Market snapshot, not real-time
S&P 500 (SPY) $769.06 +0.21% Nasdaq-100 (QQQ) $716.08 −0.20% Dow (DIA) $534.27 +0.26% Russell 2000 (IWM) $301.72 +0.50% 10-Year Treasury (IEF) $93.38 +0.48% Crude Oil (USO) $130.91 +0.19% Gold (GLD) $413.84 +3.84% US Dollar Index (UUP) $27.88 −0.92% Volatility (VXX) $19.05 −3.05% Semiconductors (SMH) $560.92 −1.55% Silver (SLV) $60.01 +4.47% Emerging Markets (EEM) $66.11 +1.18% Bitcoin (BTC) $69,297.78 +7.14% Ethereum (ETH) $2,251.69 +17.48% S&P 500 (SPY) $769.06 +0.21% Nasdaq-100 (QQQ) $716.08 −0.20% Dow (DIA) $534.27 +0.26% Russell 2000 (IWM) $301.72 +0.50% 10-Year Treasury (IEF) $93.38 +0.48% Crude Oil (USO) $130.91 +0.19% Gold (GLD) $413.84 +3.84% US Dollar Index (UUP) $27.88 −0.92% Volatility (VXX) $19.05 −3.05% Semiconductors (SMH) $560.92 −1.55% Silver (SLV) $60.01 +4.47% Emerging Markets (EEM) $66.11 +1.18% Bitcoin (BTC) $69,297.78 +7.14% Ethereum (ETH) $2,251.69 +17.48%

August 2026 · Interim Note

WhiteFiber Fell 21% in a Day, and Estée Lauder Quietly Covered Most of It

August 19, 2026 · Not a full monthly letter — a short check-in between them

Same desk, same rule as Tuesday, and a much stranger day. WhiteFiber fell 21.02% and NBIL fell 19.72%, while Estée Lauder rose 16.30% on an earnings beat and a raised margin target. The net damage to the book was $443 — less than a third of Tuesday’s, on a day with far bigger single-name moves. 12 of 32 positions closed higher. No trades.

−$443Net book impact, vs. −$1,481 Tuesday
+$247Estée Lauder alone
−$186NBIL, the 2× Nebius fund
12 of 32Positions higher on the day

One good position covering three bad ones

Figure 1 · Dollar attribution

Estée Lauder paid for most of the AI-infrastructure damage

+$247 EL +$48 TEM +$46 NOW −$186 NBIL −$114 SNDK −$90 AMD −$80 WYFI −$72 PANW −$243 Other 24 −$443 NET

Share counts × the change in each position’s closing price, August 18 to August 19, 2026. Estée Lauder’s +$247 alone covers NBIL and Nebius combined (−$235), and 76% of the four-name AI-infrastructure loss. Source: Massive Market Data closes, applied to my own share counts.

This is the chart I would have wanted on Tuesday and didn’t have: a day where the percentage headlines and the dollar reality point in opposite directions. WhiteFiber’s 21.02% decline is the number a screen would show, and it cost $80. Estée Lauder’s 16.30% gain is a smaller percentage and worth three times as much, because it is a much larger position. Diversification did its job, and I want to be clear that this was structure working rather than a call I made on the day.

Positions moving more than $15 · August 19, 2026 close
PositionCloseDayBook impactvs. my cost
Estée Lauder Companies (EL)$98.01+16.30%+$247+41.80%
Tempus AI Inc (TEM)$61.25+24.09%+$48+23.17%
ServiceNow Inc (NOW)$127.20+6.45%+$46+50.07%
Fortinet Inc (FTNT)$152.85−3.27%−$15+81.32%
Broadcom Inc (AVGO)$362.48−4.61%−$18+23.52%
CrowdStrike Holdings (CRWD)$201.63−5.30%−$23+112.78%
Space Exploration Technologies (SpaceX) (SPCX)$139.65−2.57%−$30+3.44%
Vertiv Holdings Co (VRT)$261.00−4.23%−$35−14.40%
Datadog Inc (DDOG)$233.52−5.07%−$37+95.79%
Nebius Group N.V. (NBIS)$223.90−9.87%−$49+148.45%
Intel Corporation (INTC)$92.80−4.02%−$54+147.93%
Palo Alto Networks (PANW)$359.76−3.84%−$72+124.40%
WhiteFiber Inc (WYFI)$21.38−21.02%−$80+60.39%
Advanced Micro Devices (AMD)$466.42−3.71%−$90+133.98%
Sandisk Corp (SNDK)$1,568.87−3.50%−$114+224.68%
GraniteShares 2x Long NBIS Daily ETF (NBIL)$25.24−19.72%−$186+173.46%

The rotation didn’t stop — but it did get selective

WhiteFiber (WYFI) daily close vs. SPY, indexed, last ~45 trading sessions through August 19, 2026. Source: Massive Market Data.

