3 AI stocks Brad Gerstner thinks still have massive upside

Artificial intelligence is no longer just about chatbots and flashy demos. According to top tech investor Brad Gerstner, the real opportunity sits in the infrastructure powering AI: data, compute, and tokens. In other words, the companies building the rails for the AI economy may have the biggest long-term upside.

In this article, we’ll break down Gerstner’s core AI thesis, what “token flow” actually means, how he thinks about portfolio management in an AI boom, and three AI stocks that could benefit over the next decade: AMD, Lam Research, and Microsoft.

AI is becoming a full economic ecosystem

Gerstner’s key message is simple: AI is turning into a full-blown economic ecosystem built around three pillars—data, compute power, and token consumption. Every time you run an AI query, generate an image, use an AI agent, or automate a workflow, you’re consuming tokens. Those tokens sit on top of massive data and compute infrastructure.

That’s why he focuses less on the consumer-facing apps and more on the companies that:

• Store and organize data
• Provide the compute power to run AI models
• Enable the token flow that powers AI usage

Historically, the biggest winners in technology booms are often the “picks and shovels” providers—the ones selling tools to everyone else. Gerstner believes AI will follow the same pattern.

Why data infrastructure is the real engine of AI

In his CNBC appearance, Gerstner highlighted data infrastructure names like Snowflake, Databricks, and ClickHouse. His argument: as AI usage grows, token consumption explodes, and that directly accelerates demand for data platforms.

He calls this effect “token flow.” The logic is straightforward:

• Every AI interaction consumes tokens
• Tokens require models to run on top of large, clean datasets
• The more tokens consumed, the more data has to be stored, processed, and served
• Data infrastructure companies see their core businesses accelerate as a result

That’s why he sees a major difference between data platforms like Snowflake and application-layer companies like Salesforce. Data platforms don’t have to “win” the AI model war. No matter which model dominates—OpenAI, Anthropic, Google, or someone new—they all need clean, well-managed data. That makes data infrastructure a leveraged bet on the entire AI economy, not just one model provider.

If you want a broader view of these structural shifts, it’s worth pairing this idea with the mega themes discussed in 3 AI mega trends that could change your life.

Letting winners run—without losing discipline

Gerstner also spent time on something most investors struggle with: how to manage a portfolio in a parabolic AI market.

He follows two key principles:

1. Let your winners run. Great companies can compound for years. Selling too early just because a stock is up often means missing the biggest part of the move.

2. Trim when things go parabolic. When a stock runs far beyond his price targets, he trims exposure and rotates capital into better risk/reward opportunities. That doesn’t mean the story is broken—it’s about risk management.

To make this practical, he uses a simple exposure framework he calls “3-6-9”:

• When fear is high and markets are washed out, he’s comfortable being closer to “large” exposure (around 90%).
• When euphoria is rampant and everything feels frothy, he scales back toward a more “medium” exposure (around 60% or 30%).

For someone with $100,000 in cash wanting to start investing in AI today, he wouldn’t go all-in. Instead, he’d put roughly 30% to work and wait for volatility and pullbacks to deploy more.

The coming wave of AI IPOs and capital rotation

Another underappreciated factor Gerstner raised: the impact of huge AI IPOs like SpaceX, Anthropic, and OpenAI.

Institutional investors don’t have unlimited capital. When a massive IPO like SpaceX comes to market, they often need to sell other holdings to free up cash. That can create short-term pressure on existing tech and AI stocks, even if the long-term AI thesis remains intact.

At the same time, he points out that U.S. capital markets are enormous—on the order of tens of trillions of dollars. So while big IPOs can cause temporary volatility and rotation, they’re unlikely to “break” the market.

Why SpaceX is really an AI infrastructure play

Gerstner is especially excited about SpaceX—not just as a space or communications company, but as a future AI infrastructure giant.

Most people focus on rockets and Starlink. Gerstner focuses on something else: data centers and tokens. In his view, Elon Musk is building some of the largest data centers on Earth (and potentially beyond), and “no one is better at turning electrons into tokens.”

Here’s what that means:

• Data centers convert electricity (electrons) into compute
• Compute runs AI models that generate tokens
• Tokens are how AI usage is metered and monetized

SpaceX’s growing AI partnerships and infrastructure footprint could make it a major player in the AI economy, not just aerospace. It’s part of a broader shift where AI, compute, energy, and data infrastructure are converging into a single strategic battleground.

We’re still early in the AI capital cycle

Across the interview, Gerstner is both bullish and realistic. He believes AI is the strongest secular trend in markets today and that we’re still in the early innings of a multi-year infrastructure buildout—similar to the rise of the internet and cloud computing.

At the same time, he expects:

• Sharp pullbacks and corrections
• Periods of consolidation
• Valuations occasionally running ahead of fundamentals

The key is to avoid two extremes: permanent fear and reckless euphoria. You can be long-term optimistic on AI while still managing risk intelligently.

3 long-term AI stocks to watch

Against that backdrop, three companies stand out as potential long-term winners across different layers of the AI stack: AMD, Lam Research, and Microsoft.

AMD: powering the AI compute race

Advanced Micro Devices (AMD) has transformed from a struggling chipmaker into a central player in high-performance computing and AI.

