The Hike Trade Breaks on a Negative 23,000 Payroll Print. September Hike Odds Collapse and Gold Rips to $4,411.

Friday, August 7, 2026, written at 9:15am Eastern, forty-five minutes before the open, so the pre-market marks below can and probably will move. The July employment report landed at 8:30 and it broke the trade that has organised this market since the March oil shock. Non-farm payrolls fell by 23,000 against a consensus of roughly 80,000, the first outright decline in months, and the Bureau of Labor Statistics took another 103,000 out of the prior two months, cutting May by 66,000 and June by 37,000 to leave June at just 20,000. Average hourly earnings rose 3.2 percent year over year against 3.5 percent expected. The unemployment rate fell to 4.1 percent from 4.2 percent, which sounds like the one piece of good news in the release and is in fact the worst part of it, because the participation rate dropped to 61.4 percent, the lowest reading in more than five years. The rate improved because people stopped looking, not because anybody got hired. September hike odds collapsed to roughly 20 percent from about 55 percent before the print, and from the 82 percent the futures market carried in late July. For five months the working assumption on every desk has been that a triple-digit crude shock forces the Fed to tighten into a slowing economy. Half of that assumption died this morning.

Overnight and Pre-Market Developments

Macro: What Actually Broke This Morning

The important thing about this report is not that it was weak. Yesterday’s jobless claims at 199,000 and the lowest Challenger layoff count in two years said the labour market was not deteriorating, and both of those are still true. Firing has not accelerated. What has stopped is hiring, and the two are different economic events with different policy implications. A firing wave shows up in claims and in credit. A hiring freeze shows up exactly where it showed up this morning, in a negative payroll number driven by local government education shedding 50,000 positions, leisure and hospitality losing 40,000 and retail trade losing 19,000, alongside a participation rate that falls because discouraged workers stop being counted.

That distinction is why the revisions matter more than the headline. A single negative month inside a noisy series is not a trend. A negative month plus 103,000 removed from the prior two is a re-levelling of the entire second quarter, and it means the economy has been running at a materially slower pace of job creation than the Fed believed while it was building the case for a September move. The three-month average is now close to zero.

Average hourly earnings at 3.2 percent is the piece that finishes the argument. The hawkish case for tightening into an oil shock was never really about oil. It was about the risk that energy costs pass through into wage demands and become embedded. Wage growth undershooting expectations while participation falls is the opposite of that dynamic. There is no second-round effect visible in this data. The Fed is left facing an inflation impulse that is entirely imported through the Strait of Hormuz, and the funds rate does not clear tankers.

The Gold Tell, Again

Gold has been the most honest instrument in this market for two weeks and it stayed honest this morning. December futures opened flat at $4,298.30 and ran to $4,411.70 by 8:45, closing out a week up 4.8 percent and a year up 27 percent. The reflex reading is that gold rallies when hike odds fall because the opportunity cost of holding it drops. That is true and it is incomplete.

What gold is actually pricing is the gap between where the Fed is going and where inflation is. Hike odds collapsing while Brent still trades above $80, and while the Hormuz terms are getting worse rather than better, means the market now expects the nominal rate to stay put or fall while the price level keeps getting pushed by energy. That is a declining real rate, and a declining real rate with a central bank that has lost the ability to respond is precisely the environment gold was built for. The equity market is reading this morning’s print as a discount-rate gift. Gold is reading it as a policy constraint. Both can be right for a while, and gold is usually right for longer.

Equities: a Rotation, Not a Rally

Look at the composition rather than the direction. Nasdaq 100 futures are up 1.16 percent, or roughly 343 points, while Dow futures are up only 0.31 percent and Russell 2000 futures were sitting lower into the print. If this were a market celebrating better growth, small caps and cyclicals would lead. They are not. Long-duration technology is leading by nearly four to one over the Dow, which is the signature of a pure discount-rate move rather than an earnings move.

