In 2021, "AI stocks" barely existed as a category. Then ChatGPT launched and NVIDIA went from $400 billion to $3 trillion. Here's how to think about AI investing in 2026 — and the five companies with the clearest long-term positions.
NVIDIA Corp
Microsoft Corp
Alphabet Inc
Meta Platforms
Palantir Tech
Updated June 2026 · Originally published October 2021
The original version of this article was written in late 2021, when artificial intelligence was still a technology story rather than a financial one. The stocks it recommended were solid companies. They were not, in any meaningful sense, AI stocks — because AI wasn’t yet the organising investment theme of the decade.
That changed in November 2022 when OpenAI released ChatGPT. Within weeks, the conversation about AI shifted from research labs and tech conferences to boardrooms, living rooms, and trading floors. And within months, the financial markets had made their judgment: AI was the most significant technology investment opportunity since the internet, and the companies enabling it were going to be worth vastly more than they had been.
Four years later, that judgment looks prescient. NVIDIA has grown from a ~$400 billion company to a ~$3 trillion company. Microsoft’s $13 billion bet on OpenAI has reshaped its product positioning across every business line. The AI infrastructure buildout — data centres, chips, electricity — is driving capital expenditure at a scale that is moving entire sectors of the economy.
This is an updated guide to the AI stocks worth understanding in 2026. Not a buy list — a framework for thinking about where the value actually sits in the AI economy, and which companies have the clearest path to capturing it.
“Every major technology cycle has an infrastructure phase and an application phase. We are still, in 2026, primarily in the infrastructure phase of AI.”
How to Think About AI Investing
Before looking at specific stocks, it helps to understand where in the AI value chain the money is actually being made — because different companies sit at very different points, with very different risk and return profiles.
At the base is compute — the chips and hardware that power AI training and inference. NVIDIA dominates this layer with its GPU architecture, and the moat is substantial: years of software investment (CUDA), a developer ecosystem that competitors have spent years trying and largely failing to replicate, and manufacturing relationships that cannot be rebuilt quickly. The risk is that the moat erodes as competitors invest billions in alternatives.
Above that is cloud and infrastructure — the hyperscalers (Microsoft Azure, Google Cloud, Amazon AWS) who are spending hundreds of billions on data centres to deliver AI services to enterprise customers. They are simultaneously AI investors (all have stakes in or partnerships with major AI labs) and AI distributors. Their size and existing customer relationships give them structural advantages, though none of them has won the AI infrastructure race decisively.
At the top is applications — the products that end users and businesses actually interact with. This is where the long-term value will ultimately accumulate, but it is also the most competitive layer and the hardest to predict. The application that dominates every industry’s AI adoption is not yet built — which means extraordinary opportunity and extraordinary uncertainty.
The company that accidentally became the engine of the AI economy.
9.5
9.0
8.0
8.8
NVIDIA was a gaming chip company for most of its history. Then it turned out that the architecture that made its GPUs exceptional for rendering video games — massively parallel computation — was also exactly what you needed to train large neural networks. The company had built CUDA, its parallel computing platform, over more than a decade. When AI went from interesting to essential, NVIDIA had a decade’s head start that competitors cannot erase quickly.
The numbers are extraordinary. Revenue grew from $27 billion in fiscal 2023 to over $130 billion in fiscal 2026. Net margins are above 50%. Every major hyperscaler, AI lab, and government computing programme in the world is buying its chips. The H100 and H200 GPUs are not just products — they are the currency of the AI economy. You cannot build a competitive AI model without access to them, and NVIDIA controls access.
The risk at this valuation is significant. A $3 trillion market cap prices in a lot of perfection. Competition from AMD, Intel, and custom silicon from Google, Amazon, and Microsoft is real, even if it hasn’t yet dented NVIDIA’s dominance. Export restrictions on advanced chips to China have created headwinds that could intensify. And the law of large numbers means sustaining the growth rates that justified this valuation becomes harder every quarter.
None of this means NVIDIA is a bad investment. It means it is a high-conviction bet on a company that may well retain its dominance — or may find that dominance slowly eroded by the very customers it is currently enriching. That is the AI investing dilemma in miniature.
The company that paid $13 billion to be on the right side of history.
9.5
7.5
4.5
8.2
Microsoft’s investment in OpenAI was one of the most consequential bets in corporate history. Satya Nadella reportedly pushed hard for it when others at Microsoft were uncertain. The result was that Microsoft got to embed GPT-4 and its successors into every product in its portfolio — Office, Teams, Azure, GitHub Copilot, Bing — before any competitor had comparable access.
GitHub Copilot has become the most widely used AI coding assistant in the world, with millions of developers paying monthly subscriptions. Microsoft 365 Copilot is being rolled out to enterprise customers at premium pricing. Azure’s AI services revenue is growing at triple-digit rates. The company’s existing relationships with virtually every major enterprise on earth give it a distribution advantage that pure-play AI companies simply cannot match.
Microsoft is also one of the more defensively positioned AI stocks. Its non-AI businesses — Office, Xbox, LinkedIn, cloud infrastructure — generate enormous, predictable cash flows that cushion any near-term disappointment in AI monetisation. At current valuations, it doesn’t need everything to go right. It just needs the trajectory to continue.