The two-day damage, and the cushion underneath it
PositionTueWedTwo-dayTwo-day $vs. my cost
WhiteFiber Inc (WYFI)−10.81%−21.02%−29.56%−$126+60.39%
GraniteShares 2x Long NBIS Daily ETF (NBIL)−15.23%−19.72%−31.95%−$356+173.46%
Nebius Group N.V. (NBIS)−7.60%−9.87%−16.72%−$90+148.45%
CoreWeave Inc (CRWV)−12.10%−2.47%−14.27%−$76+9.56%
Sandisk Corp (SNDK)−9.01%−3.50%−12.20%−$436+224.68%
Vertiv Holdings Co (VRT)−6.80%−4.23%−10.75%−$94−14.40%

The last column is why I did not trade: after the worst two consecutive days this book has had, five of these six positions are still ahead of what I paid, most of them by a wide margin. Vertiv is the exception, and it went underwater on Tuesday, not Wednesday.

WhiteFiber, my smallest AI-infrastructure position, fell 21.02% to $21.38 — a sharper single-day drop than any name in Tuesday’s selloff. I could not find a single dated headline explaining it. It reads as the same debt-financing rotation I wrote up Tuesday, concentrated hardest in the smallest and least liquid name in the group. Notably, CoreWeave fell only 2.47% on Wednesday after Tuesday’s 12.10% — the cohort split rather than falling together, which is the subject of today’s Opinion piece.

How much cushion is actually left

Same measurement as Tuesday’s note, run again after a second bad session — how far each name would have to fall from Wednesday’s close to reach what I paid, and how many repeats of Wednesday that would take.

PositionCloseMy costFall to reach costRepeats of that day
Sandisk Corp (SNDK)$1,568.87$483.21−69.2%33
Nebius Group N.V. (NBIS)$223.90$90.12−59.7%9
GraniteShares 2x Long NBIS Daily ETF (NBIL)$25.24$9.23−63.4%5
WhiteFiber Inc (WYFI)$21.38$13.33−37.7%2
Estée Lauder Companies (EL)$98.01$69.12−29.5%— rose that day
Vertiv Holdings Co (VRT)$261.00$304.90+16.8%— already below

Assumptions, stated plainly. The last column compounds that single session’s percentage move repeatedly until the position reaches my average cost. Nothing about markets works that way — daily returns do not repeat, and a run of that length would long since have triggered a rethink rather than a spreadsheet. It is not a forecast and carries no probability. It is a unit of measurement: it converts “still well ahead of cost” from a reassuring phrase into a countable number of bad days, which is the only form in which that phrase is worth anything. WhiteFiber is the one to watch: two more days like Wednesday and a position that is up 60% on cost is at breakeven. That is what a 21% daily move does to a cushion, and it is the argument for the position being small rather than for selling it.

Estée Lauder, and one number about Nebius

Estée Lauder, a Core position, rose 16.30% to $98.01 after a genuine fiscal fourth-quarter beat — adjusted EPS of $0.39 against a $0.32 estimate, accelerating organic sales for a fourth straight quarter, and a raised operating-margin target range for fiscal 2027. Tempus AI also jumped 24.09% to $61.25 on volume well above normal, though I found no dated news explaining it; I am logging it as unexplained rather than guessing at a cause.

The number underneath Nebius’s decline

Separate from Wednesday’s price action, Nebius disclosed a real improvement in its own contract economics this week: Q2 contract payback fell to roughly 1 year and 10 months, down from a historical 2–3 year range, on contracts averaging over $1 billion each at $20 million-plus of annual contract value per megawatt, with 70% customer prepayments. That is a company-reported operating figure, not a price target — and it is the cleanest evidence I have that this week was a financing-cost repricing rather than a change in the demand picture I bought the stock on.

Bitcoin rose 7.14% and Ethereum rose 17.48% after President Trump met with crypto executives at the White House alongside the SEC and CFTC chairs, on reported optimism about regulatory clarity and the stalled Clarity Act. I hold neither, and Coinbase is not in my book — a sideline observation, not a position update.

No trades today

The Blackout Rule and Holding Period exist for exactly this kind of session. WhiteFiber, Nebius and NBIL are all still well ahead of my cost basis after two rough days in a row. This moved the price, not the thesis. I made no trades on either day.

Sources cited in full on the linked News and Opinion articles. Prices via Massive Market Data; dollar impacts computed from my own share counts. This note is educational and reflects my personal views only. It is not investment advice.