You’ve seen the charts. They look like a vertical wall. It’s the kind of growth that makes your stomach do a flip, half from excitement and half from that nagging fear that you’re the last one to the party. Everyone and their cousin is asking: is nvidia still a buy, or are we just watching a very expensive bubble start to hiss?
Honestly, the answer isn't a simple yes or no. If you're looking for a "get rich quick" play in 2026, you might be a few years late. But if you’re looking at the fundamental plumbing of the global economy, things get much more interesting.
The world is being rebuilt. Literally. Jensen Huang, Nvidia's CEO, calls it the "next industrial revolution," and while that sounds like typical Silicon Valley hyperbole, the data from data center capex (capital expenditure) suggests he’s not just blowing smoke. We aren't just talking about chatbots anymore. We are talking about drug discovery, weather forecasting, and autonomous robotics that actually work.
The Trillion-Dollar Question: Why Nvidia Keeps Defying Gravity
Why do people keep betting against this thing only to get steamrolled?
It’s the moat. Most people think Nvidia is a hardware company. They think, "Oh, Intel or AMD or some startup like Groq will eventually make a faster chip, and then Nvidia is toast." That misses the entire point of the Cuda ecosystem. Cuda is the software layer that developers have been building on for nearly two decades. It’s sticky. It’s hard to migrate away from. Imagine trying to tell every developer in the world to stop using Windows and switch to a brand-new OS overnight. It doesn't happen.
Then there’s the Blackwell architecture. When Nvidia moved from Hopper (H100) to Blackwell (B200), the jumps in performance weren't incremental. They were massive. We're talking about a reduction in energy consumption and a surge in inference speeds that make the previous generation look like a calculator.
📖 Related: 3 000 pesos in us dollars: Why the Exchange Rate is Often a Lie
But here’s the rub.
The valuation is eye-watering. You’re paying a premium for perfection. When a company is priced this high, even a "good" earnings report can send the stock tumbling if it isn't "miraculous." We saw this in late 2024 and throughout 2025—volatility is the price of admission here. If you can’t handle a 20% swing in a single week, you shouldn't even be looking at the ticker.
The Hyperscaler Addiction
Microsoft, Meta, Alphabet, and Amazon. These are the "Hyperscalers." They are effectively locked in an arms race. If Mark Zuckerberg stops buying H100s or B200s, and Sundar Pichai at Google keeps buying them, Meta loses. It’s a classic Prisoner’s Dilemma.
- Meta’s Llama 4 and 5 models require exponentially more compute.
- Microsoft’s Copilot integration is an insatiable vacuum for GPU cycles.
- Sovereign AI: Now, entire countries (Saudi Arabia, UAE, Japan) are building their own sovereign data centers.
Is this sustainable? Maybe not forever. But right now, the demand still outstrips supply. Even as supply chains have eased compared to the nightmare of 2023, the lead times for the top-tier chips remain significant.
Is Nvidia Still a Buy? Let’s Talk About the "Air Pocket"
There is a legitimate fear in the market called the "air pocket." This is the period where big tech companies have spent hundreds of billions on chips but haven't yet shown the massive revenue gains to justify that spend to their own shareholders.
If Wall Street decides that AI ROI (Return on Investment) is a myth, the spending stops.
📖 Related: Ross Rennie Canada Prudential: What You Actually Need to Know
If the spending stops, Nvidia’s revenue doesn't just slow down—it craters.
We haven't seen that yet. In fact, companies like ServiceNow and Adobe are starting to show real, tangible bottom-line growth from AI features. But it’s the primary risk factor. You have to ask yourself if you believe generative AI is a fundamental shift in how work gets done or if it's just a better version of "Clippy" from the old Microsoft Word days.
The Competition is Coming (Slowly)
AMD is the closest rival. Their MI300 and MI325X series chips are impressive. They are cheaper. They have more memory in some configurations. For certain "inference" tasks—that's when an AI model is actually running and answering your questions—AMD is a very viable alternative.
Then you have the "in-house" chips.
- Google has the TPU.
- Amazon has Trainium and Inferentia.
- Apple has its own silicon.
They want to stop paying the "Nvidia Tax." But designing a chip is one thing; building a software library that everyone knows how to use is another. Nvidia has a 15-year head start. That is a lifetime in technology.
Understanding the "Price to Perfection"
When asking is nvidia still a buy, you have to look at the PEG ratio (Price/Earnings-to-Growth). Surprisingly, for a long time, Nvidia’s PEG ratio was actually lower than many "boring" consumer staples companies because their earnings were growing faster than their stock price.
That’s a freak occurrence in finance.
But eventually, the law of large numbers catches up. You cannot grow at 200% year-over-year forever. The math literally stops working. At some point, Nvidia becomes a cyclical company again. It’s happened before. In 2018, the "crypto winter" crushed Nvidia’s stock because miners stopped buying GPUs. In 2022, the gaming slump did the same.
This time is different because the buyer isn't a teenager in a basement or a crypto bro in a warehouse; it’s the Fortune 500. But "different" doesn't mean "invincible."
The Geopolitical Wildcard
We can't talk about Nvidia without talking about Taiwan. TSMC (Taiwan Semiconductor Manufacturing Company) makes almost every high-end Nvidia chip. If anything happens in the Taiwan Strait—a blockade, an invasion, even a major earthquake—Nvidia’s stock price is the least of our worries, but it would effectively go to zero overnight because they have no "Plan B" for manufacturing.
The U.S. government is also constantly tweaking export rules. They don't want the most powerful chips going to China. Every time a new restriction is announced, Nvidia loses a chunk of its total addressable market. They’ve been clever at creating "slowed down" versions of chips for the Chinese market, but the cat-and-mouse game with the Department of Commerce is a constant drag on the bull case.
Practical Steps for Potential Investors
If you’re staring at the "buy" button, don't just market-order your life savings. That's a recipe for a heart attack.
- Dollar Cost Averaging (DCA) is your friend. Instead of buying all at once, break your investment into four or five chunks. Buy one chunk now, and set an alert to buy another if the stock drops 10%.
- Check your exposure. If you own an S&P 500 index fund, you already own a massive amount of Nvidia. It’s one of the largest weights in the index. You might be "over-concentrated" without even realizing it.
- Watch the "Inference" shift. The big money in the next two years isn't in training models (which is what Nvidia dominates); it’s in running them. If Nvidia loses the inference market to cheaper chips, the party is over.
- Look at the "Nvidia Ecosystem." Sometimes the better buy isn't the chipmaker itself, but the companies building the power substations, the liquid cooling systems (like Vertiv), or the fiber optics that make these data centers possible.
Nvidia is no longer a "hidden gem." It is the sun around which the entire tech sector orbits. Buying it now is a bet that we are still in the early innings of the AI era. If you think we're at the "Peak of Inflated Expectations" on the Gartner Hype Cycle, you should wait for the inevitable "Trough of Disillusionment." But if you think the world's computing infrastructure is being permanently replaced, then the current price might actually look cheap in five years.
It’s a volatile, high-stakes, and incredibly complex business. It's definitely not for the faint of heart. Keep your position sizes sane and your eyes on the data center revenue. That is the only number that truly matters.
Actionable Insight: Before buying, review the most recent quarterly earnings transcript—not just the headlines. Look specifically for "Data Center Revenue" growth vs. "Inventory Levels." If inventory is rising faster than sales, it's a major red flag that the cycle is peaking. If you decide to enter, use limit orders rather than market orders to avoid getting hosed by the stock's extreme intraday volatility.