Wednesday was the precise inverse of Tuesday - and anyone who read yesterday's diary knows why.
| Close | Change | |
|---|---|---|
| Nasdaq 100 ETF | 717.30 | -0.90% |
| S&P 500 ETF | 769.79 | -0.20% |
In dollars, per the ETFs tracking the indices.
Again the Nasdaq moved harder than the S&P - only this time in the other direction.
The reversal: those who rose on expectation fell on the numbers
Yesterday we wrote that the four companies reporting in the evening jumped during the session, before publishing a single number. Today we got the other side of that trade.
| Company | 3 Aug | 4 Aug (pre-report) | 5 Aug (post) | Net over two days |
|---|---|---|---|---|
| AMD | 484.64 | 518.58 (+7.00%) | 482.05 (-7.04%) | -0.53% |
| Astera Labs | 321.05 | 361.67 (+12.65%) | 318.43 (-11.96%) | -0.82% |
| SpaceX | 114.53 | 125.33 (+9.43%) | 108.27 (-13.61%) | -5.47% |
| Arista | 184.89 | 190.51 (+3.04%) | 197.31 (+3.57%) | +6.72% |
The figure that sums up the week
AMD and Astera Labs returned almost exactly to where they stood before the rise.
Minus 0.53% and minus 0.82% across two days - after rising 7% and 12.65% and then falling 7% and 12%.
And this happened despite the reports being good. AMD published record revenue of $11.5 billion and a 107% jump in Data Center. Astera Labs posted 104% growth and guided to another 40% in a single quarter.
The conclusion is not that the reports were weak. The conclusion is that the price had already embedded them before they were published.
And Arista, the only one that did not jump before its report, was the only one to rise after it.
And in Israel: two crashes that look irrational - until you read the guidance
SolarEdge: minus 30.48%
The stock fell to $33.90 from $48.76.
And what the company reported sounds like good news:
- Revenue of $346.2 million, growth of 20%
- The sixth consecutive quarter of gross margin expansion
- A return to non-GAAP operating profitability - for the first time since the second quarter of 2023
- Positive free cash flow
CEO Shuki Nir: "Our second-quarter results mark an important milestone in SolarEdge's turnaround."
So why did the stock crash 30%
The answer is in the guidance, not the quarter.
Third-quarter guidance: $310 million to $340 million. The midpoint is $325 million - roughly 6% below the $346.2 million the company had just reported.
The company is guiding to a decline next quarter, not to continued growth.
And there is a second, sharper detail. The release states explicitly that the guidance excludes potential tariff refunds, and that only including $11.5 million of IEEPA refunds received in July does the midpoint imply a non-GAAP operating profit in the third quarter.
Meaning: without the tariff refunds, the guidance does not imply an operating profit.
And that is exactly the same line we saw at Caterpillar - which booked $392 million of IEEPA refunds inside operating profit. Two companies in the same week whose reported profitability leans on tariff refunds.
And a note worth attention: in the same release SolarEdge states it continues to advance the SolarEdge SST to address the significant opportunity in AI factories. It too is aiming at data centers.
Taboola: minus 27.50%
From $5.29 to $3.84.
Here too the company's headline is positive: "Taboola Reports Strong Q2 2026 Financial Results, & Raises Full-Year ex-TAC Gross Profit and Adjusted EBITDA Guidance."
What was reported:
- Revenue of $476.8 million, up just 2.4%
- Gross profit of $139.5 million, up 2.9%
- ex-TAC gross profit of $192.4 million, up 11.8%
And again - the guidance
Third-quarter guidance: revenue of $460 million to $473 million. Midpoint $466.5 million - below the $476.8 million just reported.
And ex-TAC gross profit is guided to $184-190 million, below the quarter's $192.4 million.
So the company raised its full-year guidance while guiding to a third quarter weaker than the second.
And on top of that: revenue growth of 2.4% is close to standing still. The 11.8% growth is in the ex-TAC measure - that is, excluding what is paid out to publishers - and that is the metric the company chooses to emphasise, not the revenue line.
Our own pieces from today
| Company | Close | Change |
|---|---|---|
| Spotify | 482.23 | +0.85% |
| TeraWulf | 18.07 | -4.29% |
| Uber | 68.18 | -5.29% |
Uber fell too - despite crossing $10 billion of annual free cash flow for the first time. And TeraWulf fell, despite a $19 billion contract with Anthropic. This was a day when good news was not enough.
Other notable moves
| Company | Close | Change |
|---|---|---|
| Eli Lilly | 1,169.86 | +4.86% |
| Zimmer Biomet | 98.16 | +2.45% |
| Iron Mountain | 127.13 | +1.29% |
| ICL | 5.23 | +1.55% |
| InMode | 15.21 | -1.04% |
| Gilat | 11.62 | -2.43% |
| Kraft Heinz | 25.73 | -3.42% |
| GlobalFoundries | 49.46 | -4.92% |
| CVS Health | 99.12 | -5.08% |
| Cipher | 18.71 | -8.19% |
| Tower | 211.14 | -11.85% |
| New York Times | 65.48 | -13.40% |
Tower kept falling - a day after traversing a 21% range and closing higher. The two big Israeli names of the week, Tower and SolarEdge, fell together on the same day.
Who reported today
The full record of companies that filed that day. Those we did not analyse are listed only:
אלי לילי, CVS Health, קראפט היינץ, זימר ביומט, GlobalFoundries, Iron Mountain, Global Payments, קרלייל, ניו יורק טיימס, Flutter Entertainment, Circle, Owens Corning, Suncor, Thomson Reuters, Shake Shack, Trump Media, Galaxy Digital, כיל, גילת, סולאראדג', InMode, Kaltura, טאבולה, אבן קיסר, TAT Technologies, BorgWarner, CDW, Charles River, Choice Hotels, Criteo, Delek US, Elanco, Eos Energy, EVgo, Freshpet, Hudson Pacific, Icahn Enterprises, Insulet, LivaNova, Louisiana-Pacific, Medline, NCR Atleos, NCR Voyix, NiSource, Perrigo, Phillips 66, Primoris, Recursion, Regal Rexnord, Royalty Pharma, SharkNinja, Shutterstock, Southwest Gas, United Therapeutics, Utz Brands, Vishay, XPEL.
What is coming tomorrow
Thursday, 6 August. We will update through the day according to who actually files.
My Angle
A personal opinion of Ilan Abramov - not advice, not a recommendation
This was the day the thesis I wrote yesterday was tested - and got a sharp answer.
Yesterday I wrote that the market bought the AI infrastructure chain on expectation rather than on numbers. Today we saw the other side: AMD and Astera Labs returned almost exactly to the price they came from. Up 7% and 12.65%, then down 7% and 12%. A round trip, with nothing in the business having changed.
And what I take from it is not about AMD - it is about timing. When a stock rises 12% on reporting day, before the report, the investor is no longer buying a result - they are buying someone else's bet on the result.
And Arista proves the point from the other side. It was the only one that did not jump before its report, and the only one that rose after it. Plus 6.72% across two days, against declines for the rest.
As for SolarEdge and Taboola - both teach exactly the same lesson, and I think it is the important one of the day. Both published a release that reads like good news. And both guided to a quarter below the one they had just reported. The market does not buy the quarter that happened - it buys the one coming.
And what troubles me about SolarEdge more than the fall itself: the company writes that only including $11.5 million of tariff refunds does the guidance midpoint imply an operating profit. That is profitability leaning on a refund rather than on the business - and it is now the second company this week, after Caterpillar, where IEEPA refunds are a material part of the picture. Worth watching whether this becomes a pattern.