That should temper the enthusiasm slightly. A negative payroll print is good for equities exactly as long as it stays a rate story and does not become a demand story. The bridge between those two states is corporate revenue, and this week’s earnings have been giving a split answer. All three major indices are still tracking their best week since the April to May stretch, which is a genuine result after a Thursday that saw the Dow drop 464 points from a record close to 53,885 while the S&P eased 0.18 percent to 7,709.96.

Quick Take: the Earnings That Matter

Atlassian (TEAM), fiscal Q4, reported after Thursday’s close

Our take: A 25 percent EPS beat and a 28 percent growth rate at this revenue scale is not a normal software quarter, and it was the guide rather than the print that produced a 32 percent move. The thesis matters more than the number. For eighteen months the bear case on seat-based developer tools has been that coding agents shrink headcount and therefore shrink seats. Atlassian just grew 28 percent and guided higher, which says agents are being provisioned as additional licensed users inside the same workflow rather than displacing the humans who buy them. If that holds through one more quarter it repairs a multiple across the whole category.

The Trade Desk (TTD), Q2, reported after Thursday’s close

Our take: A five percent revenue miss does not produce a 29 percent drawdown on its own. What produces it is a miss plus a guide down in a business the market had already begun to suspect is on the wrong side of the AI transition. Independent demand-side platforms exist to arbitrage inventory that the walled gardens price inefficiently, and generative ad tooling inside those gardens compresses exactly that arbitrage. Set this against Atlassian and Cloudflare and the lesson is that there is no such thing as a software trade at the moment. There is software AI expands and software AI eats, and they diverged by sixty percentage points in a single session.

Microchip Technology (MCHP), fiscal Q1 2027, reported after Thursday’s close

Our take: Analog and embedded is the last part of semiconductors to participate in this cycle and it has now turned hard, with 13 percent sequential growth and a 38 percent annual rate off a genuinely depressed base. Management credited data centre, aerospace and defence, industrial and automotive. The read-through is that the AI buildout has stopped being a story about accelerators and has become a story about the entire bill of materials in a rack, which is why we would treat the analog names as a cheaper and later expression of the same trade rather than as a separate recovery.

Vistra (VST), Q2, reported this morning pre-market

Our take: On the numbers this is a miss, with hedge losses cutting into a 31 percent EBITDA gain and a reaffirmation where the market wanted a raise. On the strategy it is the most interesting thing a utility has done this year. Helix puts Vistra’s own equity alongside Nvidia and KKR in a dedicated digital infrastructure vehicle, which converts the AI power story from a contracting exercise into an ownership position in the compute itself. Constellation raised guidance on Wednesday on 920 megawatts of new agreements and Vistra is now taking equity risk in the same demand. The power leg has been the most consistent confirmation of the buildout for two weeks running.

Asia and Europe

Korea reversed almost the entire memory panic. After Thursday’s 4.59 percent collapse and sidecar halt, the Kospi opened at 6,365.07 and pushed through 6,400 within minutes, trading up about 1.8 percent to 6,411.69 with Samsung Electronics up 3.15 percent and SK Hynix up 2.54 percent on 265.5 billion won of foreign net buying. The immediate catalyst was bargain hunting after Wall Street weakness had already been discounted, but the substantive catalyst was the industry report that 2027 DRAM and HBM capacity is fully committed. Japan’s Nikkei was effectively flat at 65,607, down 0.12 percent, with cheaper crude offsetting the semiconductor drag. Hong Kong’s Hang Seng recovered 0.54 percent to 25,668 and the mainland was the strongest market in the region, with the Shanghai Composite up 1.02 percent to 3,940 and the CSI 300 firmer alongside it.

Europe keeps grinding to records without needing a catalyst. The Stoxx 600 closed at 658.19 on Thursday for a third consecutive record, having touched 660.91 intraday, with Spain, France and Italy also setting record closes. This morning the DAX is up 0.96 percent at 26,416, the FTSE 100 up 0.72 percent at 10,946 and the CAC 40 up 0.53 percent at 8,746, with the Euro Stoxx 50 up 0.81 percent at 6,555. The pattern we flagged yesterday still holds. European outperformance is not defensive positioning, it is levered participation in the same semiconductor cycle through names like Soitec, AT&S and STMicroelectronics, without the memory concentration that made Korea the shock absorber this week.

