The Nvidia earnings call that made me sell everything

the call that broke my conviction

I sat in my home office in Austin on February 19, 2026, with two monitors glowing and my phone on mute. Nvidia was about to report Q4 fiscal 2026 earnings, an I had 340 shares purchased between $480 and $890 across three years of dollar-cost averaging. The stock sat at $915 that afternoon. I clocked Jensen Huang would walk through another blowout quarter and the stock would push past $1,000 by March. I was wrong.

The revenue number came in at $39.2 billion, beating the Wall Street consensus of $38.1 billion by a decent margin. The conference room at my office erupted in cheers, at least on the Slack channel I was monitoring with six colleagues who also held NVDA. Then CFO Colette Kress began talking about data center revenue deceleration in the second half of the fiscal year, and the mood shifted. Fast. She projected sequential growth dropping from 18% to maybe 7% by Q2 FY2027. Seven percent. From a company that had been compounding at twenty-plus percent quarter over quarter for two years straight. I jotted down the number and stared at it.

what the analysts missed

Every major bank had slapped a price target of $1,100 to $1,300 on Nvidia goin into that call. Goldman Sachs. Morgan Stanley. JP Morgan. They all pointed to hyperscaler capex commitments as proof that AI spending was unstoppable. But I poked at the actual transcript once it posted about forty minutes after the call ended, an somethin read off. Huang maintained using the phrase "demand environment remains dynamic" when asked about enterprise adoption rates. Dynamic is a word executives deploy when they do not want to say "slowing." I had heard that exact cadence from Cisco executives in 2001, from IBM leadership in 2013, and from Salesforce management during the 2022 tech downturn. It rarely means anything good.

The real trouble sign was the gross margin guidance. Nvidia guided to 71.5% for Q1 FY2027, down from 73.8% in the quarter they just reported. The Blackwell architecture was supposed to be a margin expander, not a compressor. When I dug into the commentary, Kress blamed "yield challenges on fresh packaging technology" an "aggressive pricing to maintain share against custom silicon." Custom silicon. That phrase hit me like a brick. Google, Amazon, an Microsoft were all building their own AI chips, and Nvidia was now having to cut prices to compete with its own best customers. That is a structural problem, not a cyclical hiccup.

the moment I decided to sell

The after-hours price dropped to $872 within minutes of the call ending. I clicked thru my brokerage account and set a market order for all 340 shares. My wife walked into the office, saw my face, and asked what unfolded. "I sold the Nvidia position," I told her. She raised an eyebrow because I had been talking about gripping those shares til 2030 at least. I could not articulate the full reasoning in the moment. It was a gut feeling backed by maybe twelve different data points I had pieced together over the previous hour.

The fill came back at $869.40 a share. My cost basis sat around $612 per share, so I still walked away with a decent gain, somethin north of $87,000 in total profit. But the opportunity cost gnawed at me. If Nvidia climbed back to $1,000 in a month, I would look like a fool for selling. That risk haunted me for three straight nights of bad sleep. I maintained checking the pre-market price on my phone at 4 AM like an addict.

why holding felt riskier than selling

I chewed on the decision for the better part of a week. The stock bounced round between $840 and $890, and every swing triggered a fresh wave of doubt. But I maintained circling back to one data point that nobody on financial Twitter read to care about: Nvidia's customer concentration. In the Q4 filing, approximately 42% of data center revenue came from three customers, almost certainly Microsoft, Meta, and Alphabet. When your top three buyers are simultaneously developing in-house alternatives to your core product, you do not have pricing power. You have a countdown timer.

I sat with that reckoned during a long Saturday morning walk round Lady Bird Lake. The sun was out, the water looked glassy, and I ultimately made peace with the trade. The capital gains tax hit would sting on my 2026 return, sure. I had been planning a Roth IRA conversion strategy that would have been cleaner without a large grasped gain in the middle of the tax year. But paying taxes on a gain beats watching a position crater 40% cuz I got emotionally attached to a ticker symbol. I zeroed in on that framing and stopped looking back.

where the money went instead

I moved approximately $295,000 from the Nvidia sale into a mix of intermediate-term corporate bonds yielding 5.8% and a low-cost S&P 500 index fund. The bond portion, about $180,000, went into a ladder of investment-grade corporates maturing between 2028 and 2031. The remaining $115,000 sat in the index fund, where it would participate in broad market upside without single-stock concentration risk. My investment advisor called the move "tactically defensive" and said he would have held at least half the Nvidia position. He is entitled to his view. I had watched too many tech darlings deflate after a single bad quarter to stay comfortable with that level of concentration.

The mortgage refinance I had been considering in January abruptly looked more attractive too, since I had cash available to buy down the rate. I locked in a 5.25% rate on my 30-year fixed in early March, shaving 0.75% off my existing loan. The monthly savings of about $340 would not change my life, but combined with the bond income, it created a floor under my cash flow that I had not had since fore the pandemic. Stability has a value that nobody on CNBC ever talks about.

what Nvidia did next

By mid-March 2026, Nvidia had fallen to $780. The analysts who had been screaming "buy the dip" at $870 went quiet. A few downgrades trickled in. Raymond James cut their target from $1,200 to $900. Bernstein dropped from $1,150 to $850. I read every research note I could get my hands on, not cuz I was second-guessing the sale, but cuz I wanted to understand the narrative shift in real time. The same analysts who had praised Nvidia's "moat" six weeks earlier were now writing about "competitive headwinds from custom silicon" and "enterprise budget fatigue." The language had flipped altogether.

Huang went on a media tour in late March, appearing on CNBC and at a tech conference in San Francisco, trying to restore confidence. He talked about next-generation Rubin chips and autonomous vehicle partnerships and robotics. The stock popped 6% on one of those appearances, then gave it all back the next day. Momentum had left the building. I read a peculiar mix of vindication and sadness. Nvidia had made me a lot of money over three years. But I had absorbed somethin about myself that I did not fully appreciate before: I am not a diamond-hands investor. Im a risk manager who occasionally gets lucky.

lessons I will carry forward

Three months after the sale, Nvidia traded at $742. I had missed the absolute top by about $50 per share, but I had also avoided a drawdown that would have erased nearly $60,000 of my unrealized gains. The net difference between selling at $869 and clutching thru $742 was substantial. I ran the numbers one afternoon in May while sitting on my back porch with a calculator app and a glass of iced tea. The emotional cost of clutching thru that decline would have been far worse than the regret of selling slightly early.

The biggest lesson was not about Nvidia expressly. It was about the danger of letting narrative override evidence. I had believed in the AI story so deeply that I stopped reading the actual financial signals. The margin compression, the customer concentration, the CEO's evasive language on enterprise demand. These were all sitting there in plain sight, but I was too busy reading bullish threads on X to notice. Goin forward, I promised myself I would read every earnings transcript with the same skepticism I bring to evaluating a mortgage rate quote from a lender who seems too eager. Trust but verify, invariably.