The Week Ahead

TL;DR

• This is the busiest earnings week of the quarter, and it runs on a single argument: whether the money promised to AI actually arrives.
• Palantir opens it Monday evening, AMD and SpaceX print Tuesday, and Caterpillar reports Tuesday at 6:30am having fallen 23% from a record under a Michael Burry short.
• Wednesday is the healthcare day: Eli Lilly, Novo Nordisk five days after a failed trial, plus Disney and Uber.
• Thursday August 6 is the one nobody is ready for. Roughly 911.5 million SpaceX shares become eligible to trade, against an IPO that sold 555.6 million.
• Friday brings the July jobs report at 8:30am ET, and five companies reporting into it, including Take-Two, three months before Grand Theft Auto VI.

What to Expect From the Stock Market This Week

Last week was megacap earnings. This week is the audit: the last big AI chip read in AMD, the software that AI is supposed to sell in Palantir and Datadog, the electricity and the engines it needs in Vistra, Oklo and Caterpillar, the consumer paying for all of it in McDonald's, Disney and Airbnb, and a payrolls print on Friday that sets the discount rate applied to every one of them.

The Board

Twenty-seven companies, an FDA decision, 911.5 million shares unlocking and payrolls. No quiet day again.

Monday, August 3: Palantir Sets the Mood

ISM manufacturing lands at 10:00am ET, with forecasters looking for a rise to about 54 from 53.3 in June. Above 50 means expansion, and it is the only genuinely quiet hour of the week.

Then the evening arrives. Palantir reports after the close, with the call at 5:00pm ET, and it will be read as a referendum on AI software whether or not that is fair. Consensus is $0.35 of adjusted EPS on about $1.81 billion, up 81%, and the unusual detail is that the street has parked itself above the company's own guidance of $1.797 to $1.801 billion. Analysts normally hide inside a guided range. Here they are explicitly betting Palantir beats a number it drew deliberately tight, with options pricing a double-digit move either way.

Updated August 4: Palantir delivered. Revenue rose 93% to $1.935 billion and the full-year guide went to $8.15-8.16 billion, past the whisper, with the stock up 12-15% in extended trading. The result is marked to market in the Palantir piece, and Snap beat the same evening with revenue up 19% and an above-consensus Q3 guide.

Two more land the same evening and both are better stories than their size suggests. Vertex reports at 4:30pm with options pricing about 6%, in a stock that fell 20.6% on this exact print last year, not because of the quarter but because a pain-drug trial failed in the same release. And Clorox reports at 4:15pm having raised its dividend to $1.25 two days beforehand, annualising to $5.00 against its own guided GAAP earnings of $4.78 to $4.98. The dividend is larger than the profit.

When Does AMD Report Earnings? Tuesday, August 4

After the close on Tuesday. Consensus is EPS of about $1.62 against $0.48 a year ago, more than triple, on revenue of roughly $11.3 billion, up 47%.

Note the bar management set for itself: AMD guided to $11.2 billion plus or minus $300 million, about 46% growth at the midpoint. Consensus sits inside that range, which means the headline is close to pre-agreed and the beat is worth very little on its own. The full AMD preview is here.

AMD is the last receipt in the AI capex argument that has driven every session this month, and it sits on the other side of the ledger from everyone who has already reported. Microsoft answered the capex question with $678 billion of contracted backlog and was rewarded. Amazon showed AWS accelerating to 36.7% and closed up more than 15% at $271.58. Alphabet guided capex up without a receipt and was punished.

Those are all buyers of compute. AMD is a seller. If the hyperscalers are spending what they say, it shows up in AMD's data centre line. A miss would be the most damaging single data point the AI trade has had, because it would mean the promised spending is not converting into orders.

SpaceX also reports Tuesday, its first results since the largest listing of the year: no history to model, no seasonality, no prior guide. And the release itself triggers the first lock-up tranche, which is why Thursday matters. Our full SPCX preview treats the print and the unlock as one event. Update, Tuesday after the close: the print beat on every line, $7.81 billion of revenue and a 9-cent loss; the full results are here. AMD beat and raised too, $11.54 billion of revenue and a $13 billion Q3 guide, and still dropped about 9% in the first extended-hours reaction: priced for blowout, delivered a beat.

Updated August 4, two Tuesday results are already in. Hut 8 reported Q2 before the open: revenue of $74.9 million missed the roughly $80 million consensus, the $177.1 million net loss was mostly unrealized bitcoin marks, and the real news was $7.5 billion of investment-grade project financing for its AI campuses. And in Seoul, SK Hynix closed its last legally gagged session before the US listing rule lapses tonight: no shareholder return announcement was possible, the stock faded about 3% on top of Monday's 7.92% drop, and Wednesday is the first session where the company can actually speak.

The Morning Before It: Caterpillar, McDonald's and Pfizer

Tuesday starts at 6:30am, not at the close, and the pre-market block is where the week's most interesting argument sits.

