If you missed Nvidia, here’s how to think about the next wave of AI stocks

If you’ve ever stared at Nvidia’s stock chart and thought, “I totally missed it,” you are very much not alone. But here’s the good news: even if Nvidia’s big move is behind you, the broader AI opportunity is not. The key now isn’t to find “the next Nvidia,” but to understand what actually drove Nvidia’s success—and how to spot similar patterns in other businesses.

In this article, I want to walk through what’s happening in the market right now, how rising interest rates affect growth stocks, why AI is still a massive long-term trend, and how two companies—AMD and Marvell Technology—fit into this next phase. This is not a stock tip or a promise of 10x returns. It’s a framework for thinking about AI stocks in a practical, long-term way.

Where we are in the market cycle

Let’s start with the backdrop, because the environment you invest in matters just as much as the individual stocks you pick.

Recently, we’ve seen the major stock indexes wobble. Some areas of the market are still up nicely for the year, but volatility has picked up—especially in September, which has a reputation for being a rough month for stocks.

On top of that, the Federal Reserve raised interest rates again by 0.25%. That pushed the federal funds rate into the 3.75%–4% range, and all 12 voting members agreed on the hike. That’s important: this wasn’t a divided decision. The Fed is still very focused on bringing inflation back toward its 2% target.

Why higher interest rates hit growth stocks harder

Higher interest rates sound like a boring macro headline, but they have very real consequences for your portfolio—especially if you own or are considering high-growth tech or AI stocks.

When rates go up, a few things happen at once:

• Borrowing gets more expensive. Companies pay more to issue debt. Consumers pay more on mortgages, car loans, and credit cards.

• Future profits are worth less today. A lot of fast-growing companies are valued based on what investors think they’ll earn 5–10 years from now. When interest rates are higher, the math used to “discount” those future earnings makes them less valuable in today’s dollars. That tends to hit growth stocks hardest.

• Safe income becomes more attractive. When you can buy high-quality bonds with yields around 5–7%, stocks suddenly have more competition. Investors may decide they don’t need to take as much risk to earn a reasonable return.

All of this doesn’t mean you should avoid growth stocks. It does mean that price and expectations matter a lot more. Chasing any AI stock just because it’s “in AI” is a recipe for disappointment.

Housing: a clear example of how tight money feels

One place where higher rates are really visible is housing. The 30-year mortgage rate has moved above 7%, while home prices are still historically high. That combination creates one of the worst affordability environments we’ve seen in years.

Existing home sales are running around 4 million homes on an annualized basis, which is very low. But interestingly, this isn’t a classic housing crash with plunging prices—it’s more like a transaction recession:

• Many existing homeowners refinanced at 2–4% mortgage rates and don’t want to give those up.

• New buyers are facing much higher monthly payments at today’s 7%+ rates.

The result: buyers can’t afford to buy, sellers don’t want to sell, and activity dries up even if prices don’t fall dramatically.

This matters for investors because housing is a big driver of economic activity and jobs. The Fed is trying to balance three things at once: inflation, a weak housing market, and a broader economy that’s still holding up reasonably well. That tug-of-war is part of why volatility may stick around.

AI spending: the story underneath the noise

While headlines bounce between “AI is overhyped” and “AI will change everything,” I like to ask a simpler question: what are the actual businesses saying with their numbers?

So far, the answer is pretty clear: the AI infrastructure buildout is still massive.

• Nvidia’s data center revenue has exploded, growing well over 100% year over year.

• Broadcom and other chipmakers are reporting strong AI-related demand.

• Hyperscalers (think Microsoft, Amazon, Google, Meta, etc.) are still spending billions on AI data centers, chips, networking, and power.

Instead of asking, “Is the AI trade over?” I think a better question is, “Where is the next wave of earnings growth going to show up?” That’s where two other semiconductor companies start to get interesting: AMD and Marvell.

What made Nvidia special—and what to look for now

Before we talk about AMD and Marvell, it’s worth pausing on what really happened with Nvidia.

Back in early 2023, Nvidia was already telling investors that something huge was happening in data centers. Data center revenue was accelerating, and the company guided for a massive jump in revenue for the next quarter. Wall Street was shocked—and then Nvidia beat even that aggressive guidance.

That was the moment when it became clear we weren’t in a normal chip cycle. We were in the middle of a structural shift toward accelerated computing and AI.

The important lesson isn’t “buy Nvidia.” It’s this:

• Data center revenue started growing very fast.

• Earnings and profit margins scaled up, not just sales.

