Trading Cheat Sheet

• SOX Performance: The Philadelphia Semiconductor Index rose 87.8% in Q2 alone, its best quarter on record, and has more than doubled in 2026 YTD.
• AMD (YTD ~150%): Forward PE around 65-70x. Q1 revenue $10.25B, up 38% YoY. Q2 guidance ~$11.2B. Data Center at $5.8B, up 57% YoY. Earnings date: Aug 4, 2026.
• Nvidia (YTD ~14%): Forward PE around 22-23x. Q1 FY2027 revenue growth 85% YoY. Q2 guidance $91B. Earnings date: Aug 26, 2026. More conservatively priced than AMD.
• TSMC (YTD ~150% over 12 months): Q1 revenue $35.9B, up 40.6% YoY. Q2 guided to $39-40.2B. Capex toward high end of $52-56B. Full-year revenue growth above 30%. Foundry for every major AI chip designer.
• Broadcom: Q2 AI revenue $10.8B (+143% YoY). Q3 AI guidance $16B missed the $17.2B estimate. Shares fell 12-13% on guidance. Did not raise full-year AI forecast of $56B. Earnings: next report late August/September.
• Hyperscaler Capex: Combined 2026 spend from the major cloud players has topped $700B, up roughly 67% YoY. Approximately 75% tied directly to AI infrastructure.
• Key Risk: Guidance delta, not absolute results. In this market, beat-and-raise is the only outcome that extends the trade. Anything less risks a cascade, not just a dip.
• Key Watch: Amazon, Microsoft, Google, and Meta earnings calls in late July. Capex commentary will move chip stocks more than the headline numbers.

Here’s where we actually are.

The Philadelphia Semiconductor Index just recorded its best quarter in history. We’re talking 87.8% in a single quarter, the biggest gain in records stretching back to 1994, topping even the dot-com era. The SOX has more than doubled in 2026, and we’re only halfway through the year. AMD rose more than 185% in Q2 alone. Intel was up 216%. Micron jumped over 240% in the quarter. The chip trade didn’t just run — it went parabolic.

And then June happened.

Volatility spiked hard mid-month, with the sector at the center of a sharp selloff that erased significant gains in days. The June turbulence was real. Broadcom’s Q3 AI chip guidance of $16 billion fell short of the $17.2 billion analyst estimate — and the company did not raise its full-year AI revenue forecast of $56 billion. That single guidance miss, combined with a slight software segment miss, sent Broadcom shares down more than 12-13%, wiping out over $270 billion in market value and dragging the broader sector with it. That’s not noise. That’s the market telling you something about how sensitive these valuations are to even minor guidance shortfalls.

What the market is telling you is this: the AI semiconductor thesis is structurally intact, but the stocks are priced for perfection heading into a Q3 earnings season that starts mid-July. The gap between the business and the price is narrower than it was six months ago. That changes the positioning calculus entirely.

The Numbers That Actually Matter

The semiconductor industry entered 2026 in the midst of what industry observers describe as the most consequential investment cycle in the sector’s history. Unlike the boom-bust cycles of previous decades driven by PC and smartphone refresh cycles, the current upcycle is structural. Artificial intelligence has become the dominant demand engine, pulling the entire semiconductor stack — logic, memory, advanced packaging, and networking — upward at the same time.

The market size confirms it. Deloitte projects global semiconductor sales to reach $975 billion in 2026, with AI chips alone approaching $500 billion — roughly half of total industry revenue for the first time. Bank of America has an even more aggressive view, projecting the global chip market hits $1.3 trillion in 2026 with potential to double to $2 trillion by 2030.

AMD’s most recent quarterly results show the demand is real. Q1 2026 revenue came in at $10.25 billion, up 38% year-over-year, with non-GAAP EPS of $1.37 beating estimates. The Data Center segment carried the quarter at $5.8 billion, up 57% year-over-year, and free cash flow hit a quarterly record. Management guided Q2 to roughly $11.2 billion in revenue, about 47% year-over-year growth. That’s not a broken story. That’s a strong story at a demanding valuation.

And that’s exactly the tension.

