Top Call
NVDA — buy the post-earnings dip. Q1 FY2027: $81.6B revenue (+85% YoY), data-center at $75.2B (92% of sales), EPS $1.87 vs $1.76 est., $80B buyback, Q2 guided to $91B. The stock slipped ~1% post-print. The market wanted a miracle and got a record — that's sentiment, not a broken thesis.
Overnight / Long-Weekend Recap
- QCOM: +11.6% Friday on confirmed data-center processor shipments to a major hyperscaler by end-2026. Upgrades from Daiwa, Tigress ($280 PT), TD Cowen. The "QCOM is just mobile" narrative is cracking.
- AMD: Q1 revenue $10.25B, data-center +57% YoY, 20+ PT upgrades. Cleaner risk/reward than NVDA at current levels — less priced to perfection.
- MU: Q2 FY2026 revenue $23.9B (vs $8.05B a year ago), Q3 guided to $33.5B; new HBM facility at Tongluo, Taiwan. HBM is the scarcest input in AI training.
- TSMC: global semis forecast raised to $1.5T by 2030; the toll-road every NVDA/AMD/QCOM chip runs through.
- ASML: UBS PT to €1,900; the only EUV game in town. Long-cycle compounder.
Trade Ideas
| Idea | Entry | Target | Stop |
|---|---|---|---|
| LONG NVDA — buy the post-earnings dip | $130–138 | $175 (12M) | < $120 |
| LONG MU — HBM scarcity play | pullback to ~$120 | $150 | $105 |
| LONG QCOM — edge-AI re-rating | $220–230 | $280 | $205 |
Bottom Line
The AI semiconductor supercycle is real, but the easy money is behind us. SOX at +65% YTD demands discipline — own the quality (NVDA, MU, TSMC), use the post-earnings NVDA dip, respect the 1999 parallels without being paralysed by them. The next 12 months will separate the investors from the tourists.
Sources: company earnings releases and guidance (NVDA, AMD, MU, QCOM), broker notes, market data as of 26 May 2026, 06:00 ET. Disclaimer: educational purposes only; not investment advice; levels reflect views at time of writing.