SPECIAL REPORT
7 AI Stocks To Buy For 2026
The AI Infrastructure Watchlist
By David Thomas, Founder, North Tech Capital
A quick reality check.
AI isn’t a trade anymore. It’s an economic build-out.
The market’s first instinct was to treat AI like a single story: “buy the AI winners.”
But what we’re actually living through looks more like the early stages of the cloud era, or the smartphone era, or the modern data centre era:
- a long capex cycle
- infrastructure constraints
- vendor concentration
- second-order winners that don’t look like “AI plays” at first glance
- and a market that repeatedly misprices the difference between hype-driven demand and durable, repeatable demand
That matters because the next few years won’t reward every company that says “AI” on an earnings call.
They’ll reward the companies that can convert AI demand into:
- durable revenues
- expanding operating profits
- and defensible cash flows
That’s what this report is designed to help you see.
Not predictions. Not hot tips. A clearer map.
Why the AI cycle is entering a new phase
The first phase of any technology cycle is narrative.
The second phase is infrastructure.
The third phase is economics.
We’re moving from narrative into economics.
In 2024 and 2025, the market paid up for exposure. Multiples expanded because investors wanted to own the idea of AI.
In 2026 and beyond, markets become more selective because the questions get sharper:
- Who is absorbing the capex?
- Who is extracting the margin?
- Who has pricing power when competition rises?
- Which business models are structurally advantaged in a higher-rate, tighter-liquidity regime?
- Which “AI winners” are actually just “capex-heavy projects” with fragile returns?
This is where many investors get chopped up.
They buy the most obvious “AI winners” after the easy gains have happened, right as the market starts demanding proof.
Proof doesn’t mean more demos.
Proof means margins. Proof means cash flow. Proof means returns on capital, and earnings growth.
The core insight: AI is a supply chain, not a product
Most investors look at AI as a product wave.
But the better mental model is a supply chain.
AI requires:
- accelerated compute
- high-speed networking
- massive storage and data movement
- hyperscale cloud infrastructure
- software platforms that make the hardware usable
- and an energy and cooling footprint that is becoming a bottleneck
If you want a durable way to invest in AI, you don’t start with “what’s exciting.”
You start with: what must exist for AI to function at scale?
That’s where the infrastructure layer comes in.
It’s also where the “winners” tend to be fewer, larger, and more durable than investors expect.
The North Tech Capital lens
We do not build discounted cash flow models and pretend we can forecast the future to the decimal.
We care about whether a business can compound.
So we focus on six things:
1) Cash flow durability
Can the company generate real free cash flow through a full cycle, or does it depend on perfect conditions?
2) Returns on capital
Does the business create value as it grows, or does growth require so much capital that returns get diluted?
3) Moat strength and pricing power
When customers have choices, who can hold margins?
4) Structural growth drivers
Is growth supported by a long runway, or is it a short-term spending burst?
5) Capital allocation discipline
Is management reinvesting intelligently, buying back shares prudently, and avoiding empire-building?
6) Liquidity and rate regime sensitivity
Some models thrive when liquidity is abundant. Others can still win when money is more expensive.
That framework is why North Tech Confidential exists. It’s also why we keep our focus on large, durable platforms.
The AI Infrastructure Watchlist
This is a research watchlist: seven public companies that sit at critical points in the AI infrastructure stack.
These are not “the only winners.”
They are, however, the businesses we believe are most likely to remain essential as AI shifts from novelty into real economic productivity.
1) Nvidia
The compute platform, not just a chip company
Nvidia is the choke point for accelerated compute, but its real advantage is platform depth.
The market often frames Nvidia as “the AI GPU leader.” That’s true, but incomplete.
The deeper point is this:
When a technology becomes critical infrastructure, ecosystems form around it. Software, tooling, developer workflows, compatibility layers, and switching costs all expand.
That’s how you get pricing power that doesn’t disappear overnight.
The key question for investors going forward isn’t “will AI continue?”
It’s “how much of AI’s economic value pool does Nvidia continue to capture, and how durable is that capture as competition rises?”
In an infrastructure cycle, competition will always rise. The winners are the ones that keep their economics anyway.
2) Broadcom
Networking + custom silicon: the less crowded side of AI
Broadcom sits in two places that are becoming more important as AI scales:
- high-speed AI networking
- custom silicon (where hyperscalers want alternatives and optimization)
As AI workloads expand, the bottleneck is increasingly not compute. It’s moving data fast enough.
That pushes value into networking and interconnect.
Broadcom’s role here is not just “AI exposure.” It’s leverage to a structural requirement: if hyperscalers want scale, they need networking that can keep up.
