Earnings, Energy, and a New Fed Regime: What Drove the Quarter and What We’re Watching Next

Phenomenal second-quarter returns across the major indices erased the anxiety of the first quarter, and strong earnings growth was the primary driver of that recovery. But the sentiment behind those gains looks more fragile than the headline numbers suggest. Beneath a quarter defined by a record-setting earnings season, a geopolitical shock that faded faster than anyone expected, a genuine regime change at the Federal Reserve, and the early innings of two structural supply stories in energy and semiconductors, we see a market that is less forgiving, more selective, and increasingly focused on which companies can actually deliver on the promises priced into their valuations. Below, we break down the major drivers of 2Q returns and the trends we’ll be watching as the second half of the year unfolds.
A Record Earnings Season, With Little Room for Error
Q1 2026 delivered the strongest earnings growth since late 2021, with the S&P 500 posting 28.6% year-over-year growth and roughly 84% of companies beating both profit and revenue estimates. It marked a sixth straight quarter of double-digit gains, and importantly, the strength stretched across nearly every sector rather than resting on a handful of tech giants alone. The biggest storyline, unsurprisingly, was the “Magnificent Seven,” where investors made clear they will keep funding massive AI infrastructure spending, now running near $650–725 billion for the year, but only for companies whose cloud businesses are growing fast enough to justify it. Alphabet and Amazon were rewarded with rallies on that basis, while Meta and Microsoft actually fell despite beating estimates, simply because their spending plans raised fresh questions about payback timelines.
That divide points to a broader shift in market mood. Investors barely rewarded companies for meeting expectations, but punished misses almost twice as hard as usual — a clear sign that with valuations already trading above historical norms, there is little patience left for disappointment. We expect this asymmetry to persist into the back half of the year, and we think it argues for greater selectivity within, rather than broad exposure to, the AI-adjacent trade.
The Strait of Hormuz: A Textbook Case of Market Desensitization
The Strait of Hormuz crisis offered a textbook illustration of how markets adapt to bad news over time. When the war first broke out in early March, fear spiked hard: the VIX jumped above 35 within just nine days, and oil prices posted their largest monthly rise on record. But each subsequent escalation landed with less force than the last. When the administration announced a formal blockade of the strait in mid-April — arguably a more serious development than the initial outbreak — stocks barely reacted, with strategists noting that markets had already priced in the worst-case scenarios.
By early May, even violent swings in oil prices barely moved equities at all, helped in large part by the strength of Q1 earnings giving investors a competing reason to stay optimistic. By June, volatility had settled back to a calm, everyday level, and when a genuine peace deal was finally signed reopening the strait, markets celebrated with a solid rally rather than reacting with fear — the mirror image of how the crisis began. The lesson for us: oil markets stayed jumpy and headline-sensitive for months, but equity investors adapted quickly, treating repeated bad news as old news well before the underlying situation had actually improved. That gap between physical-market risk and equity-market complacency is worth watching closely if tensions in the region resurface.
Fed Regime Shift
Kevin Warsh took over as the Fed’s 17th chair in May 2026, confirmed in the most partisan vote for the role in modern history, and inherited about as difficult a backdrop as any incoming chair has faced: core PCE inflation running near 3.3%, up from 3.0% at the end of 2025, pushed higher largely by the Hormuz-driven spike in oil prices. Despite that, the labor market has held up well, with private employers adding roughly 117,000 jobs a month through May and unemployment steady around 4.3% — a sharp turnaround from the near-stagnant hiring of 2025.
That combination — sticky inflation paired with a resilient labor market — has flipped rate expectations on their head. Investors began the year pricing in two or three cuts by December; they are now pricing in essentially none, and some are even betting on a hike. Warsh, despite a reputation as more dovish, used his first press conference to hammer home the Fed’s commitment to the 2% inflation target and price stability, comments markets read as unmistakably hawkish. He has also signaled a desire to eventually pare back forward guidance and rethink how the Fed models inflation, favoring more attention to supply-side forces like AI-driven productivity and energy shocks over the Fed’s traditional demand-driven framework. The bigger story here is less about any single rate decision and more about a genuine regime shift in how the Fed communicates and thinks about its mandate — layered on top of an economy where a geopolitical energy shock, not the usual demand-driven cycle, is doing most of the work.