Caterpillar reports at 6:30am ET against consensus of $6.25 on $19.31 billion, implying 32% earnings growth. It does so having fallen 23% from a record close in five weeks, and that fall has two names on it: Michael Burry shorted the stock at $1,060.98, and Baird cut its target from $1,200 to $900 on the argument that state-level pushback against data centre siting threatens the high-margin power business from 2027. Options price only 6.1% in a stock that has cleared 10% twice in a year.

Then the consumer gets tested. McDonald's reports before the open with US comparable sales expected between 0.5% and 1.1%, down from 3.9% in Q1, because the quarter laps last year's Minecraft promotion. And Pfizer's report lands the same morning with its call at 10:00am, carrying a dividend that has now been frozen at $0.43 for four straight quarters on a yield near 6.9%, which is a board protecting flexibility rather than a board in trouble.

Wednesday, August 5: Eli Lilly, Novo Nordisk, Disney and Uber

Wednesday is the densest day of the week and it splits cleanly in two.

Healthcare first. Eli Lilly reports before the open, call at 10:00am ET, the largest healthcare name of the week and the rare megacap growth story with nothing to do with AI.

Novo Nordisk reports the same morning, and it arrives in a considerably worse mood. On July 31 the Phase 3 ZEUS trial of ziltivekimab failed to reduce major adverse cardiovascular events, taking roughly $30 billion of market value with it and removing the company's most advanced attempt to be something other than a GLP-1 business. Full-year guidance still calls for adjusted sales to fall 4% to 12%. The one line growing is the Wegovy pill, which did DKK 2.26 billion in its first US quarter, about double what analysts modelled. Two obesity giants, thirty minutes apart.

The FDA also decides on Moderna's mRNA-1010 seasonal flu vaccine, which an advisory committee backed 9-0. As we set out in the Moderna breakdown, that date carries more weight than any quarter the company reports this year.

Then the consumer half. Disney reports before the open with consensus at $1.85, up about 15%, and the argument sitting where it always does now: parks against streaming profitability. Uber reports before the open too with options pricing an 8.3% move and the autonomous question deciding the reaction rather than the quarter. Shopify joins them, guiding gross profit to grow slower than revenue (updated August 5: it grew past 30% instead, revenue rose 34%, and the stock surged), and AppLovin closes the day after the bell guiding 52% growth at an 84% EBITDA margin while the stock sits a third lower this year.

Circle reports Wednesday morning too, with its webcast at 8:00am ET: consensus sees revenue up toward $744 million with EPS down more than 80%, and our preview explains why that ugly comp misleads.

And the memory print of the week closes it. SanDisk reports fiscal Q4 and full-year results after the bell at 1:30pm PT (4:30pm ET), against company guidance of $7.75 billion to $8.25 billion of revenue and $30 to $33 of adjusted earnings per share, with consensus sitting at the top of that range or just above it. A year ago the same quarter did 29 cents on $1.90 billion. An Investor Day follows on August 13, and options price the print somewhere between 17.5% and 25% depending on whose screen you read. This is the only scheduled catalyst in a complex that has been trading purely on Seoul's opening bell, and whether last week's rally was a dead cat bounce turns on it more than on anything else this week.

ADP employment and ISM services land the same day. ADP is the first read on Friday.

Thursday, August 6: 911.5 Million SpaceX Shares Unlock

The most mechanically important date on the calendar, and the least discussed.

Roughly 911.5 million SPCX shares become eligible to trade. The widely quoted $116 billion value assumed a share price near $127; at Tuesday's post-earnings-rally close of $125.33 the tranche is worth about $114 billion. For scale, the entire IPO sold 555,555,555 shares. More stock turns sellable on Thursday than the offering that created the public float.

It is a staircase rather than a cliff, running in tranches to December 8, and Musk and a select group of insiders stay locked until mid-2027. Full schedule in the SPCX lock-up calendar.

There is a second, conditional tranche of 455.8 million shares, and it is worth knowing precisely why it almost certainly will not arrive. Those shares release early only if SPCX closes at or above $175.50 on at least five of the 10 trading days in the window running into earnings. That trigger is 30% above the $135 IPO price.

The stock is trading near $110, roughly 19% below the IPO price and close to its all-time low. The trigger sits about 60% above the current price, and it needs to be cleared five times in ten sessions, this week. It is not happening.

That is worth saying plainly rather than leaving as a scary number, and it cuts a specific way: those 455.8 million shares do not disappear, they roll to December 8. So the realistic August supply is the 911.5 million, and the failure to trigger quietly makes the December date bigger.

Eligible to trade is not the same as sold, and lock-ups are heavily pre-positioned, so plenty is in the price. But the sequencing is hostile: earnings Tuesday, supply Thursday. A disappointing first public quarter followed 48 hours later by the largest tranche of new stock is the worst possible ordering, and it is the one the calendar handed us.