• The addressable market (AI compute) turned out to be much larger than analysts assumed.

• Customers (big cloud providers) were committing huge amounts of capital.

Those are the kinds of patterns I look for when I evaluate other AI-related companies today.

AMD: not the next Nvidia, but a powerful second platform

Advanced Micro Devices (AMD) is one of the clearest examples of a company that’s benefiting from the AI infrastructure boom—but in its own way.

AMD is already a major player in CPUs, but its AI story goes well beyond that. The company now has exposure across most of the AI stack:

• GPUs (Instinct accelerators for AI workloads)

• CPUs (EPYC processors)

• Networking

• Software

• Rack-scale systems (Helios)

Instead of just selling individual chips, AMD is trying to become more of a complete AI infrastructure platform.

What the numbers are saying about AMD

Here’s where it starts to rhyme with Nvidia’s earlier story:

• Total quarterly revenue recently hit about $11.5 billion, up 50% year over year.

• Data center revenue alone reached roughly $6.7 billion, up 107% year over year.

• Data center now makes up around 58% of AMD’s total revenue.

• Data center operating income was about $2.1 billion, with a 31% operating margin.

A year ago, the data center segment was actually losing money (partly due to a one-time export control charge). Now it’s not just growing—it’s becoming highly profitable. That’s a big deal.

I don’t think AMD needs to “beat” Nvidia to be a compelling investment. The more important question is: can AMD capture a meaningful share of an AI compute market that’s itself getting much larger? If the answer is yes, earnings can grow a lot over time even if Nvidia remains the dominant player.

Helios: AMD’s move into full AI systems

One of the more interesting pieces of AMD’s strategy is Helios, its rack-scale AI infrastructure solution. Instead of just selling chips, AMD is packaging together:

• Instinct MI450 GPUs

• EPYC CPUs

• Networking

• Software

into a full system that hyperscalers can deploy in their data centers.

And this isn’t just a slide in a presentation. AMD already has real customer commitments:

• Anthropic plans to deploy large-scale racks of Instinct MI450 GPUs using Helios.

• Microsoft is expanding its partnership with AMD to use Helios and next-generation EPYC processors in Azure.

Again, I’m not saying this guarantees Nvidia-like returns. But these are the kinds of signs I look for: major customers, real deployments, and revenue and profits moving up together.

Marvell Technology: a “picks and shovels” play on custom AI chips

Marvell Technology (MRVL) is a very different AI investment from AMD. While AMD competes more directly in AI compute (CPUs and GPUs), Marvell is increasingly a behind-the-scenes player in custom AI silicon and networking.

Here’s the idea: Nvidia sells powerful, general-purpose AI accelerators. But some of the biggest cloud providers—Google, Amazon, Microsoft, Meta—are designing their own chips tailored to their specific workloads. Custom chips can improve performance, power efficiency, and cost for those companies.

The catch? Designing cutting-edge semiconductors is very hard. That’s where Marvell comes in.

You can think of Marvell as an AI chip architect and infrastructure partner. It helps large customers design and build specialized chips and supporting components for their data centers.

Marvell’s numbers and why they matter

Marvell’s latest results show how AI is pulling its business forward:

• Record quarterly revenue of about $2.74 billion, up nearly 40% year over year.

• Data center revenue up 46% year over year.

• Over $600 million in operating cash flow.

Management has been clear that AI-related demand—both in custom silicon and networking—is a key driver of this growth.

Google partnership: a sign of where things are headed

One recent development really stands out: Marvell expanded its relationship with Alphabet (Google).

They entered a commercial agreement that covers a wide range of custom semiconductor products tied to Google’s TPU (Tensor Processing Unit) ecosystem, including:

• AI inference accelerators

• Storage controllers

• Network interface controllers

• Memory interface controllers

• Near-memory compute

This isn’t just one small chip. It’s Marvell potentially touching multiple layers of Google’s custom AI infrastructure. Google even structured the deal with commercial incentives through warrants, which is a strong signal of how strategic this relationship is.

For our purposes as investors, the takeaway is simple: this is not what a company does if it’s slowing down AI spending. It’s what a company does when it’s doubling down.

The next phase of AI: an ecosystem, not a single stock

The first phase of the AI boom was easy to summarize: “Buy GPUs.” Nvidia was the obvious winner. But the next phase is broader and more complex.

