It’s hard to move the needle on shares of Nvidia (NASDAQ:NVDA) these days. Even though the stock has continued to drag its feet into the new year, with shares currently down just over 3% year to date, there have been several bullish developments that I thought have reignited a bit of enthusiasm for the GPU maker. AI demand isn’t just staying hot; it’s hot enough that supply can’t keep up.
Whether we’re talking about the high-bandwidth memory shortage, fabs running at full capacity, or other bottlenecks, the supply side still seems to be in a tough spot as it tries to keep up with demand.
Of course, it’s far easier to dismiss the situation as the inflating of some sort of AI bubble that will end in a painful burst that drags down all the high-flying AI plays from across the board, from Nvidia to other semiconductor firms that have been aggressively spending to ramp up.
Given lengthy lead times and the boom-and-bust nature of the semiconductors (especially the semiconductor equipment makers), it certainly seems like there’s a risk that supply overshoots, perhaps at some point over the next three years.
Either way, until the AI-native companies, the mega-cap tech titans, and other firms scrambling to keep up choose to scale at a slower pace, I do think the GPU trade has another couple of good years up its sleeves. In the meantime, though, investors are going to be more skeptical of new developments, partnerships, or commentary delivered by visionaries like Nvidia’s top boss Jensen Huang.
While shares of Nvidia may suggest it’s been a rather uneventful year so far, I’d argue that the $4.4 trillion titan has delivered far more exciting developments than just about any other tech company.
Vera Rubin is a big driver, but expectations are already quite heightened
With Vera Rubin going into full production several months earlier than expected, there are reasons to get more bullish on the firm as the significant efficiency gains pave the way for even more demand. The company seems to be well ahead of schedule, and the big performance gains are, indeed, real. If the firm can keep delivering positive surprises like this, I do think it’s just a matter of time before investors start reacting more positively.
Undoubtedly, much cheaper inference tokens from the Very Rubin era could lead to greater consumption, which, in turn, would increase the hunger for computing power. The Jevons paradox may very well play out with AI. As more compute comes online while firms make AI more useful and applied, perhaps those waiting for an AI bubble to go bust might be waiting for a few years longer. Add the potential efficiency gains of ASICs (application-specific integrated circuits) into the equation, and perhaps the cost of AI compute is about to take a nosedive as demand continues marching higher.
As Nvidia gears up to sell H200 chips to the Chinese market, with a 25% surcharge, while already focusing on what Huang sees as a year of physical AI, perhaps it’s time to view the recent sluggish start to 2026 as more of an opportunity to nibble on weakness than a sign that it’s time to take profits or heed the words of the great Michael Burry by placing a few bearish bets against the company as the AI trade looks to reverse course, perhaps in a devastating way.
The bottom line
Though the future looks bright for the next generation of AI-powered technologies (agents, applied AI, and robotics), there’s lingering doubt about the timeline and fear over how bad the bust will be when it happens. As always, time will tell. With Vera Rubin already in full production, I do think there’s reason to be just a bit more bullish, even though the 45.5 times trailing price-to-earnings (P/E) multiple leaves little room for error going into its coming quarters.