Forget Rate Cuts — Wall Street Just Priced In a Fed Hike

Markets spent most of July pricing two separate tail risks at once — a widening war in the Middle East, and a bubble bursting inside the AI infrastructure trade — and this week both began to unwind, for reasons that have almost nothing to do with each other.

The Federal Reserve held rates steady on Wednesday, but three regional presidents dissented in favour of a hike, an unusually hawkish signal that pushed September rate-hike odds toward two-thirds to four-fifths and sent the 30-year Treasury yield to levels last seen in 2007. That happened the same week core inflation data actually cooled. The tension between those two facts — a hawkish Fed dissent and a disinflationary print — is arguably the single most important thing happening in markets right now, and it isn't resolved.

Then, over the weekend, President Trump said the US would hold off further strikes on Iran, citing what he called the "perimeters of a deal" to reopen the Strait of Hormuz. Oil fell sharply Monday morning, unwinding a chunk of the war premium that has been built into energy prices since the conflict first erupted at the end of February.

Separately, and more violently, the AI trade's leverage got tested — and broke in at least one place. Leopold Aschenbrenner's Situational Awareness, a hedge fund that had grown to roughly $45 billion betting on AI infrastructure, was forced to sell its entire public stock book to Citadel after margin calls — a forced-selling event that intersected with an already-brutal month for memory-chip stocks and pulled the Philadelphia semiconductor index down roughly a fifth from its June highs.

Underneath all of that volatility, the thing equity investors are actually meant to be pricing — corporate earnings — kept doing what it's done all year: beating. Roughly 300 S&P 500 companies have now reported, with about 85% beating estimates and aggregate profit growth tracking above 45%. That's the "signal from noise" tension this week leaves us with: the macro and positioning story was genuinely chaotic, but the fundamentals underneath it were not.

Equities: Capital rotated hard within equities rather than out of them. It moved toward mega-caps with AI monetisation already showing up in reported numbers — Microsoft's Azure, Amazon's AWS and advertising — and away from more leveraged or story-driven AI-adjacent names, particularly memory chips and newly public AI infrastructure plays. That split explains why the Dow notched its fourth consecutive monthly gain in July even as the Nasdaq fell for the month.

Bonds: The curve is sending a genuinely confused signal. The 30-year yield sits at its highest level since 2007, reflecting the hawkish Fed dissent and oil-driven inflation fear, while the 10-year eased back off its own highs into Monday morning as the Iran de-escalation reduced some of that same inflation risk. Long-end and short-end are, for now, telling two different stories.

Dollar: Softer into Monday, hitting its weakest level since mid-June after Japanese and US authorities intervened directly in currency markets to support the yen — itself a sign of how uncomfortable policymakers have become with the pace of currency moves.

Commodities: Oil fell more than 5–6% Monday morning on the Iran de-escalation news, a sharp reversal after a month in which crude had rallied more than 20%.

Gold: Whipsawed in the opposite direction to oil all week — falling when the conflict escalated, rising when it didn't — a reminder that gold is currently trading as an inflation-and-dollar hedge first, and a pure safe-haven second.

Bitcoin: Quietly decoupled from the rest of the "risk" complex. It gained roughly 7.5% in July even as equity-adjacent risk assets like semiconductors and momentum-factor funds were getting hit hard — a sign crypto is trading more on its own liquidity dynamics this month than as a leveraged proxy for risk appetite generally.

What this tells us: capital isn't fleeing risk broadly. It's being unusually selective about which risk it's willing to hold — rewarding balance-sheet-backed AI exposure and punishing leveraged or narrative-only exposure to the same theme.

What happened: The FOMC left rates unchanged on Wednesday, 29 July, but three regional Fed presidents dissented in favour of a 25-basis-point hike — an unusually hawkish split for a hold decision.

Why expectations changed: Markets had been debating hold-versus-cut; the dissent repriced the conversation toward hold-versus-hike, with September hike odds moving into the 60–80% range depending on the source, and the 30-year yield touching its highest level since 2007.

Immediate market impact: Duration-sensitive and momentum-driven trades came under pressure mid-week before a sharp Thursday–Friday rebound.

Longer-term implications: This is happening under Kevin Warsh, still a relatively untested Fed chair, and it raises a real question about whether policy is responding to an oil-driven inflation spike that may already be reversing, rather than to the underlying trend.

What we're watching next: Friday's jobs report. June's print undershot badly; another soft number would sit awkwardly next to a Fed board leaning hawkish.

What happened: Trump said over the weekend the US had agreed to hold off further strikes on Iran, describing "the perimeters of a deal" that would include reopening the Strait of Hormuz. Oil dropped sharply Monday morning.

Why expectations changed: After months of episodic escalation since late February, this is the most concrete step toward de-escalation the conflict has produced.

Immediate market impact: Oil down more than 5%, gold firmer, the dollar softer, the 10-year yield easing off its Thursday peak.

Longer-term implications: If it holds, this removes a real plank of the hawkish Fed case — much of the recent inflation scare has been energy-driven, not demand-driven.