AI infrastructure now touches:

• Compute (GPUs, CPUs, accelerators)

• Custom silicon designed for specific workloads

• Networking and connectivity

• Memory and storage

• Power and cooling

That’s why I don’t think of AI as “one stock” or even “one type of stock.” It’s an entire ecosystem. Nvidia dominated the first obvious wave, but the next several hundred billion dollars of spending will likely be spread across multiple players.

AMD and Marvell are two names that, in my view, deserve to be in that conversation because they sit in front of large, growing, and very real spending cycles.

AMD vs. Marvell: different roles, different risks

Even though both companies are tied to AI, they’re not interchangeable:

AMD

• More direct competitor to Nvidia in AI compute.

• Exposed to GPUs, CPUs, software, networking, and full rack-scale systems (Helios).

• Thesis: AMD becomes a strong second platform in AI compute and continues to gain data center share.

Marvell

• Less about general-purpose compute, more about custom silicon and connectivity.

• Helps hyperscalers build their own specialized AI chips and infrastructure.

• Thesis: as big tech companies build more custom AI hardware, Marvell’s design and infrastructure capabilities become increasingly valuable.

You don’t necessarily have to choose one or the other. Some investors might own both as different ways to participate in the same long-term trend.

The real risk: learning the wrong lesson from Nvidia

To me, the biggest danger here isn’t that AMD or Marvell are bad companies. It’s that investors learn the wrong lesson from Nvidia’s success.

The wrong lesson is: “If a company is growing fast and mentions AI, it might become the next Nvidia.”

The right lesson is to focus on the earnings trajectory and the quality of growth:

• Are revenues accelerating in a sustainable way?

• Are analyst estimates being revised higher over time?

• Is the addressable market actually expanding, not just in a slide deck?

• Are major customers deploying the products at scale?

• Are margins and profitability improving as the business grows?

Those are the questions I’d ask for any AI-related stock—not just AMD and Marvell.

What I’d watch going forward

If I were tracking AMD and Marvell over the next few years, here’s what I’d keep an eye on.

For AMD

• Instinct GPU ramp: Does MI450 adoption meet or beat expectations?

• Helios deployments: Do hyperscalers roll out Helios at scale?

• CPU share: Does AMD keep gaining share in data center CPUs?

• Software progress: Can AMD keep closing the gap with Nvidia’s CUDA ecosystem?

• Margins: As AI revenue grows, do operating margins and earnings expand meaningfully?

For Marvell

• Custom silicon revenue: How quickly do new hyperscaler programs ramp?

• Google partnership: Does this relationship grow into a multi-billion-dollar stream over time?

• Data center growth: Can the company maintain strong growth in this segment?

• Customer concentration: How dependent does Marvell become on a handful of very large customers?

• Profitability: Is AI-driven revenue coming with healthy margins, not just top-line growth?

How this fits into a long-term investing plan

If you’re a long-term investor, the goal isn’t to perfectly time every move. It’s to build a portfolio that lines up with your risk tolerance, time horizon, and financial goals—then selectively add exposure to big secular trends like AI in a thoughtful way.

That might mean:

• Keeping AI-related stocks as a portion of your total stock allocation, not your whole plan.

• Prioritizing companies where earnings and cash flow are actually showing up, not just hype.

• Being comfortable with volatility—because AI and semiconductor names can move sharply in both directions.

If you like thinking about big trends and how they play out over years, you might also enjoy how I approached a different high-volatility area in this article on why most people miss Bitcoin’s big moves. The core idea is similar: most of the opportunity is missed not because people are dumb, but because they either chase too late or get scared out too early.

If you feel like you “missed it”

Feeling like you missed Nvidia can be emotionally painful. It can trigger FOMO, regret, and the urge to “make it back” by swinging harder at the next hot thing. That’s usually when people get hurt.

Instead of trying to replay the last cycle, I’d encourage you to:

• Focus on where the next earnings inflections might show up.

• Be honest about your risk tolerance and time horizon.

• Use a consistent investing process rather than reacting to headlines.

For me, AMD and Marvell are two companies that check several of the boxes I care about: strong data center growth, real customer demand, and a direct line into the ongoing AI infrastructure buildout. They’re not the only names worth watching, and they’re not guaranteed home runs—but they’re good examples of how to think more clearly about this space.

If you’re still building your investing foundation, it can also help to zoom out and revisit the basics of long-term stock investing. You might find my broader thoughts on evaluating individual stocks and opportunities useful as a complement to AI-focused ideas.

At the end of the day, the goal isn’t to perfectly pick the next Nvidia. It’s to build a resilient, thoughtful plan that lets you participate in big trends like AI without betting your entire financial future on any single stock.