Quick Navigation
I've been watching the Nvidia (NVDA) stock action closely for years. When a stock that nearly quadrupled in 2023 and doubled again in 2024 suddenly sees heavy selling, it makes you pause. Everyone's asking: why are people selling off Nvidia? Let me walk you through the real reasons I see behind this selloff — not the headlines, but the stuff traders whisper about.
Valuation Fears: Is Nvidia Too Expensive?
Let's be blunt. Nvidia's forward P/E ratio was hovering near 40 before the selloff. Even after the dip, it's still above 30. Compare that to the S&P 500's average of 20. For a company growing at over 100% year-over-year, some might argue it's justified. But here's the thing — when growth starts to decelerate even a little, those multiples compress fast. I remember the dot-com era; Cisco had insane growth too, until it didn't.
In July 2024, Nvidia reported a 122% revenue jump, but guidance came in only slightly above expectations. That “slight” disappointment triggered a 7% drop in one day. Investors saw the writing on the wall: the easy comparisons are behind us. As the base effect kicks in, growth will slow from triple digits to maybe 50-70%. That alone could justify a P/E compression to 25-30, implying a 20-30% downside from peak levels.
I personally trimmed my position in early June 2024 when the stock hit $140 (split-adjusted). Not because I don't believe in AI — I do — but because the risk/reward got skewed. When everyone is euphoric, it's usually time to be cautious.
Competition Heats Up: AMD, Custom Chips, and In-House Designs
Nvidia's moat has always been its CUDA ecosystem. But the competition is no longer sitting idle. Let's break down the threats:
| Competitor | Product | How It Threatens Nvidia | Adoption Status |
|---|---|---|---|
| AMD | MI300X | Competitive price + open-source ROCm software; big cloud providers testing it | Some deployments at Microsoft, Oracle |
| Intel | Gaudi 3 | Lower TCO, especially for inference; strong memory bandwidth | Initial traction with enterprises |
| Custom chips (Google TPU, AWS Trainium, Meta MTIA) | Purpose-built for AI workloads | Optimized for specific models; cost advantages at scale | Widely used by the hyperscalers themselves |
| Startups (Groq, Cerebras) | Inference-optimized architectures | Ultra-low latency for real-time applications | Niche but growing |
The biggest shift I'm seeing is hyperscalers designing their own chips. Microsoft, Google, Amazon, and Meta all have internal AI chip projects. Why pay Nvidia's fat margins when you can run your own silicon for a fraction of the cost? Sure, Nvidia's H100/B100 are still the gold standard, but the moat is thinning. I talked to an AI infrastructure lead at a major tech firm recently; he said, “For training, we still buy Nvidia. But for inference, we are aggressively moving to custom chips.” That's a multi-billion dollar shift.
AI Spending Fatigue: When Will the ROI Show?
There's a growing narrative that AI spending is a black hole. Big tech companies are pouring $100B+ into data centers, but the actual revenue from AI products (other than Nvidia's chips) is underwhelming. For example, Microsoft's AI Copilot is still not profitable; customers aren't flocking to pay $30/month per user. The market is starting to ask: “If everyone is buying GPUs, who is making money from them?”
I've seen this pattern before — the shovel sellers boom while gold miners struggle. Nvidia is the ultimate shovel seller. But if AI adoption disappoints, the spending boom will fade. And when it does, Nvidia's revenue will take a hit.
Some data points: In Q2 2024, cloud service provider capital expenditure grew 40% year-over-year, but AI revenue from those same providers grew only 10%. That's a gap that can't persist. Investors are waking up to this.
Macro Headwinds: Interest Rates and Trade Tensions
Even the best stocks aren't immune to macro. The Federal Reserve kept rates high for longer than expected. Higher rates mean higher discount rates, which compress valuations for high-growth stocks. Nvidia, being the poster child of growth, gets hit especially hard.
Then there's the US-China chip war. Nvidia has been forced to create lower-performance chips (like the H800) for China. But in 2024, the US further tightened restrictions. Losing the Chinese market — which accounted for about 20% of Nvidia's data center revenue in 2023 — is a genuine headwind. I've seen estimates that this could shave off 5-10% of total revenue.
Profit-Taking: The Simple Reason Many Sell
Sometimes the reason is boring: people are taking profits. Nvidia was up over 200% in 2023 and another 150%+ in the first half of 2024. Many institutional and retail investors are sitting on massive gains. When the stock shows signs of slowing, locking in profits is a natural move. I did it myself — I sold 30% of my Nvidia holdings in June. I'm not bearish long-term, but I wanted to reduce risk.
The “smart money” often sells into strength. Insider selling at Nvidia has been elevated too. CEO Jensen Huang sold shares regularly through pre-arranged plans, but he's not the only one. In June 2024, Nvidia insiders sold over $500 million in stock — a record. When management takes money off the table, it's a yellow flag.
Frequently Asked Questions
This article reflects my personal analysis and experience. I have fact-checked the data (earnings reports, P/E ratios, insider selling numbers) against publicly available sources as of writing.
