The plausible way Nvidia can get there, and how that path could be derailed.
The AI revolution has catapulted Nvidia (NVDA 4.14%) stock to all-time highs, with its market cap reaching $3.38 trillion as of this writing, now second only to Apple today.
But Wall Street doesn’t see Nvidia stopping here, with some analysts seeing Nvidia surpassing Apple, and hitting a $5 trillion valuation by 2026. Here are the reasons for Wall Street’s optimism, and what could also disrupt the bullish thesis.
Hopper, Blackwell, Rubin, and the software to tie it together
Nvidia’s rise is sure to garner a lot of skeptics. But the bulls have a compelling case that one, the AI revolution is real, and will only grow bigger in the years, and two, that Nvidia will continue to dominate the space, generating not only growth but maintaining its super-high margins.
There have been several bullish predictions on both fronts in recent weeks. At its own recent AI event, Advanced Micro Devices (AMD 1.24%) CEO Lisa Su predicted the AI data center chip market will reach $500 billion by 2028, up from just $45 billion in 2023. And recent Wall Street banks discovered that Nvidia’s new Blackwell chip is now sold out for the next 12 months.
Two Wall Street sell side analysts also just weighed in with very high price targets. Back in June, boutique firm Rosenblatt increased its price target on shares from $140 to $200. That $200 price target is the highest on the street, and just a hair shy of a $5 trillion market cap. Of note, Wall Street analysts usually give price targets based on what they believe the stock will do in the next year.
Rosenblatt’s analysts upped their target after being encouraged by Nvidia’s new Blackwell chip, which is just being shipped now. And the analysts see the demand continuing with the launch of Rubin, which will probably come out at the end of 2025 or beginning of 2026. Of note, Nvidia announced last year it would be increasing the pace of new chip architectures every year, versus a prior two-year cadence.
But Nvidia can’t just hit $5 trillion by growing earnings; it also has to maintain a high multiple to get there. That’s where Rosenblatt sees Nvidia’s non-chip offerings helping. Remember, Nvidia isn’t just a chip provider, but increasingly a full system solutions provider, providing networking infrastructure and full data center server system reference designs. It also has newer software offerings that help developers use and improve its hardware to specific AI outcomes.
Software earnings, with their recurring, subscription-like nature, generally garner a higher valuation that chip earnings, which can be more cyclical. As the AI industry matures, Rosenblatt sees Nvidia’s mix of software revenue increasing, keeping its P/E ratio high.
Then just last week, Bank of America raised its Nvidia target to $190 from $165. Analyst Vivek Arya boosted his target based on his view that free cash flow margins and growth could sustain not just this year but next year, bringing in over $200 billion over that time. Arya also notes new partnerships with other major enterprise companies such as Accenture can help Nvidia sustain its dominance in the enterprise.
Image source: Getty Images.
What could derail the path to $5 trillion
The big risks for Nvidia investors are if the two factors mentioned above, AI’s benefits and continued investment, as well as its competitive moat, don’t hold up.
As for the sustainability of the AI revolution, there are virtually no technology executives saying it will slow down. That may only change if the big cloud companies see a lack of returns on their AI spend. But even if a return is delayed, these companies are still likely to continue investing for fear of being left behind. So, it looks as though it would take a serious downturn in their revenues and earnings, likely a recession, for that to happen. But that doesn’t look like it’s happening with the U.S. economy remaining strong and the low unemployment situation what it is today.
More likely is the penetrating of Nvidia’s moat and the rise of alternatives to its high-priced chips. To preserve its moat, Nvidia believes its CUDA software, built over the past 15 years, gives it an entrenched advantage.
However, there are many big tech companies now getting behind open-source alternatives like Pytorch and Tensorflow, and working on abstracting away the CUDA software kernel so that its programing can be used for other non-Nvidia chips. If the CUDA moat breaks down, then the industry becomes more competitive.
With AMD’s MI300 line of GPUs becoming more available and virtually all large cloud companies producing more of their own in-house custom accelerators, alternatives to Nvidia are growing, which could lead to pricing pressure. Remember, it’s not normal for a chip company to have gross margins in the mid-70% range as Nvidia does today. And it’s especially abnormal when those chips are the most expensive in the industry and being bought in biggest numbers.
NVDA Gross Profit Margin data by YCharts
Another risk is that as the industry moves from training at the big cloud players to inference across enterprises, that lower-cost inferencing chips made by Nvidia’s competitors may win out. Once a model is trained, custom ASICs and even some CPUs can be utilized for inferencing, especially smaller or medium-sized models. So enterprises will probably tend to go for the best cost-performance for their needs, which may or may not be Nvidia.
As many think the inferencing market will dwarf the training market, it’s possible alternatives may give Nvidia a run for its money, literally, as business and consumer inferencing applications emerge today and in the years ahead. Given that Nvidia’s current valuation implies continued growth and sustained high profits, the industry’s pivot to inferencing is another thing to monitor.
The AI market is dynamic and evolving
It’s unclear today whether Nvidia will capitalize positively on its current lead, or whether competitors will emerge as the AI revolution evolves to its next inning.
The market appears to be bifurcating, to super-high-end training for the race to artificial general intelligence on one hand, where Nvidia is likely to continue its lead, and then to pragmatic everyday inferencing at lower costs on the other. Therefore, investors need to keep abreast of current developments in AI, and be unafraid to adjust one’s view as new data points come to light.