Nvidia's Central Role in the AI Trade
Nvidia (NVDA) is the most important company for the AI trade. It plays four roles at once: vendor, investor, banker, and guarantor of the AI buildout. It makes the chips, provides financing, helps build the data centers and infrastructure, and supports its customers directly. What matters now is whether this web of deals and transactions can keep generating growth over the next few years.
The earnings will be huge. Profit is set to double, and free cash flow will reach almost $200 billion. The numbers are astronomical. Yet investors act almost ungrateful and keep asking what comes next.
Stock Reaction and Ownership
The stock has fallen in the last four quarters after earnings, and in seven of the last eight it was down a month later. This is not institutional investors rotating out. It reflects the law of large numbers hitting a $5 trillion company. The growth of the last three or four years has been so extreme that investors almost do not believe it can continue.
Nvidia is also owned by nearly everyone. It is a top holding in most indexes. The chip index (SOX) is up almost 150% this year. Nvidia is everywhere.
The right view looks past the next two or three quarters to the long term. Jensen Huang has shown how powerful he is within the AI trade. Leading means spotting where the bottlenecks are and investing in the companies that will be future leaders. Nvidia's venture portfolio and outside investments back the companies building the next economy. The bet: Nvidia stays the leader over the next 5 to 10 years. Nvidia has held a position in the company since 2016 and is not selling.
Data Centers and Customer Mix
Data centers are the key focus. Over the last few years, most customers were the big four: Google (GOOGL), Amazon (AMZN), Microsoft (MSFT), and one more. SpaceX will be another hyperscaler spending heavily going forward.
Nvidia now wants to widen its customer base, pulling more revenue from industrials, neoclouds, and new companies. It started breaking down these numbers last quarter. Investors want to see that breakdown. If growth from neoclouds and what Nvidia calls "S ACI" runs much higher than growth from hyperscalers - a trend that should continue - then Nvidia sees rising growth.
Is the Demand Real?
Is AI demand real, or is it increasingly propped up by Nvidia-backed financing? The demand is real. People are starting to use AI in new ways. The data centers built by hyperscalers are only the first step in a massive buildout. The next step is all the other companies, neoclouds, and industries that will start using the compute already built. That shift is beginning to play out, and diversification will bring more demand for Nvidia over the coming years.
Margins and Memory Costs
Gross margins remain healthy, above 75%. Higher costs tied to memory and the Ruben chip ramp could pressure them. Memory costs have risen a lot and are hurting. Nvidia even redesigned some chips to use less memory. But the drop is minimal, from 75% to 73%, against hundreds of billions of dollars in revenue. This is not a small-margin business, so it will not hurt the bottom line much.
The bigger question is where memory costs go from here. The core problem: the world needs more fabs to build more memory. Supply is constrained by the ability to build it. Nvidia's comments on memory will matter beyond itself - they will affect Apple (AAPL) and SpaceX, and the whole AI complex.