What we're watching next: Whether the "perimeters of a deal" becomes an actual signed agreement, or joins a list of prior pauses this year that didn't stick.

What happened: Leopold Aschenbrenner's Situational Awareness, an AI-focused hedge fund that had grown to roughly $45 billion in assets, was forced to sell its entire public stock portfolio to Ken Griffin's Citadel at a discount after margin calls tied to reportedly heavy leverage.

Why expectations changed: It's the clearest evidence yet that some of the AI infrastructure trade's gains were leverage-driven rather than purely fundamentals-driven — and that unwind can happen in days, not months.

Immediate market impact: The Philadelphia semiconductor index fell roughly 20% from its June highs; SK Hynix, Micron and SanDisk all fell 30–50% in July alone, in some cases their worst monthly performance since the 2008 financial crisis.

Longer-term implications: Forced-selling drawdowns tend to behave differently from fundamentals-driven ones — they can overshoot and partially reverse once the forced seller is done. Some buy-side managers are already treating this one that way.

What we're watching next: SpaceX reports its first-ever quarterly earnings as a public company this Tuesday, a real-time test of whether public markets will re-rate this trade on fundamentals now that forced-seller flow has largely cleared.

What's actually being debated on institutional desks this week, roughly in order of how often it's coming up:

Whether "circular financing" structures inside AI infrastructure deals — compute-for-equity arrangements, chip vendors financing their own customers — are flattering reported demand more than underlying end-demand justifies. Whether the big hyperscalers' roughly $720–745 billion combined 2026 capex guide is now a floor or a ceiling, given Meta alone saw free cash flow collapse by more than 90% this quarter as capex consumed nearly all operating cash flow. Whether China's rapidly scaling domestic memory manufacturing structurally caps the pricing power that drove the AI-memory trade in the first place. Whether a hawkish three-person dissent on a still-new Fed board is a stabilising or destabilising dynamic heading into September. And, a smaller but recurring thread: retail investor sentiment has stayed unusually bearish for well over a year even as the S&P sits near highs — the kind of gap between sentiment and price that desks watch closely without ever fully trusting.

Growth: US Q2 GDP growth came in softer than expected, though the headline figure masked real underlying resilience — business investment, led by AI-related capex, continued to do a lot of the work.

Inflation: Core PCE, the Fed's preferred gauge, cooled to roughly 3.3% year-on-year from 3.4%, with the monthly core print undershooting consensus. Headline PCE eased to around 3.7% from 4.1%. This is a genuinely disinflationary read.

Employment: June nonfarm payrolls badly missed expectations, adding well under half the consensus forecast, following downward revisions to the prior two months. July's report lands this Friday.

Central banks: The Fed held, but with a hawkish three-person dissent. The Bank of England also held, at 3.75%, on a 6–3 vote, with the three dissenters favouring a move to 4%; UK inflation ran at 2.6% in June, with policymakers expecting it to climb again later in the year on energy pass-through. The Bank of Japan held as well, while authorities intervened directly in currency markets to support the yen.

Bond markets: An unusually inverted signal between short and long-dated conviction — the long end pricing in higher-for-longer inflation risk even as the short end starts to digest a genuinely cooler core inflation print.

Bottom Line: This is a macro dashboard arguing with itself. Soft jobs data and cooling core inflation point toward easier policy; a hawkish Fed dissent and an energy-driven headline inflation scare point the other way. Both the Fed and the market are visibly still deciding which signal to trust — and that unresolved tension, more than any single data point this week, is the actual story.

Money moved away from leveraged or narrative-driven AI infrastructure exposure — memory chips, semiconductor equipment, newly public AI-adjacent companies — and toward two different kinds of safety: mega-cap platforms with AI monetisation already visible in reported numbers, and old-fashioned cash-generative value, particularly energy and UK-listed banks and miners, which benefited from strong commodity prices and (for refiners specifically) wide margins tied to the same Middle East disruption that's been driving the broader volatility.

This is being driven more by positioning and near-term guidance credibility than by valuation. Several of the stocks that fell hardest this week aren't expensive on a growth-adjusted basis; what changed was investor confidence in the near-term payoff on capex.

Is it sustainable? Only partially. The rotation into proven AI monetisers looks structural. The rotation out of memory looks more like a forced-liquidation air pocket that could partially reverse once the market finishes digesting this week's forced-seller flow — though a genuine deceleration in hyperscaler capex would make it structural too.

Alphabet — reported the week prior, on 22 July, but set the tone the rest of the sector was measured against all week. Its post-earnings decline established just how little patience the market currently has for AI capex without a clear near-term payoff.

Microsoft — reported Tuesday. Azure growth accelerated into the low-40% range year-on-year, and the market treated its capex plans as the most credible of the week's cohort, sending shares to a multi-month high.

Meta — reported Wednesday, the same day as the Fed decision. Revenue grew strongly, but earnings missed on a mix of legal costs, severance tied to layoffs, and rising depreciation; free cash flow fell by more than 90% quarter-on-quarter as capex absorbed nearly all operating cash flow. Shares fell sharply.