The Valuation Problem Nobody Says Out Loud

AMD has surged roughly 150% year-to-date and now trades at approximately 65-70 times forward earnings. Nvidia, by contrast, trades around 22-23 times forward earnings despite posting record revenue with 85% year-over-year growth in its most recent quarter. The market is pricing Nvidia more conservatively than AMD, which implies the consensus is building more execution risk into the challenger than the incumbent. That’s a useful signal.

Slight tangent, but it matters: Nvidia’s forward multiple has actually compressed significantly from its historical averages. Its 5-year average PE is north of 70x. At 22-23x forward, the market is pricing in a lot of continued earnings growth without giving much credit for upside surprises. That’s a different kind of valuation risk than AMD’s — less about being overpriced, more about whether the earnings growth keeps coming.

The spending backdrop is real. Combined 2026 capex from the major hyperscalers has now topped $700 billion, according to multiple analyst estimates, up roughly 67% year-over-year and representing the third consecutive year of 60%+ growth. Approximately 75% of that spend is tied directly to AI infrastructure. The contracts are real. The question is whether it’s already in the price.

What Q3 Earnings Season Actually Requires

Investors are now looking ahead to the second-quarter earnings season, which begins mid-July. For semiconductors specifically, the bar is not just beating — it’s beating and raising. The S&P 500 is expected to report year-over-year earnings growth of roughly 23% for Q2 2026, which would mark the second consecutive quarter of earnings growth above 20% and the seventh consecutive quarter of double-digit earnings growth for the index.

The bar is high. The market raised it. That’s the problem.

The Broadcom situation illustrated this perfectly. Record Q2 revenue of $22.19 billion, up 48% year-over-year. AI semiconductor revenue of $10.8 billion, up 143%. And the stock still fell 12-13% because the Q3 AI revenue guidance of $16 billion came in below the $17.2 billion estimate, and the full-year AI forecast wasn’t raised. When the math lands at $56 billion rather than higher, traders read it as a ceiling rather than a floor. In this market, that’s the only thing that extends the move — beat and raise. Anything else, and the risk is a cascade, not just a dip.

Three Scenarios for the Second Half

Bull Case: TSMC, Nvidia, AMD, and Broadcom all beat Q2 consensus and raise full-year guidance. Hyperscaler capex commentary on earnings calls confirms the $700B+ spending cycle is accelerating into 2027. AMD approaches its 52-week high, and the SOX index consolidates near all-time highs into year-end.

Base Case: Results are solid but guidance is cautious. Companies acknowledge the Broadcom dynamic. Stocks consolidate in a 10-15% range around current levels through Q3. The passive bid provides a floor — passive investing continues to absorb capital at record rates and that bid doesn’t disappear on a guidance miss. The move doesn’t break, but it doesn’t sprint either.

Bear Case: One or two major hyperscalers trim capex guidance on Q2 earnings calls, citing digestion of prior investments. Memory oversupply concerns resurface. The SOX retests the June lows. AMD gives back another 15-20% from current levels. Given the scale of the Q2 rally — 87.8% in a single quarter — a mean reversion of that magnitude would still leave the sector up massively on the year. But it would hurt if you’re sized wrong.

Where the Interesting Risk-Reward Lives

TSMC is the most interesting risk-adjusted position in this environment. Q1 2026 revenue came in at $35.9 billion, up 40.6% year-over-year, beating consensus. Q2 is guided to $39-40.2 billion. The company raised its full-year revenue growth guidance to above 30% and is spending toward the high end of a $52-56 billion capex range for 2026, communicating confidence in the multi-year demand outlook. As the foundry for every major AI chip designer — Nvidia, AMD, Google, Apple, Broadcom — TSMC captures the AI buildout regardless of which chip designer wins the arms race. That’s a structurally different kind of semiconductor exposure.

The options market is worth watching closely heading into mid-July. Retail options activity has entered a new regime, with traders absorbing record levels of premium in recent weeks. That level of options concentration means the gamma dynamics around earnings dates will be extreme. Volatility sellers will be tempted. The risk is that in a sector this crowded, a guidance miss creates a cascade — not just a dip.

The discipline that matters most right now is position sizing. The AI semiconductor thesis is one of the strongest in a generation. The valuations are among the most demanding in the market. Both things are true at the same time. The traders who navigate the second half best will be the ones who hold both realities simultaneously — without letting either one blind them to the other.

For informational and educational purposes only. Not investment advice. Trading involves risk, including loss of principal.