The S&P 500 just posted its strongest quarter since 2020 — roughly 14% — and the Nasdaq 100 surged approximately 20%, entirely on AI infrastructure enthusiasm. Q3 opens with elevated valuations, a hawkish Fed at 3.50–3.75%, and an earnings season where every miss will be punished. The rally was real. The test that follows is harder.
Q2 2026 is over. The scoreboard is unambiguous.
The S&P 500 posted approximately 14% in gains for the quarter — its strongest performance since Q2 2020, when markets were rebounding from pandemic-era lows. The Nasdaq 100 did better still: roughly 18–20%, driven by relentless buying of AI compute infrastructure names. For a quarter that began with genuine uncertainty around geopolitical conflict and Federal Reserve policy, the outcome was decisive.
The question that matters tonight, with Q3 now open and a major earnings season three weeks away, is whether the market has priced in too much of the good news — and not enough of what could go wrong.
The rally had a clear driver. It was not broad-based economic strength. It was AI.
Semiconductor manufacturers, cloud infrastructure providers, and the physical hardware supply chain that feeds AI training and inference clusters produced the quarter's outsized returns. According to analysis from investing.com and multiple equity research desks, AI-related spending is expected to account for approximately 60% of Q2 earnings per share growth, with aggregate EPS projected to rise roughly 22% year-over-year for S&P 500 constituents.
That is a concentrated driver. When one theme accounts for more than half of aggregate earnings growth, the index's performance becomes a referendum on that theme's sustainability, not a diversified signal about economic health.
The quarter also featured institutional buying pressure that reinforced index concentration. The Nasdaq-100 underwent its quarterly rebalancing effective June 22, 2026, and the additions tell the story clearly: CoreWeave (CRWV), Astera Labs (ALAB), Nebius Group (NBIS), Rocket Lab (RKLB), and Teradyne (TER) were added. CoreWeave — a GPU-as-a-service cloud provider — joined the index just 15 months after its March 2025 IPO, according to Nasdaq.com. That pace of index inclusion reflects the urgency with which passive funds are being forced to absorb AI infrastructure exposure.
Astera Labs, which builds high-speed connectivity semiconductors (PCIe, CXL, Ethernet) for AI data centers, saw heightened trading volume following the rebalance announcement, driven in part by passive fund mechanical buying pressure and end-of-quarter window dressing by institutional managers. This is how a momentum theme becomes embedded in index structure — and how it becomes harder to exit without market-wide consequences.
Index composition changes are mechanical events, but they carry interpretive weight.
Five companies removed from the Nasdaq-100 — Charter Communications, Cognizant, Insmed, Verisk Analytics, and Zscaler — represent a broad cross-section of legacy tech, healthcare IT, and cybersecurity. The five additions are unified by a single theme: proximity to AI compute infrastructure.
[OPINION] The index is not just tracking the market. It is directing capital flows that reinforce the very theme it reflects. Every passive fund benchmarked to the Nasdaq-100 was required to increase its position in CoreWeave and Astera Labs after June 22. This mechanical demand is separate from fundamental valuation — it is structural reinforcement of the AI trade regardless of price level.
This matters for H2 because the same mechanism works in reverse. If AI infrastructure sentiment deteriorates — following an earnings miss by a major hyperscaler, a guidance cut, or a demand signal that disappoints — the index's AI concentration amplifies the selloff through mechanical passive fund de-risking.
The equity rally occurred in spite of, not because of, Federal Reserve policy.
The Fed held the federal funds rate at 3.50–3.75% at its June 16–17 FOMC meeting, according to Schwab market data. More significantly, the Fed removed language signaling future rate cuts — a technical shift that markets read clearly: the easing cycle expected at the start of 2026 is not happening, and the next move may be up.
Chair Kevin Warsh's public posture has been consistently hawkish. Core PCE remains above target; labor market data has been resilient. Bank of America has updated its outlook to anticipate potential rate hikes in September, October, or December 2026 if inflation remains elevated and the labor market holds.
For equities, this creates a specific headwind: rising yields compress price-to-earnings multiples, and a stronger dollar makes international earnings less valuable in reporting currency. Neither effect is disabling at current levels, but both become material if the rate path shifts more aggressively.
