Several weeks ago (March 17th), Nvidia CEO Jensen Huang gave the keynote for the companies GTC (GPU Technology Conference) event in San Jose, California. At the press Q&A event the next day, Huang stated that “his company would see upward of $1 trillion in data center revenue through 2027.” That is WOULD, not COULD.
He went even further by adding “that the $1 trillion figure only refers to sales of Nvidia’s Grace Blackwell and Vera Rubin chips where the final result will likely be higher.” Huang also reiterated his remarks at an earlier GTC in Washington D.C. that the “company had a through-line to $500 billion in revenue by the end of 2026.” Remember, he is the CEO of a publicly traded company (thus what he says is material information which amounts of some level of guidance as there will be those who may make investing/trading decisions based on what was said). $1 trillion is a huge amount. $1 trillion is a billion x 1000.
Let’s put this into better context. Nvidia’s TOTAL revenue for fiscal year 2025, was $130 billion (a record breaking amount for them and nothing to sneeze at) while the above article notes they had $216 billion for their fiscal 2026 (they operate on an advanced calendar year). This was also built on the massive AI spending bubble that powered a lot of this business where many companies were just dumping unquestioned money into it throughout 2024-2025. We’re seeing signs this bubble is about to burst (OpenAI pulling the plug on Sora; refer to what it was potentially costing them per day where some estimated that it was at least $1 million/day).
The few months I was doing intensive video upscaling/enhancing (using local model apps like Topaz and Aiarty Video AI that kept my RTX processing constantly, increased my electric bill by $150 for those months). Translation: this stuff requires a lot of electricity (and at data center levels, the additional cooling). There isn’t much “economies of scale” even if using renewal energy (because there are initial capital costs associated with that which will take years to pay off). And just like my use is mainly for nothing mission critical, theres a lot of “AI slop” burning up those processing cycles at a greater scale.
Simpler translation: it costs money and businesses are going to pass that on to customers. How many (b2b and b2c) are paying now for premium AI services? Sora being shutdown is just the tip of the iceberg when it comes to that answer. But I digress.
Huang is saying just this one segment of their chip business, will do $1 trillion from between his March 2026 prediction through 2027. Who is going to pay for that? Is he intending for his company to charge massive premiums to get to that $1 trillion? He seems/is oblivious to the fact that many of these other companies operate in the same “infinite growth at all costs” mentality that Nvidia does where there is this point on a graph (which any competent CFO will look at), where such spending does not compute in terms of their reality of margins, year-over-year growth, P/E ratios, etc.
I know Huang isn’t stupid (he started off as an electrical engineer). But engineers don’t necessarily make the best executives to run the actual business side or even the marketing/PR side by becoming this megaphone that is now making these material statements as the CEO. Nvidia has been a massive recipient of this AI bubble (which the financial media and Wall Street are tepid in calling a bubble because they make money from these cycles).
Maybe he believes he has the same sort of reality distortion field that Steve Jobs had (not seeing it but maybe I am just blind/jaded) and is going to use that mind trick to will the industry into making those massive expenditures that have been built on excessive hype (greater than that of the dot com bubble era which is that now tiny peak around that 1999-2000 period; back then, it wasn’t tiny in comparison to the all time highs it was hitting on the NASDAQ).
That previous bubble is what will happen with the one that has formed on the far right of this chart (and this bubble isn’t just AI, it also includes the crypto bubble). History has shown that it’s not sustainable. Once a CEO of a major company starts making prognostications like this (where part of it feels like a “pump and dump”), leaves me wondering how much of that air is going to come out. Time as always, will determine how potentially right or wrong I am on this.