Most people still associate AMD with gaming and PC chips, but the real story now is data centers and AI accelerators. As AI models get larger and more complex, the demand for powerful, energy-efficient chips is exploding. Without these chips, AI models can’t be trained or deployed at scale.

Here’s why AMD matters in the AI era:

• Data center focus: AMD’s data center business has become one of its key growth engines, driven by AI workloads and cloud computing demand.
• AI accelerators: Its Instinct line of AI accelerators is gaining traction as enterprises and cloud providers look for alternatives in the AI hardware market.
• Secular tailwind: Global AI infrastructure spending could reach hundreds of billions of dollars over time, and AMD is positioned at the heart of that cycle.

AMD has also executed a remarkable strategic turnaround. It’s moved up the value chain from being seen as a lower-tier CPU provider to a serious high-performance computing and AI infrastructure player. That shift is crucial because markets tend to reward companies that become more strategically important over time.

Other strengths include:

• Diversified exposure across gaming, embedded systems, enterprise, and cloud
• Improved balance sheet and profitability
• A management team focused on long-term execution rather than short-term hype

There are risks—semiconductors are cyclical, competition is fierce, and valuations can be volatile—but the core question for long-term investors is: will AMD be more important to the global economy 10 years from now? If AI adoption keeps accelerating, the answer increasingly looks like yes.

Lam Research: the quiet AI enabler in chip manufacturing

Lam Research (LRCX) is not a household name, but it plays a critical role in the AI supply chain. It doesn’t design chips—it builds the advanced equipment used to manufacture them.

Without companies like Lam, the cutting-edge AI chips powering today’s models simply couldn’t be produced at scale.

Lam specializes in semiconductor manufacturing equipment—highly complex, extremely precise systems that enable chipmakers to push toward smaller, faster, and more powerful designs. As AI demand grows, chipmakers must invest heavily in new tools and processes to keep up.

Key reasons Lam Research is strategically important:

• High barriers to entry: Developing leading-edge chipmaking equipment requires decades of engineering expertise and deep customer integration. It’s not easy to disrupt.
• Memory demand: AI workloads consume huge amounts of high-bandwidth memory and storage. Producing next-generation memory chips requires advanced fabrication tools—an area where Lam plays a major role.
• Model-agnostic exposure: No matter which AI models or platforms win, they all need advanced chips. And advanced chips need advanced manufacturing tools.

Financially, Lam has shown it can navigate semiconductor cycles while maintaining strong profitability and free cash flow. It has also consistently returned capital to shareholders through dividends and buybacks, all while investing heavily in R&D.

As AI, cloud computing, automation, and high-performance computing all scale up, global semiconductor demand is likely to remain structurally strong. That, in turn, supports a long runway for companies enabling chip production.

If you’re interested in how AI is reshaping the chip landscape more broadly, you may also want to look at our deep dive on Micron’s AI boom in Micron’s AI chip boom and what it means for investors.

Microsoft: embedding AI across the enterprise

Microsoft (MSFT) may be the most obvious AI stock on this list—but that doesn’t make its position any less powerful. It’s no longer just an operating system and Office company. Microsoft is rapidly becoming one of the central pillars of the AI economy.

What makes Microsoft so formidable is its breadth. It’s integrating AI across:

• Cloud infrastructure (Azure)
• Productivity tools (Office, Teams, Copilot)
• Developer tools and platforms
• Cybersecurity
• Search and knowledge work
• Enterprise workflows and automation

Instead of asking customers to adopt entirely new platforms, Microsoft is layering AI into products businesses already rely on. That dramatically lowers friction and speeds up adoption.

Key advantages for Microsoft in the AI race:

• Azure and cloud scale: AI workloads are driving a new wave of cloud spending, and Azure is one of the main beneficiaries.
• Enterprise trust: Large organizations already trust Microsoft with mission-critical systems. That trust is a huge edge when selling AI solutions that touch sensitive data and workflows.
• Recurring revenue: Its subscription model (Office 365, Azure, etc.) creates predictable cash flow and makes it easier to monetize new AI features.

Productivity is one of the biggest near-term opportunities. AI integrated into Office, Teams, and developer tools can automate repetitive tasks, summarize meetings, generate content, analyze data, and more. When software directly saves time and money, enterprises are usually willing to pay for it.

On top of that, Microsoft has:

• One of the strongest balance sheets in the world
• Massive free cash flow to fund AI infrastructure buildouts
• A long history of successfully navigating major tech shifts—from PCs to cloud to AI

There are risks around competition, regulation, and high expectations, but investors aren’t just betting on a single AI product here. They’re betting on a global ecosystem where AI is being woven into almost every Microsoft offering.

How these three stocks fit into the AI stack

While AMD, Lam Research, and Microsoft operate in different parts of the market, they’re all plugged into the same AI megatrend:

• AMD powers the compute infrastructure that runs AI models.
• Lam Research enables the manufacturing of the advanced chips those models run on.
• Microsoft embeds AI into the software and cloud platforms enterprises use every day.

Together, they offer three distinct but complementary ways to gain exposure to long-term AI growth—hardware, manufacturing infrastructure, and software/cloud.

As always, none of this is investment advice. AI stocks can be volatile, and it’s essential to do your own research, understand your risk tolerance, and think in terms of years, not weeks. But if Brad Gerstner’s thesis is right, the companies controlling data, compute power, token generation, and AI infrastructure could help define the next decade of market winners.