AI After Nvidia: Who's Scaling, Who's Failing, and Who's the Dark Horse

Nvidia Corporation (NVDA) just put another giant quarter on the board, and the useful part was not the victory lap. Revenue hit about $96 billion. Data center alone did about $89 billion. Management pointed the next quarter near $108 billion and said customer forecasts would support even faster growth if supply could keep up. Jensen Huang’s line was blunt: AI is doing useful work, tokens are profitable, and compute is revenue. So NVDA delivered, and AI is delivering revenue. But now what. Who still earns a seat as the AI space only gets more and more crowded? My take is simple, there are three names to watch out for. One is scaling into the current generation of AI, one is failing the trust test, and one is a dark horse most people aren't paying enough attention to. ## Who’s scaling: Micron (MU) If Nvidia is the factory floor of the AI Boom, Micron Technology (MU) is the fuel that's keeping the machine running. High-bandwidth memory sits next to the big AI chips. Without enough of it, the fancy processors sit half-starved. Nvidia’s call kept circling supply pressure, and memory has been the loudest choke for months. Micron just reported a record fiscal third quarter with revenue around $41.5 billion, more than four times the year-ago level, and guided the following quarter near $50 billion. Gross margins crossed into the mid-80s. Management says the full 2026 high-bandwidth memory book is spoken for under multi-year deals, customers want more than Micron can ship, and buyers have already put roughly $22 billion of deposits and related support on the table to lock supply. I like this kind of American industrial story when the product is physical and the buyer has to prepay to get in line. The boom needs more memory than the industry can make, and Micron is one of the few shops that can deliver the high-end stack Nvidia and the big clouds actually use. What would change my mind is a real break in those take-or-pay contracts or a sudden flood of new capacity. Until then, Micron is the scaling name because scarcity is still doing the work. ## Who’s failing: Super Micro (SMCI) Super Micro Computer (SMCI) still builds the boxes a lot of AI customers want. Selling servers and passing the trust test are different jobs. Federal prosecutors unsealed a case tied to a co-founder and others over an alleged scheme that moved billions of dollars of advanced servers toward China through diversion paths. The stock took a brutal hit. Super Micro says it was not charged as a company, current senior management was unaware of the scheme, and an independent probe supports that view. It has fired people and promised tighter export controls. Taiwan later added fresh indictments of individuals tied to illegal AI server flows into China, again without charging the companies as entities. I am not here to play courtroom. I am here for the investment story. A tier 3 vendor in the middle of such heavily controlled exports cannot afford to play games in the race to win AI. It's China vs. America and if Trump gets a whiff that someone is helping the other side, it's hard to want to put my money anywhere near that. Growth that rides the gray edge of the rules can reverse into legal-file risk overnight. Super Micro can still ship product. The failing label is about trust, governance, and whether buyers and regulators will treat the brand as safe infrastructure. ## Who’s the dark horse: Nebius (NBIS) Most people still talk about AI like it's only ChatGPT, Claude, Nvidia and Google. Nvidia’s call kept pointing at the other half of the picture: AI clouds, enterprises, sovereign buyers, and the cloud layer that rents full Nvidia stacks to labs and companies that cannot wait in the hyperscaler line. That broader customer bucket is already huge and growing faster than pure hyperscale. Nvidia even flagged cloud partners heading toward multi-gigawatt GPU capacity as the year closes. Nebius Group (NBIS) is my dark horse pick in that lane. It is not the household name CoreWeave became, and that is partly the point. Nebius has been stacking real capacity contracts with giants like Microsoft and Meta, posting hyper-growth revenue as AI cloud took over the company, and drawing partnership heat from the Nvidia ecosystem. Management has talked about holding some capacity back rather than selling the whole future at yesterday’s price. Dark horse does not mean risk-free. These clouds burn cash to build power and clusters. Debt and execution can punish any of them. What I like is the seat they occupy: when Nvidia says demand is broadening past the five famous clouds, Nebius is built for that sentence. Plenty of investors are missing what's really being said. Nvidia’s own customer map says that corner is becoming a real second revenue engine. ## Bottom Line Nvidia just reminded everyone the AI factory is still expanding under supply stress. On my watchlist, Micron is still scaling on the memory choke, Super Micro is still failing the trust test even while it sells servers, and Nebius is the dark horse in the cloud lane Wall Street under-counts. *I am not a financial advisor, and nothing here is financial advice. This is one person's opinion about a story. Do your own homework and talk to a licensed professional before you put real money at risk.*