Amazon — reported Thursday. Revenue and earnings both beat comfortably, helped by a large one-off investment gain, and AWS growth accelerated to its fastest pace in more than four years. Shares jumped double digits.

Apple — also reported Thursday. Revenue and earnings both beat, but shares still fell, as guidance flagged supply constraints and margin compression for the current quarter. This is this week's cleanest illustration of a pattern worth understanding on its own — see Market Myth, below.

Friday: US July nonfarm payrolls. Why markets care: June's print badly missed, and this is the freshest read on whether the labour market is genuinely cracking or simply normalising. Bullish outcome: a moderate, in-line print with a steady unemployment rate supports a soft-landing narrative without reviving hike fears. Bearish outcome: either a sharp miss (recession repricing) or a strong beat (reinforces the hawkish Fed dissent) — this is a "good news can also be bad news" release. Most likely outcome: a modest, unspectacular print, with the market's reaction probably more volatile than the number itself deserves.

Tuesday: SpaceX's first-ever earnings as a public company. Why markets care: the stock has fallen by roughly half from its post-IPO peak, and this is the first real fundamentals check against that move. Bullish: a clean print stabilises sentiment across the broader AI-infrastructure complex. Bearish: any confirmation of bearish theses already circulating on Wall Street about the durability of its AI-related revenue could trigger another leg lower. Most likely: a noisy, mixed reaction — there isn't enough reporting history yet to know how the market will read it.

Through the week: roughly a fifth of the S&P 500 reports, including AMD (a direct read on AI-chip demand), Caterpillar (global industrial and construction confidence) and McDonald's (lower-income consumer health).

In the UK: Diageo's full-year results and strategy update, alongside the construction PMI. Why UK readers care: Diageo is a read on consumer staples pricing power in a soft-drinks and spirits market that's had a difficult year; the construction PMI feeds directly into the Bank of England's next policy decision.

Investment Thesis: High-bandwidth memory, DRAM and NAND remain, on the structural case, capacity-constrained relative to AI datacentre demand, with a small number of producers controlling the overwhelming majority of qualified supply.

Why Now: This segment just had its sharpest, fastest drawdown of the entire AI cycle — individual names down 30–50% in a single month, even though the group is still up on the order of 60% since its spring lows and has added several trillion dollars in combined market value over that stretch. A violent drawdown, still-enormous prior gains, and a forced seller that has largely finished clearing its book is exactly the setup value-conscious buyers look for, provided the demand story is actually intact.

Bull Case: SK Hynix has beaten estimates for six consecutive quarters, capacity for 2026–27 was reportedly committed well in advance of this month's panic, and multiple asset managers are already describing the pullback as a valuation opportunity rather than a fundamental break.

Bear Case: China's ChangXin Memory Technologies (CXMT) completed a Shanghai listing explicitly to fund further domestic DRAM capacity expansion, and the same supply shortage that drove prices higher a few months ago is now easing as high-bandwidth-memory lines ramp globally at SK Hynix, Micron and Samsung alike. If hyperscaler capex growth genuinely decelerates from its current run rate, memory is usually the first place that shows up.

Key Catalysts: SpaceX's earnings this Tuesday (an indirect read-through), Nvidia's own results later in August, and any fresh hyperscaler capex commentary.

Risks: This week proved real leverage sits inside parts of this trade. Further forced selling could resume with little fundamental warning if it does.

What Would Change Our View: A confirmed slowdown in hyperscaler capex guidance — as opposed to one company's single-quarter cash-flow wobble — would be the strongest bearish signal. A confident, beat-and-raise 2027 outlook from one of the other major producers would support the bull case.

This week was a live demonstration. Two of the five biggest tech reporters beat headline expectations on revenue and earnings — and fell anyway. The other two also beat — and rallied hard.

The mechanism is simple, even if it doesn't feel that way watching it happen: share prices react to the change in expectations about the future, not the level of last quarter's results. A beat that arrives alongside guidance the market reads as weaker — supply constraints, margin compression, a capex bill that's starting to outrun cash generation — can matter more to the share price than the beat itself. That's what happened to the two decliners this week: strong quarters, undermined by forward-looking detail investors didn't like.

Conversely, a beat that comes with credible evidence the growth is durable — accelerating cloud revenue, a capex plan the market judges "defensible" — gets rewarded even when the headline numbers aren't dramatically different in kind.

The practical takeaway: the print tells you what already happened. The guide tells you what management thinks happens next. Markets price the second one.

This briefing is provided for general educational and informational purposes only and does not constitute financial advice or a personal recommendation. It does not take into account your individual circumstances, financial situation, or investment objectives. The value of investments can fall as well as rise, and you may get back less than you invested. Past performance is not a reliable indicator of future results. If you are unsure about the suitability of any investment, you should seek advice from an FCA-authorised financial adviser. Alpha One is not authorised to provide regulated financial advice.