The AI chip leaderboard is shifting fast in May 2026. Intel is up roughly 240% year to date, and AMD is breaking out on MI400 momentum.
Meanwhile, Nvidia still owns the data center, but its share of new accelerator deployments is slipping. The question on every trader's screen: should you trim NVDA and add INTC plus AMD to your portfolio now?
This is the AI chip rotation thesis. It is real, but it deserves a clear-eyed playbook before you reposition.
The Catalyst: Intel Up 14% on Apple Deal, AMD MI400 Momentum
Intel ripped 14% on May 8 after reports that Apple is finalizing a chip foundry deal. According to CNBC, the agreement could deliver 25% of Apple's chip orders to Intel by 2030.
Bank of America estimates that workload alone could generate $10 billion in annual foundry revenue. For a company that lost money in its foundry segment for years, that is a structural validation.
AMD is running its own playbook. The Instinct MI400 launches in the second half of 2026 with 432GB of HBM4 memory and 40 petaflops of FP4 compute.
S&P Global projects MI400 will generate $7.2 billion in its first year. AMD's data center GPU revenue is forecast to grow 114% year over year to $15 billion in 2026.
Has Nvidia's Moat Narrowed in 2026
Nvidia's accelerator market share has slipped from above 90% in 2024 to roughly 68% in early 2026. That is the bear case in one line.
The bull case is more nuanced. Nvidia's absolute revenue is still climbing because the overall AI accelerator market is expanding faster than competitor share gains.
Total Nvidia AI accelerator revenue could still exceed $150 billion in 2026, keeping Nvidia at 75% to 85% share of total spend. Share loss in percentage terms does not equal revenue contraction.
The real risk is not market share. It is margin compression as hyperscalers diversify suppliers and pressure pricing on next-generation GPUs.
Comparing Forward PE: NVDA, INTC, AMD, AVGO
Valuation is where this rotation gets uncomfortable. The four major AI chip names trade at very different multiples right now.
| Ticker | Forward PE | Catalyst | Valuation Read |
|---|---|---|---|
| NVDA | ~26x | Blackwell Ultra ramp, Rubin launch | Most reasonable of the four |
| INTC | ~120x | Apple foundry deal, SK Hynix tie-up | Priced for flawless execution |
| AMD | ~45x | MI400 launch, hyperscaler wins | Premium but defensible |
| AVGO | ~40x | Custom ASIC for Google, Meta | The quiet AI play |
Nvidia is now the cheapest of the four on forward earnings. That is a remarkable inversion of 2024 sentiment.
Intel's 120x forward PE assumes the Apple deal closes on time and at scale. Any slip in qualification timelines could compress that multiple fast.
Trader Playbook: Pair Trade or Rotate Capital
Two structures fit this rotation. The first is a long INTC plus long AMD versus short or trimmed NVDA pair trade.
This expresses the view that share losses at Nvidia translate into share gains at the challengers. It hedges out broader semiconductor beta.
The second structure is a partial capital rotation. Trim 20% to 30% of your NVDA position and redeploy into INTC and AMD on a barbell.
This keeps Nvidia as a core holding while adding asymmetric exposure to the rotation thesis. It is the more conservative path for long-term portfolios.
If you prefer one stock instead of three, AVGO is the dark horse. Custom ASIC revenue from Google and Meta is growing without the headline-cycle volatility of INTC or AMD.
For deeper context on the broader setup, see our analysis on Intel turnaround dynamics and AI stocks beyond Nvidia.
What Could Re-Accelerate Nvidia: Blackwell Ultra, Rubin Roadmap
The rotation thesis assumes Nvidia stays static while challengers catch up. That ignores the Rubin roadmap.
According to Tom's Hardware, Rubin enters full production this year and ships to partners in the second half of 2026. Rubin Ultra follows in Q2 2027.
Rubin offers 3.3x the FP4 compute of Blackwell Ultra. It pairs HBM4 memory with NVLink 6 and delivers 1.2 ExaFLOPS of FP8 training per rack.
If Rubin hits its targets, Nvidia's per-rack training economics widen the gap against MI400 again. The rotation window could close fast.
Watch for two signals. First, hyperscaler capex commentary on the Q2 calls. Second, MI400 shipment quality reports from cloud customers in October.
Conclusion
The AI chip rotation is a real opportunity, but it is not a one-way trade. Nvidia's moat has narrowed, not collapsed.
The smart move is a partial rotation, not a full exit from NVDA. Trim, add INTC and AMD on weakness, and watch Rubin execution closely.
Review your AI chip holdings today on Gotrade. Open the app, check your NVDA weight, and add INTC or AMD to your watchlist before the next catalyst hits.
FAQ
Q: Should I sell all my Nvidia stock to fund Intel and AMD?
A: No. A partial trim of 20% to 30% is more defensible than a full exit, since Nvidia still trades at the lowest forward PE of the four major AI chip names.
Q: Is Intel's 14% jump sustainable?
A: It depends on Apple foundry deal execution. Qualification and production ramp can take two to three years, so near-term volatility is likely.
Q: When does AMD's MI400 actually ship?
A: Second half of 2026, with production volumes ramping into 2027. Watch Q3 earnings for early shipment quality data.
Q: Why include AVGO in this conversation?
A: Broadcom's custom ASIC business with Google and Meta is a quieter AI growth story. It avoids the headline volatility of pure GPU names.
Q: What could derail the rotation thesis?
A: A strong Rubin launch from Nvidia in late 2026. If Rubin meets the 3.3x compute claim, share losses at Nvidia could stabilize quickly.