Around it, two prints are worth the attention. Datadog reports before the open with its call at 8:00am, and it is the highest-variance name of the day: options price 13%, the last report moved the stock more than 30% against an 11.5% implied move, and full-year guidance implies the company earns in the second half roughly what it earned in the first while revenue grows near 30%. Airbnb reports after the close having missed EPS in three of the last four quarters while beating on revenue every single time. The Trade Desk, DraftKings and MP Materials share that evening, each previewed.

Friday, August 7: The July Jobs Report, and GTA 6 Three Months Out

Nonfarm payrolls and the unemployment rate, 8:30am ET, and this is the number of the month. Capital Economics is looking for about 130,000 as state and local hiring fades.

The reason it matters is June: payrolls at 57,000 against 115,000 expected, with 74,000 of downward revisions and participation at a five-year low. We argued then that the jobs report matters more than CPI now, and the Fed has since made that more true. On July 29 it held at 3.50% to 3.75% on a 9-3 vote, with three members voting to hike, and the 30-year then pushed above 5.19%.

That is the trap, and it is two-sided. A hot number validates the three hawks and drives long yields higher, which is a discount-rate problem for every high-multiple stock regardless of earnings. A cold number confirms the labour market is cracking with inflation still at 3.3%, which is the stagflation read. The band in the middle that the market likes is narrower than usual.

Five companies report into that print, all of them before the open, which is an unkind piece of scheduling. Take-Two is the one with the search volume: bookings for the quarter are expected to fall about 4% and nobody cares, because full-year guidance of $8.0 to $8.2 billion leans on a single date, November 19, when Grand Theft Auto VI ships. This is the first call since pre-orders opened, and the number everyone wants is the one publishers never give. The full TTWO preview is here.

The other two worth knowing are both electricity. Vistra reports with profits expected up 140% and a stock nearer its 52-week low than its high, sitting on a 20-year Meta agreement for 2,609 megawatts that is not in the guidance the market is valuing it on. And Oklo reports as a roughly $6.76 billion company expected to book about $3.8 million of revenue this year, from radioisotopes rather than electricity, which is the purest expression of how much of this theme is still a promise.

One more date sits outside the week but lands on it: Replimune's FDA decision has a Sunday August 2 goal date, so the market reacts at Monday's open. The stock rose as much as 127% on a 10-3 advisory vote after two prior rejections, and we explained why we would not take a position into it.

Economic Calendar This Week: Fed, Inflation and Jobs Data

Strip out the earnings and the macro week is short, front-loaded at the end and entirely about the labour market.

Monday brings ISM manufacturing at 10:00am ET, seen rising to about 54 from 53.3. Wednesday brings ADP employment and ISM services, and ADP is the first read on Friday's payrolls rather than an event in its own right. Friday at 8:30am ET is nonfarm payrolls and the unemployment rate, and it is the only macro release this week with the power to override everything the earnings said.

Index by index, the setup is different in each: the S&P 500 sits 1.7% below its 52-week high after a flat July, the Nasdaq 100 is 8% below its record after its worst month since March 2025, and the Dow was the only major index to gain. You can watch how each of this week's prints lands across all 500-plus names, sector by sector, on our live S&P 500 heatmap.

There is no CPI, no PPI and no Fed meeting. The Fed already moved on July 29, holding at 3.50% to 3.75% on a 9-3 vote with three members voting to hike, after which the 30-year pushed above 5.19%. That leaves payrolls carrying the entire macro week, which is why a light calendar is more dangerous than a heavy one: there is nothing else to average it against.

The Playbook

• Tuesday is a two-event evening. Do not hold short-dated premium through both. AMD and SpaceX print the same day with correlated tape risk and no way to manage in between. Options are pricing an AMD move of roughly 12%, which is the market telling you it does not know either.
• The guide beats the quarter, every time this season. Consensus already sits inside management's own range at AMD, onsemi, Shopify and Datadog. Position for the outlook. That is the lesson Apple taught last week, when a record margin lost to a soft September guide.
• Where implied moves look small against recent realised ones, buy volatility rather than sell it. Caterpillar at 6.1% has cleared 10% twice in a year. That reflex, not the opposite one, is what worked through July.
• Selling premium into a payrolls Friday is not the same as selling it into an earnings print. Macro gaps through strikes and does not respect an expected move. A defined-risk iron condor with real wings beats a naked strangle this week.
• Cash through Thursday is a legitimate position. The SPCX supply date is the only event on this calendar with a known direction of pressure and an unknown size.

The One-Line Read

This week hands you the last big AI chip read in AMD, the software case in Palantir and Datadog, the electricity behind it in Vistra and Oklo, the consumer paying for it in McDonald's, Disney and Airbnb, roughly 911.5 million SpaceX shares becoming sellable on Thursday and a payrolls print that settles whether the Fed's three dissenters were right: last week asked whether the megacaps could justify their spending, and this week asks whether anyone can still afford the discount rate.