And unlike many “AI software” stories, this is tied to real build-out demand.
Important note for Global Tech 15 followers: Broadcom (AVGO) is a core holding precisely because it sits in this infrastructure layer.
3) Taiwan Semiconductor Manufacturing Company
The manufacturing choke point
In most technology cycles, the “obvious” winners aren’t always the most durable.
But manufacturing chokepoints are different.
When demand is real, it doesn’t matter how many AI startups exist or how many chips are designed.
You still need a small number of manufacturers that can deliver at the leading edge.
That’s why Taiwan Semiconductor Manufacturing Company matters.
It’s not about being a consumer brand.
It’s about being the hardest part of the stack to replicate.
In a world where governments care about supply chain security and companies care about reliable access, chokepoints tend to gain strategic value.
4) Microsoft
AI distribution at scale
The best AI products don’t win because they are clever.
They win because they are distributed.
Microsoft’s advantage is not “having AI.” Everyone has AI.
Microsoft’s advantage is enterprise embed:
- productivity workflows
- corporate IT environments
- and a cloud platform that already runs mission-critical workloads
When you can layer AI into existing workflows, you don’t need customers to change behavior. You just need them to adopt upgrades.
That is a structurally powerful model.
From a valuation-first perspective, the question is never “is Microsoft good?”
The question is always: “how much are we paying for that quality?”
5) Amazon
The industrial backbone of cloud demand
Amazon is an infrastructure business at its core.
Retail is the consumer-facing engine, but the strategic value is the cloud platform and logistics infrastructure it has built.
As AI becomes more mainstream, companies don’t want experiments. They want deployment.
Deployment happens in cloud environments with security, compliance, and enterprise-grade scalability.
This is where hyperscalers become the “picks and shovels” of AI adoption.
The risk for investors is that capex intensity rises across the industry. The opportunity is that winners gain scale and amortize costs better than everyone else.
6) Alphabet
AI-native research + distribution
Alphabet has two advantages that rarely show up cleanly in simplistic AI debates:
- world-class AI research capability
- distribution at global scale
The path to monetization can come in phases. Markets sometimes punish that.
But in infrastructure cycles, the companies with deep internal capability and distribution can often absorb volatility and still win long-term.
The investor’s job is to separate “near-term noise” from “long-term positioning.”
7) Meta Platforms
A cash-flow machine that uses AI to protect the engine
Meta is not the classic “AI infrastructure” pick.
But it belongs here because AI is increasingly a profit protection mechanism:
- better targeting
- better recommendation systems
- better monetization efficiency
- and lower marginal costs of content relevance
Some companies use AI to chase growth.
The stronger models use AI to reinforce an already dominant cash engine.
That kind of feedback loop tends to be underestimated early in cycles.
What to do with this watchlist
If you only take one lesson from this report, take this:
The goal isn’t to own “AI.” The goal is to own durable economics inside the AI build-out.
Practical ways to use this list:
1) Build your own “infrastructure-first” watchlist
Add companies you already follow, but categorize them by where they sit:
- compute
- networking
- manufacturing
- cloud
- software platforms
- energy and cooling
Most portfolios are unintentionally concentrated in one slice.
2) Track the cycle through bottlenecks
Infrastructure cycles tend to reveal themselves through bottlenecks.
When the bottleneck shifts, the winners shift.
First it was compute. Then it became data movement. Next it may be power and cooling.
3) Keep valuation discipline
A great company can be a poor investment at the wrong price.
AI cycles create moments where investors pay for perfection.
Your edge comes from refusing to do that.
Final Word
The AI build-out is no longer theoretical.
Capital is being deployed. Infrastructure is being laid. And the companies that matter are starting to separate from the ones that merely captured attention.
The opportunity now is not to chase headlines, but to understand where durable economics are forming as AI moves from experimentation to real-world deployment.
That’s the lens we use at North Tech Capital.
A quick reminder
You’re now subscribed to North Tech Confidential.
Through the newsletter, we’ll continue to:
- break down the structural forces shaping technology markets
- cut through short-term noise and headline-driven volatility
- and explain how long-term investors should think about positioning capital across cycles
We publish with discipline, not urgency. With context, not predictions.
And with a long-term framework built around fundamentals, valuation, and capital flows.
Keep an eye on your inbox for the next issue.
In the meantime, you can explore our latest research and market commentary by visiting the homepage.
Disclosure
North Tech Capital provides research and market commentary for informational and educational purposes only. Nothing in this report constitutes investment advice or a recommendation to buy or sell any security.