Sector Spotlights

Semiconductors and memory. The timing here is close to comic. On Wednesday night SanDisk and Western Digital printed the two best quarters in the history of storage and lost roughly nine and sixteen percent because their guidance implied the price curve might flatten. On Friday morning the industry reported that Samsung, SK Hynix and Micron have sold every wafer of 2027 DRAM and HBM output and are filling customers at 60 to 70 percent of requested volume, with NAND allocations finalising this month. SK Group has put 2027 AI chip demand growth at 60 to 100 percent and storage demand at 50 to 60 percent. Sold-out capacity two years forward with rationed allocation is not a flattening price curve, it is a steepening one. We said yesterday we would be slower to buy the memory dip than the tape suggested, because a multiple built on a price curve takes time to rebuild. This report shortens that timeline meaningfully, and Korea has already voted.

Software. Atlassian up 32 percent, Cloudflare up 16 percent, the Trade Desk down 29 percent, and one day earlier Datadog down 20 percent and HubSpot down 24 percent. Anyone still trading software as a single factor exposure has had an expensive week. The organising question is whether a given business sells the seats, capacity or distribution that agents consume, or sells an intermediation layer that agents route around. Atlassian and Cloudflare sit on the first side. The Trade Desk sits on the second. Datadog and HubSpot sit in the awkward middle, where the product is genuinely needed but the pricing model is being rewritten mid-contract, which is why both beat and still lost roughly a fifth of their market value.

Energy and power. Brent is down about eight percent on the week and WTI 5.3 percent, and almost none of that is fundamental. OPEC+ approved a sixth consecutive monthly output increase for September, which is supply, but the Houthis struck southern Saudi Arabia and injured eleven civilians, hit Najran airport, attacked a Saudi-linked Red Sea vessel and are maintaining a declared blockade, which is not. The physical picture has not eased. What eased was the war premium that had been priced when Brent traded above $100 in late July. The electricity side of energy is doing something more durable, with Constellation raising and signing 920 megawatts on Wednesday and Vistra committing a billion dollars of equity to a compute vehicle this morning.

Space and orbital compute. SpaceX rose more than six percent on Thursday as its first lock-up tranche released up to 911.5 million shares, roughly doubling the tradable float to about twelve percent of shares outstanding. The stock had closed at an all-time low of $108.27 on Wednesday after a 14 percent fall, so the supply arrived into a name already twenty percent below its $135 issue price, and it went up anyway. That is a cleaner signal than the Starmind headline itself, because it tells you insiders are not the marginal seller they were assumed to be. The second tranche of 319 million shares comes on August 12.

Geopolitics: the Deal Advances and the Terms Deteriorate

The Hormuz negotiation has now been moving in two directions at once for three sessions. Iran says the arrangement with Oman is agreed in principle and the shipping lane coordinates are fixed. Simultaneously Iran is seeking to bar United States and Israeli vessels from the corridor entirely, ban Israel-linked cargo, and charge other users fees covering insurance and environmental costs, with a parliamentary committee reviewing a bill that would fine violators up to 20 percent of cargo value. The United States has rejected this outright, with an official stating that any temporary routes will carry no approvals, no permissions, no tolls and no charges. Secretary Rubio has described progress without finality.

This is not a deal converging. It is two parties agreeing on a map and disagreeing on who is allowed to use it, which is the part that determines whether crude actually flows. Brent falling eight percent this week on an arrangement whose central term Washington has publicly refused strikes us as the market pricing a headline rather than an outcome. We would not chase crude lower here.

Separately, tomorrow is the expiry of the ten-day ultimatum Trump issued to Russia on July 28, after which 100 percent secondary tariffs on buyers of Russian energy are threatened, primarily affecting China and India. A Senate sanctions bill cleared a procedural vote 86 to 12 but has stalled over the tariff authority it would grant the president. A second energy clock runs out on a Saturday, which means the market prices it Monday morning rather than today.

Key Events Today and Ahead