks about networking—until it fails.
Arista Networks ($ANET) builds high-performance network infrastructure that enables seamless data flow across massive AI clusters. When workloads demand ultra-low latency and high throughput, Arista’s software-defined networking becomes essential to maintaining cluster efficiency. Downtime or inefficiency carries enormous cost—Arista captures value by guaranteeing full-speed operation.
Astera Labs ($ALAB) operates inside the data path—ensuring high-speed connectivity between GPUs, CPUs, and memory within AI systems. As cluster density increases, bottlenecks shift from the network edge to chip-to-chip communication—precisely Astera’s domain. In high-performance AI environments, if components can’t communicate fast enough, the entire system slows.
Manufacturing: The Long-Cycle Constraint
No chip manufacturing capability means no AI scaling. No manufacturing tools means no advanced chips.
ASML’s EUV lithography machines require over one year to produce, cost more than $200 million each, and have no credible substitutes. Every advanced chip—from NVIDIA’s H100 to Apple’s M-series—requires their equipment. Lam Research’s etch and deposition tools are embedded in virtually every major wafer fab worldwide.
Names: $ASML $LRCX
Long-cycle constraints. Structurally harder to disrupt than any software moat. Discussion热度 lags far behind what’s warranted.
ASML Holding ($ASML) is the sole supplier of EUV lithography systems—the most advanced chipmaking tools available and a prerequisite for producing cutting-edge semiconductors. Multi-year order backlogs and no viable competitors mean ASML controls a critical choke point in the global chip supply chain.
Lam Research ($LRCX) supplies the etch and deposition equipment that forms the backbone of semiconductor manufacturing. Its tools are deeply embedded across all major wafer fabs—making Lam a recurring, indispensable partner in chip capacity expansion. As AI demand drives continuous capacity buildouts, Lam secures long-cycle revenue directly tied to global semiconductor manufacturing growth.
Misclassification: The Source of Alpha
This is the part most investors overlook—and the most asymmetric opportunity on the entire map.
There exists a class of companies that the market prices as Type A—but whose operations and financial reality are already Type B.
Consider $CIFR (Cipher Digital) and $IREN (IREN Limited).
The market still sees Bitcoin miners.
What they’re becoming is far more valuable: AI power infrastructure and HPC data center platforms.
These companies locked in low-cost power and built infrastructure before demand emerged—exactly the two things hyperscalers are now scrambling to secure.
Cipher Digital has already begun executing its transition—signing 15-year leases with investment-grade hyperscale tenants (its third AI/HPC campus) and securing a $200 million revolving credit facility from top-tier global banks. These are not speculative moves—they are long-cycle revenue commitments.
IREN executes the same strategy across multiple sites—integrating energy acquisition with scalable data center development. Its advantage is speed: it already controls the land, power, and infrastructure needed to pivot to AI workloads.
The market still sees miners. Their balance sheets already look like infrastructure companies.
This gap will close. And when it does, it won’t be slow.
Portfolio Overview
This is not a random collection of stocks—it is a system.
Each position corresponds to a specific, binding constraint in the AI stack—and each constraint must be solved for the system to function. That is discipline.
- Power: $CEG $GEV $VST $WMB
- Grid: $PWR $ETN
- Thermal: $VRT
- Memory: $MU
- Networking: $ANET $ALAB
- Manufacturing: $ASML $LRCX
- Misclassification: $CIFR $IREN
The Cognitive Shift Most Investors Haven’t Completed
We are moving from compute scarcity to infrastructure scarcity.
This means:
- GPUs are no longer the sole narrative
- Power, grid, memory, and thermal management are now the dominant profit drivers
- Returns follow constraints—not hype
Most portfolios remain anchored in the old world.
Risk: Discipline Matters Just as Much
This framework fails under specific conditions—and those deserve honest treatment.
Hyperscaler capex slowdown. If Amazon, Google, and Meta slow infrastructure spending due to margin pressure or weaker-than-expected demand, the assumption of inelastic demand weakens. This is the top risk to monitor—watch quarterly capex guidance as a leading indicator.
Faster-than-expected bottleneck resolution. Government intervention in transformer manufacturing, accelerated nuclear permitting, or restructuring of the interconnection queue could compress the premium on constrained infrastructure. These changes are slow—but real.
Regulatory friction. Power and grid infrastructure intersect with utility regulation, environmental review, and rate-setting agencies. An adverse regulatory shift in this domain could structurally and persistently cap return potential.
The key distinction: this is not a product-cycle bet. Product cycles can reverse in a quarter. Industrial constraints take years to build—and years to unwind. That asymmetry is the point.
Finally
In every industrial era, wealth was not created by the companies that built the trains.
It was created by the companies that owned the rails, coal, and right-of-way.
AI’s rails are measured in megawatts, transformer lead times, and per-rack thermal capacity.
Most investors chase AI. The real opportunity lies in owning what AI cannot operate without.
In every system, headlines follow innovation—profits follow constraints. We focus on constraints, not narratives. Our current return is approximately 60%. As AI infrastructure accelerates, this is not the end of the trade—it remains early. We believe we’re only in the third inning.
Join TechFlow official community to stay tuned
Telegram:https://t.me/TechFlowDaily
X (Twitter):https://x.com/TechFlowPost
X (Twitter) EN:https://x.com/BlockFlow_News
@BitfuryGeorge