The nonfarm payrolls report scheduled for Thursday, July 2 — moved one day earlier due to the Independence Day holiday — carries unusual weight in this context. A strong print (above consensus of approximately 113,000–115,000 jobs) reinforces the case for the Fed to hold or hike. A weaker print might briefly revive rate-cut hopes, but the structural shift in Fed rhetoric makes that scenario less credible.
The S&P 500's H2 setup is defined by a narrow path between two failure modes.
The first failure mode: earnings disappointment at elevated valuations. Analysts at multiple research desks note that companies missing Q2 estimates have seen average stock-price drops of approximately 4.2%, according to investing.com — meaningfully larger than the historical average. The market has a "razor-thin margin for error," as one analysis framed it, because valuations have risen faster than the earnings base can support if growth decelerates.
The critical test will arrive in mid-July, when hyperscalers — Microsoft, Alphabet, Meta, and Amazon — report Q2 results. Their AI capital expenditure guidance and commentary on monetization timelines will set the tone for the sector. [OPINION] If any major hyperscaler signals that AI-driven revenue growth is slower to materialize than the market has priced, the correction will be sharp and fast.
The second failure mode: rate shock. The market's current positioning does not fully price a rate hike. If the July 2 payrolls print and subsequent inflation data force the Fed's hand before September, equities will need to reprice. The Nasdaq's AI concentration makes it more sensitive to this adjustment than broader indices, because high-multiple growth stocks are the most vulnerable to discount rate increases.
There is a useful distinction to make here, and most commentary conflates it: the business of AI infrastructure is strong. The stock prices of AI infrastructure companies may or may not reflect that business correctly.
Microsoft's Three Mile Island nuclear deal, hyperscaler capex running at extraordinary scale, and the Nasdaq-100 adding CoreWeave within 15 months of its IPO — these are all real signals of genuine demand. AI training and inference infrastructure is being built because the demand exists and is growing. This is not a phantom.
What is genuinely uncertain is the timeline for that infrastructure spend to convert into earnings per share growth that justifies current multiples. Investors are beginning to ask the question that should have been asked earlier: at what point does the AI capex cycle produce durable margin expansion rather than just revenue growth?
[UNCERTAIN] The honest answer is that no one has a clean data point here yet. The companies spending the most on AI infrastructure are also the companies whose earnings reports in July will be scrutinized most intensely for evidence that the spend is working.
The following represents the author's analysis and should not be taken as financial or investment advice.
Q2 2026 was the AI trade winning on the scoreboard. The S&P 500's 14% gain and the Nasdaq's 20% surge were real, earned through genuine earnings growth, not pure multiple expansion. That matters as context.
[OPINION] But the setup going into Q3 is structurally more difficult than the Q2 entry point. The rally has already happened. The index has been recomposed to embed AI infrastructure exposure at the mechanical passive level. The Fed has removed the easing language that would otherwise provide a floor. And the earnings season that begins in three weeks must justify the valuations that this quarter's rally has established.
The risk/reward is asymmetric in a way it was not in April. A strong earnings season confirms what the market has priced and produces modest incremental gains. A disappointing earnings season — one where hyperscalers guide conservatively on AI monetization, or where the Fed's hawkish turn accelerates — produces a correction from an elevated base.
That asymmetry does not make equities a bad investment. It makes position sizing and risk management more important in H2 than they were in H1. The companies that will hold up best are those where AI revenue growth is already visible in the income statement — not those where it remains a forward story priced into multiples.
CoreWeave's inclusion in the Nasdaq-100 is the right metaphor for this moment. The market just officially embedded a company that IPO'd 15 months ago into one of the world's most-tracked indices. That is how confident — or how committed — passive capital has become. The next 90 days will reveal whether that confidence is earned.
The best quarter in six years just closed. The S&P 500 gained approximately 14%, the Nasdaq 100 approximately 18–20%, and the Nasdaq-100 was recomposed to increase its direct AI infrastructure exposure. This was a good quarter for anyone long the AI trade.
Q3 opens with the Fed holding at 3.50–3.75% and potentially moving higher, valuations that leave little room for disappointment, and a July earnings season that must deliver evidence of AI monetization — not just AI capex.
The rally was real. The test that follows is harder than the one the market just passed.