The company that invented modern AI — and briefly looked like it was going to lose because of it.
9.2
7.2
3.5
8.0
Google’s researchers invented the transformer architecture — the technology that underlies virtually every modern AI language model, including the ones that ChatGPT runs on. The paper “Attention Is All You Need,” published by Google researchers in 2017, is arguably the most consequential piece of computer science research of the decade. And then OpenAI and Microsoft used it to build products that briefly looked like they might threaten Google’s core search business.
That threat has moderated. Google’s Gemini models are genuinely competitive with GPT-4. AI Overviews have been integrated into Google Search and, despite initial stumbles, are driving engagement rather than undermining it. Google Cloud’s AI services are growing rapidly and taking share from Azure in certain enterprise segments. DeepMind continues to produce research that defines the frontier.
The investment case for Alphabet rests on a simple observation: it is the best-positioned company in the world to benefit from AI-enhanced search, and it trades at a discount to Microsoft and NVIDIA on almost every valuation metric. The antitrust risks are real — a judge ruled that Google holds an illegal monopoly in search in 2024, and the remedies process is ongoing. But the underlying business is extraordinarily strong, and the AI transition is going better than the darkest 2023 predictions suggested.
The biggest open-source AI bet in history — and it’s working.
8.8
8.0
4.0
8.3
Meta’s AI strategy is unusual among the big tech companies: it has committed to releasing its Llama models as open source, making powerful AI available to anyone who wants to build with it. Mark Zuckerberg has been explicit about the reasoning — Meta needs AI to be a commodity so it can compete on distribution rather than model quality, and open-sourcing its models puts competitive pressure on OpenAI and Anthropic while giving the developer community tools that embed Meta’s approach into the wider AI ecosystem.
The business impact has been striking. Meta AI, embedded across Facebook, Instagram, WhatsApp, and Messenger, now has over 400 million monthly active users — making it one of the most widely used AI assistants in the world by raw numbers. AI-powered ad targeting improvements have driven revenue growth above analyst expectations for several consecutive quarters. The advertising business, which many wrote off in 2022 during the iOS privacy changes crisis, has emerged stronger and more AI-dependent than before.
Meta’s Reality Labs division — the metaverse investment that generated enormous scepticism — has also found firmer footing with the commercial success of Ray-Ban Meta smart glasses, which use AI as their primary differentiator. The glasses that let you ask an AI about what you’re looking at are selling. The metaverse living room that Mark Zuckerberg showed in 2021 is still not here. But the hardware strategy has evolved into something more immediately commercial.
The company that spent 20 years building the thing everyone suddenly needs.
7.8
8.5
8.5
7.5
Palantir is the most polarising stock on this list. Its bulls argue it is uniquely positioned at the intersection of AI and national security — the company that governments and intelligence agencies trust with their most sensitive data, now offering that same infrastructure to commercial enterprises through its AIP (Artificial Intelligence Platform) product. Its bears argue that the valuation has long since detached from any reasonable near-term earnings trajectory and that the fandom around CEO Alex Karp borders on the cultish.
Both sides have a point. Palantir’s US government business is genuinely irreplaceable — it sits inside the most sensitive data environments in the world, has taken years and significant security clearances to access, and cannot be easily displaced by a competitor. Its US commercial business has been growing at over 50% annually, driven by enterprises paying for AIP, its platform for deploying AI on proprietary data without exposing that data to third parties. The “AI Bootcamp” model — where Palantir brings clients in for intensive sessions to demonstrate AIP’s capabilities — has proven unusually effective at converting prospects.
The valuation risk is real and should not be dismissed. Palantir trades at over 80x forward earnings — a multiple that requires flawless execution for many years to justify. Any disappointment in commercial growth, any loss of a major government contract, any broader market rotation away from high-multiple growth stocks, and the share price could fall dramatically. This is a stock for investors with high conviction and a long time horizon, not a defensive position.
Final Thoughts
AI investing in 2026 is not the same as AI investing in 2021, or even 2023. The obvious picks have already been made. NVIDIA at $3 trillion is not the same opportunity as NVIDIA at $400 billion. The easy money — buying the shovel-makers before everyone understood there was a gold rush — has largely been made.
What remains is harder: understanding which companies have durable advantages in an AI economy that is still being built, which are temporarily benefiting from a buildout that will eventually slow, and which are genuinely undervalued relative to their long-term position. That requires real analysis, real conviction, and real patience.
The five companies above represent different bets within the AI theme — from the infrastructure layer (NVIDIA) to the platform layer (Microsoft, Alphabet, Meta) to the enterprise AI application layer (Palantir). A diversified position across them gives broad AI exposure. A concentrated position in any one of them is a specific conviction bet that deserves specific research.
Whatever you do, don’t invest in AI stocks because AI is exciting. Invest because you understand the specific company, the specific competitive position, and the specific price you’re paying for that position. The excitement is already in the price. What matters now is whether the underlying businesses can justify it.
Disclaimer: This article is for informational purposes only and does not constitute financial advice. Stock prices shown are approximate and for illustrative purposes only. Always conduct your own research before making investment decisions.