Energy: Oil Normalizes While Nuclear Becomes a Structural Story
The oil side of the energy story is entering a calmer chapter. With the Strait of Hormuz reopened and Gulf flows gradually normalizing, most major forecasters — the EIA, JPMorgan, Morgan Stanley — now see Brent easing from the crisis-era $100+ range back toward $75–80 a barrel by 2027 as supply rebuilds, though full normalization of Gulf production could stretch into next year, and any renewed friction in negotiations or shipping could snap the war premium right back.
Layered on top of that near-term oil story is a much bigger structural shift: AI data centers are driving the fastest growth in U.S. power demand in decades, and traditional grid buildouts simply cannot move fast enough — pushing hyperscalers directly toward nuclear. Every major hyperscaler has now signed at least one nuclear deal, with 13 announced projects committing over 9.8 GW of capacity as of May 2026, including Amazon’s $500 million backing of X-energy’s small modular reactor technology and Microsoft’s deal to restart Three Mile Island. Small modular reactors in particular are drawing the bulk of investment because they are factory-built, faster to deploy than a decade-long conventional plant, and can provide the continuous, carbon-free baseload power that intermittent wind and solar cannot match on their own — something Goldman Sachs has flagged as the real bottleneck for data center buildouts. That dynamic is rippling into commodities as well: uranium prices have stabilized around $84–86 a pound, with analysts projecting a move toward $100–125 as reactor demand accelerates.
Beyond nuclear, hydrogen fuel cells and on-site microgrids with battery storage are emerging as secondary options for operators seeking backup or supplemental power. Put together, the picture for investors is less “oil versus alternative energy” and more a genuine bifurcation: oil is normalizing as a geopolitical-risk trade, while nuclear and adjacent power technologies are becoming a structural, multi-decade growth story tied directly to AI infrastructure spending — less a bet on energy prices, and more a bet on which suppliers can actually deliver reliable power fast enough to keep pace with data center demand.
Memory Chips: The AI Buildout’s Clearest Supply Crunch
If energy is one structural bottleneck of the AI buildout, memory is the other — and arguably the clearest supply crunch story of 2026. Unlike past chip cycles, analysts are calling this one structural rather than temporary. Samsung, SK Hynix, and Micron together control over 95% of global DRAM production, and all three have shifted the bulk of their manufacturing capacity toward high-bandwidth memory (HBM), the specialized stacked memory that feeds AI accelerators like Nvidia’s GPUs. The economics explain why: a single HBM3E module sells for roughly $60–100, versus $5–10 for an equivalent amount of conventional DDR5 — an easy choice for manufacturers when capacity is constrained.
The price impact has been extreme. DRAM contract prices surged 80–90% in Q1 2026 alone, and TrendForce is now forecasting another 58–75% increase in Q2 for conventional DRAM and NAND. Micron has confirmed its HBM output is effectively sold out for the year and has retired its consumer-facing Crucial brand entirely to focus on data-center products. This differs from prior boom-bust chip cycles in two important ways: AI infrastructure spending, projected near $650 billion in 2026 across the major hyperscalers, is a structural and still-growing demand source rather than a temporary surge; and new fab capacity takes years to bring online — Micron’s new Idaho and New York facilities will not ship product until 2027–2030 — so simply “making more chips” cannot fix the imbalance quickly. Most forecasters, including Intel, Gartner, and IDC, now point to 2027 or even 2028 before conditions normalize.
The fallout is already visible well outside the AI industry. Laptop, smartphone, and automotive manufacturers are facing 15–20% price hikes and production constraints, with Tesla and other automakers openly warning that the shortage will limit output in 2026. For investors, the read-through mirrors the nuclear and data-center energy story: this is not a cyclical dip to trade around, but a multi-year capacity reallocation in which the companies controlling scarce, high-margin production — Micron, SK Hynix, Samsung — sit in a structurally advantaged position, while downstream consumer hardware makers absorb margin pressure they cannot fully pass on to price-sensitive buyers.
Looking Ahead
Taken together, the second quarter told a consistent story: fundamentals were genuinely strong, but the market’s tolerance for anything less than strong has narrowed considerably. Earnings, geopolitics, monetary policy, and the AI infrastructure buildout all pointed toward the same conclusion — a market rewarding proof over promise, and increasingly discriminating between AI beneficiaries with a clear path to monetization and those still asking investors to trust the spending. As we move into the third quarter, we’ll be watching earnings guidance for signs that the AI capex cycle is translating into durable revenue, tracking how quickly Gulf oil production normalizes against any renewed geopolitical friction, and following the two structural bottlenecks — power and memory — that may end up defining which companies can actually scale the AI buildout they’ve promised.


