Doug Clinton thinks Nvidia and Google are the safest AI investments you can make right now. The entire tech press nodded in agreement when he said it in April 2026. Meanwhile, Google just committed to spending up to $185 billion on AI infrastructure this year alone—a number so staggering it makes venture capital look like pocket change.
So which is it? Are these companies printing money or lighting it on fire?
The Case for Calling Anything “Safe” in AI
Clinton, who runs Intelligent Alpha, isn’t some random talking head. He’s paid to identify winners in the AI space, and his picks make sense on paper. Nvidia makes the chips that power every serious AI operation. Google has search dominance, cloud infrastructure, and enough cash to outspend competitors into oblivion.
These aren’t scrappy startups hoping their Series B doesn’t run dry. They’re established giants with actual revenue, actual products, and actual moats. When financial analysts talk about “safe bets,” they mean companies that probably won’t implode if the AI hype cycle takes a breather.
But here’s what bothers me about that framing: safe compared to what? Compared to investing in a company that just launched its first chatbot? Sure. Compared to literally any other sector of the economy? That’s where it gets interesting.
The $185 Billion Question
Google’s capital expenditure guidance for 2026 sits somewhere between $175 billion and $185 billion. Read that number again. That’s not R&D. That’s infrastructure—data centers, chips, cooling systems, the physical backbone of AI at scale.
Sundar Pichai is betting the company’s future on AI infrastructure paying off. Not might pay off. Not could pay off. Has to pay off, because you don’t spend that kind of money on a maybe.
This is either the smartest move in tech history or the most expensive mistake. There’s no middle ground when the numbers get this big. Google needs AI to generate returns that justify building what amounts to a small country’s worth of computing power.
Nvidia’s Different Problem
Nvidia doesn’t have Google’s spending problem. They have a dependency problem. Every major AI company needs their chips, which sounds great until you remember that every major AI company is also trying to design their own chips.
Google has TPUs. Amazon has Trainium. Microsoft is working with AMD. The list goes on. Nvidia’s current dominance is real, but it’s also the exact kind of dominance that makes customers nervous enough to fund alternatives.
The company that sells picks and shovels during a gold rush does well—until everyone starts making their own tools. Nvidia knows this. Their customers know this. The question is whether Nvidia can stay far enough ahead that custom chips remain a poor substitute.
What “Safe” Actually Means Here
When Clinton calls these stocks safe, he’s making a specific claim: these companies will survive and probably thrive regardless of how AI shakes out. Even if the current AI boom cools off, Nvidia still makes gaming GPUs and data center chips. Google still owns search and YouTube.
That’s different from saying the stocks can’t drop 30% if sentiment shifts. It’s different from saying they’re guaranteed to beat the market. Safe means they’re unlikely to become cautionary tales.
For investors who want AI exposure without betting on which startup will win, that logic holds. You’re not trying to pick the next OpenAI. You’re buying the infrastructure layer that wins regardless of which applications succeed.
The Honest Take
Are Nvidia and Google safe AI bets? Safer than most, sure. But calling anything in AI “safe” in 2026 requires ignoring some pretty significant risks. Google’s spending could outpace returns. Nvidia’s moat could erode faster than expected. The entire AI market could consolidate in ways that hurt even the winners.
Clinton’s probably right that these are the safest public market plays if you want AI exposure. That doesn’t mean they’re safe in any absolute sense. It means they’re the least likely to completely blow up in your face.
Sometimes that’s the best you can hope for.
🕒 Published: