Oil Puts the September Hike Back on the Table as Brent Hits $89. Intel Upsizes Its Raise to $20B.

Tuesday, August 11, 2026, written at 9:20am Eastern, ten minutes before the bell, so the marks below are live and several will move within the hour. The index level is not the story and neither is the earnings tape. The story is that four trading days ago July payrolls printed minus 23,000 and the market declared the hiking cycle finished, and since then crude has risen four consecutive sessions and put the September hike back to a coin toss. Nothing about growth changed. Brent changed. That is a different market from the one that closed last Friday, and almost nobody has repriced equities for it.

Top Call: the Fed’s Reaction Function Now Runs Through Brent

September hike odds were roughly 67% a week ago, fell to somewhere between 40% and 44% on the payroll miss, and are back to about 52% this morning. The 2-year yield is at 4.241%, up more than three basis points, and the 10-year is at 4.725%, closing on its highest level since January. None of that move came from labour data or from a Fed speaker. It came from Brent at $89.41, up $1.69 and higher for a fourth straight session, on a US-Iran negotiation that has visibly stopped negotiating.

This is the uncomfortable configuration, and it is worth naming plainly. A central bank being pushed toward tightening by an energy supply shock it has no ability to influence, into a labour market that has already begun shedding jobs, is the textbook policy-error corner. The single cleanest confirmation that the market understands this is gold. Spot is at roughly $4,445 to $4,447, a two-month high, up 0.6% on the same morning that rate-hike odds rose eight points. Gold rallying alongside rising policy rates is not a growth trade and not a rate-cut trade. It is a hedge against the committee getting this wrong in either direction.

Equities are not engaging. Nasdaq futures are up 0.34%, the S&P is up 0.16% at 7,789, the Euro Stoxx 50 set another record, and the VIX sits at 15.54. That gap between what the front end is pricing and what equity volatility is pricing is the actual risk into tomorrow morning.

The Tape

Futures have oscillated around unchanged since roughly 6am, printing modestly negative on some screens and modestly positive on others, which is the correct description of a session with Tokyo shut and the only data that matters twenty-four hours away. Note that Monday’s cash session was a broad, quiet fade with small caps down 0.56%, roughly five times the S&P’s decline. Small caps are the most rate-sensitive cohort in the index complex, and they have now underperformed on both of the last two sessions in which the front end sold off. That is coherent, and it is the one part of the equity market actually trading the rates story.

Intel: the Raise Cleared, and It Cleared Bigger

Yesterday the framing was that the AI buildout had become a financing story, because Intel had to ask shareholders for $15 billion. Overnight the market answered, and the answer was more constructive than the framing deserved. Intel priced 210,526,315 shares at $95.00 and upsized the deal to $20 billion, a third larger than announced, for net proceeds of approximately $19.7 billion. The underwriters hold a 30-day option on a further 31,578,947 shares. The deal closes tomorrow.

Three readings, in order of importance.

The first is that twenty billion dollars of equity was absorbed in a single overnight book. Whatever anyone thinks about the wisdom of the buildout, the marginal buyer of AI infrastructure paper exists and is deep. That is a genuinely bullish fact for the entire complex, because the bear case circulating for the past week has been that this cycle’s capital requirements would eventually exceed the market’s appetite to fund them. Tested at size, appetite won.

The second reading cuts the other way. A book gets upsized when demand exceeds supply, but a company accepts the upsizing when it wants the money. Intel did not raise $15 billion and stop. It took every dollar available. The initial $15 billion was therefore a floor on the funding gap rather than a measure of it, and management has now told us, through the size rather than the press release, that the capital need at the foundry was larger than the first number suggested.

The third is the price. Intel cleared at $95 having traded above $100 before the announcement. The capital was available, and it was available at a discount. That is the honest accounting: funding is not scarce, but it is no longer free, and the cost is being paid in dilution by existing holders rather than in coupon by the company.

Tonight: Super Micro Is the More Important Print Than CoreWeave

Both report after the close and the market has the emphasis backwards.

Super Micro (fiscal Q4) already pre-announced the substance on July 21, which makes the setup unusual. Revenue is guided to land near the low end of the $11.0 billion to $12.5 billion range against consensus of $11.56 billion, and consensus EPS is $0.96, up more than 124% year on year. But gross margins are estimated at 15% to 17% against a guide of 8.2% to 8.4%, which is not a beat so much as a different business. And the company took in more than $60 billion of new orders in the quarter, with backlog at a record.

Our take: the margin number deserves care rather than celebration. A server assembler does not double gross margin on execution; it does so on customer and product mix, which is by definition not repeatable on demand. The question tonight is not earnings quality, it is the balance sheet. More than $60 billion of orders against roughly $1.3 billion of cash is the same working capital gap Intel just closed with $20 billion of equity, at a company with a fraction of Intel’s rating and access. If the funding answer arrives tonight as a convertible or an equity raise, that is the second AI financing event inside 48 hours, and the market will begin pricing capital rather than demand as the binding constraint on this cycle. If management defers the question, the stock trades the $60 billion and the question returns next quarter, larger.

CoreWeave reports at 5:00pm Eastern with consensus revenue of $2.5 billion, up 109% year on year, against its own guide of $2.45 billion to $2.60 billion, and a consensus loss of $1.17 a share. Guided adjusted operating income is $30 million to $90 million, recovering from first-quarter lows. Capex for 2026 is $31 billion to $35 billion.

Our take: the revenue line is close to pre-determined and the loss is expected, so the print turns entirely on backlog. It stood at $99.4 billion on March 31, up from $66.8 billion at end-2025. The street looks for roughly $104.4 billion. Cantor Fitzgerald models approximately $131 billion. That dispersion, some twenty-six billion dollars wide on a single disclosed metric, is the most informative thing about this stock: nobody outside the company has a reliable model of contract timing. A print near $104 billion will read as deceleration against the Q1 step-up even though it is a record, and a print near $131 billion will be discounted on the grounds that backlog at a company spending $35 billion a year is a liability as much as an asset. The asymmetry is poor, which is why Super Micro’s funding disclosure matters more to the sector tonight.

Rocket Lab: the Loss Was Deal Costs, the Backlog Was Not

Rocket Lab reported after Monday’s close and it was the strongest space print of the cycle. Revenue was a record $234 million, up 62% year on year. Backlog rose 137% to $2.36 billion. Net loss narrowed to $49.26 million. Management guided to another record revenue quarter in Q3.

The EPS shortfall against consensus was caused by $8.6 million of deal costs tied to Mynaric, Motiv and the pending Iridium acquisition, which is to say the miss is the strategy. That distinction matters more than usual here. Rocket Lab is using a launch business compounding above 60% to buy its way up the value chain, and Iridium in particular converts the company from a launch provider into the operator of an in-orbit constellation with recurring service revenue. That is a business model change, not a bolt-on, and the fact that the core kept setting records through the integration is the actual result. Yesterday this note said the print would turn on the Neutron schedule. It turned on backlog instead, and a backlog up 137% is a better answer than a reaffirmed launch window.

Semis: Demand Keeps Not Being the Problem

TSMC reported July revenue of NT$467.58 billion, about $14.5 billion, up 44.7% year on year. That is the cleanest available read on end demand in the entire complex and it says the same thing it has said every month this year. There is no demand problem. Wedbush now expects TSMC to reach 180,000 monthly 3-nanometer wafer starts early in the fourth quarter, two to three months ahead of prior expectations and up from roughly 150,000 in the first half.

Separately, TSMC and Sony announced a $4.69 billion joint venture, Advanced Vision Semiconductor Manufacturing Corp, to build next-generation image sensors in Kumamoto alongside TSMC’s existing JASM plant. Sony contributes ¥465 billion, about $2.92 billion, as controlling shareholder, and TSMC ¥282 billion, with volume production targeted for 2029 and further funding contingent on Japanese government support.

The joint venture is worth a moment because it is the opposite of the trade everyone is in. TSMC is committing several billion dollars and a four-year runway to smartphone image sensors, a mature and competitive non-AI product line, at the peak of AI-driven demand for its leading-edge capacity. Read charitably, that is a company with enough capacity to afford diversification. Read carefully, it is the most concentrated beneficiary of the AI cycle quietly buying insurance against its own concentration, funded partly by a foreign government. Both readings can be true. Neither is bearish today, and both are worth remembering in 2029.

Macro: NFIB Argues With Payrolls

The NFIB Small Business Optimism Index rose 2.4 points in July to 99.8, its highest reading since August 2025 and above the 52-year average of 98.0. Hiring plans contributed the most to the increase, and the employment index rose to 102.1 after four consecutive monthly declines. The Uncertainty Index rose two points to 91 against a historical average of 68.

This directly contradicts the payroll report that reset the entire rates market last Friday. Small firms surveyed across July said they intend to hire; the establishment survey for the same month said employment fell 23,000, with 103,000 of downward revisions to May and June. Both cannot be describing the same economy. The reconciliation is either that intentions have not yet met reality, which is the benign reading, or that the payroll print overstated the deterioration, which is the hawkish one. Given that the uncertainty index sits twenty-three points above its long-run average, the honest answer is that small business owners do not know either.

For today’s purpose the direction is what counts. A firmer labour signal stacks on top of the oil signal, and both point the same way into tomorrow’s CPI.

Asia and Europe: Tokyo Shut, Hong Kong Leads the Way Down

Japan was closed for Mountain Day, which removes the region’s largest and most liquid market from the tape and makes everything below it thinner than it looks. Hong Kong took the worst of it, giving up 1.1% to 25,652 as optimism about a near-term easing of US-Iran tensions drained away. Note the internal split: Hong Kong fell more than twice as much as Shanghai, which is foreign money trimming China exposure faster than domestic money, with Beijing’s promised 2026 fiscal support still on paper rather than in the economy. The Shenzhen Component’s smaller 0.40% decline says the domestic growth complex held up better than the offshore index, the same pattern that has run for weeks.

Europe is holding record territory without extending it. The Euro Stoxx 50 edged 0.2% higher to a new record while the Stoxx 600 traded around the flatline, the DAX slipped twelve points to 26,311, and the FTSE 100 eased to 10,858. The continent’s problem is the same as the Fed’s: an oil-driven inflation impulse arriving with a lag into an ECB that had considered itself finished.

Oil and Geopolitics: Diplomacy Went Backwards

Brent rose $1.69 to $89.41 and WTI added $1.75 to $83.88, both extending a fourth consecutive session of gains, with Brent at its highest in almost two weeks. This is not a bid on demand. It is a bid on the visible failure of talks.

President Trump said the United States is “only semi-negotiating with them” and issued sweeping new demands on Iran, including compensation for people the Islamic Republic has killed in past conflicts, after Tehran reiterated its own call for reparations. Iran’s foreign minister said there is no possibility of restarting negotiations at present. Iran’s Foreign Ministry was more specific: “As long as the U.S. naval blockade continues, the necessary conditions for the reopening of the Strait of Hormuz do not exist.” Pakistan’s defence minister offered the only constructive note, suggesting the parties are closer than they appear, which is the sort of third-party optimism that has been wrong at every prior stage.

The separate Iran-Oman track survives, but the mechanism has not improved. Passage would require maritime service fees paid through a newly established corporation, which remains a toll booth on a waterway Iran does not own, and the fee dispute has been deferred rather than settled. The consistent position in these notes has been that the premium is structural until vessel counts recover, and nothing this morning argues otherwise. Both sides added conditions overnight rather than removing them. That is the definition of moving backwards.

Elsewhere, Ukraine says it has zero Patriot interceptors remaining and is seeking resupply before winter, the season in which Russia has historically targeted energy infrastructure with ballistic missiles. A Ukrainian drone strike killed at least thirteen people inside Russia. This is a second, slower energy risk for the fourth quarter, and it is currently priced at nothing.

Corporate Headlines Worth the Time

Sea Limited is the standout print of the morning, and it is not an AI story, which is precisely why it is useful. Q2 EPS of $0.86 beat consensus of $0.79, revenue rose 48.1% to $7.79 billion against $7.27 billion expected, and Shopee’s GAAP revenue rose 48.9% to $4.9 billion on record GMV and gross order volume. Operating cash flow was $1.51 billion against $5.99 billion of cash. Management now expects Shopee to reach $1 billion of adjusted EBITDA for the full year. A platform growing revenue at 48% while converting it into operating cash at that rate is doing something none of tonight’s AI reporters can currently claim, and the stock is being marked accordingly.

Babcock & Wilcox is up roughly 34% pre-market after beating and raising full-year guidance, and it ordered a gigawatt of turbine sets from Siemens. That order is the AI trade expressed through the grid rather than through silicon, and it is another week in which a power generation name has produced a larger percentage move than the chip names it exists to serve. AECOM raised backlog to a record on the same theme.

Elsewhere: Aramark beat on adjusted EPS at $0.52 against $0.48 with revenue of $5.06 billion against $4.94 billion, and backed 20% to 25% full-year adjusted EPS growth. Alcon beat at $0.84 against $0.75 on revenue of $2.782 billion. Simon Property posted FFO of $3.29 against $3.22. Hims beat on revenue at $753.2 million. On the other side, On Holding missed on net sales at CHF 850.3 million against CHF 879.57 million, Upwork fell about 18% on a full-year revenue guidance cut, ACV Auction guided full-year revenue lower despite a small EPS beat, and Archer Aviation posted a $45.5 million net loss. Plug Power guided full-year revenue growth of 15% to 16%.

Key Events Today

• 6:00am: NFIB Small Business Optimism, released at 99.8, up 2.4 points, highest since August 2025
• 10:00am: Existing home sales
• 1:00pm: Treasury auctions $58 billion in 3-year notes. Worth watching given the front-end move; a soft auction into a 4.24% 2-year would confirm the hike repricing has real money behind it
• After the close: Super Micro fiscal Q4; CoreWeave Q2 at 5:00pm
• 4:30pm: API weekly crude inventories, into a fourth day of gains
• Tomorrow 8:30am: July CPI. Headline expected around +0.1% month on month and 3.3% to 3.4% year on year, core at +0.32% month on month and 2.5% year on year, against June’s +0.25% and 2.4%
• Thursday 8:30am: July PPI
• September 15-16: FOMC, currently priced near a coin toss

What We Are Actually Watching

Three threads, and they meet tomorrow morning at 8:30.

The first is that the inflation print carries an unusually wide wedge. Headline is expected near 3.3% to 3.4% while core is expected at 2.5%. That gap is energy, and it exists because the oil move is recent enough that it has not yet propagated into core. The consequence is that a benign core print tomorrow proves very little. It measures a period before Brent went to $89, and the committee knows that. A firm core print, by contrast, means passthrough has already started, which would be the genuinely hawkish outcome and would take September from a coin toss to a decision.

The second is that the AI cycle’s constraint has moved from demand to capital, and the evidence keeps arriving in the same form. TSMC at plus 44.7%, Super Micro with $60 billion of quarterly orders, CoreWeave with a backlog somewhere between $104 billion and $131 billion, Rocket Lab up 137%: not one of these is a demand problem. Intel’s $20 billion answered the funding question affirmatively and at a discount. Super Micro either answers it again tonight or defers it. The tell to watch is not whether the money is available, because today it clearly is. It is the price of it.

The third is that the equity market is not positioned for the rates market it is sitting inside. The front end has moved fifteen or so basis points and repriced September by eight points in two sessions, small caps have registered it, and nothing else has. The VIX at 15.54 into a CPI print that could decide a central bank meeting is the cheapest expression of complacency currently on the board. That does not make the tape wrong. It makes it thin.

Yesterday this note argued the Intel raise put financing at the centre of the AI trade and that the reaction underrated it. The overnight book cleared at $20 billion, so the financing question got a better answer than the framing implied, and that is worth saying plainly rather than quietly. What did not improve is the Strait, and the Strait is now doing more work in the rates market than the Fed is.

Marks above are pre-market and will move by the open. We will mark today’s calls against tomorrow’s print.