Skip to main content
Open Hunch It

Neocloud selloff shock! Basket returns -35.6%

TL;DR

An equal-weight basket of Neocloud stocks returned -35.6% in the recent short window. The market benchmark barely moved. And the losses inside the group were far from equal.

The whole group fell, and the market did not

Neocloud stocks had an ugly stretch. These are the companies that rent out AI computing power to other businesses. We lined up an of them against a broad market . The basket holds 2 stocks, so it is a small sample, not the whole industry. Over the recent short window the basket returned -35.6%. The benchmark returned -1.5% in the same window. The difference, -34.1%, is the part the group lost on top of the market.

Glossary
equal-weight basket
A pretend portfolio that puts the same amount in each stock, so no single name dominates it. It is a quick way to see how a group traded.
benchmark
A broad market yardstick you compare a stock or a group against. When a stock falls and the benchmark stays flat, something about that stock is doing the work.
AI cloud basket against the market
The basket's last day is also its lowest point of the window, while the market benchmark ended close to where it started.
Data table
dateseriesvalueseries_role
2026-08-10Industry basket0.0%target
2026-08-11Industry basket3.7%target
2026-08-12Industry basket31.5%target
2026-08-13Industry basket29.5%target
2026-08-14Industry basket35.1%target
2026-08-17Industry basket33.1%target
2026-08-18Industry basket20.3%target
2026-08-19Industry basket12.3%target
2026-08-20Industry basket10.7%target
2026-08-21Industry basket9.3%target
2026-08-24Industry basket6.2%target
2026-08-25Industry basket10.2%target
2026-08-26Industry basket8.0%target
2026-08-27Industry basket8.5%target
2026-08-28Industry basket4.6%target
2026-08-31Industry basket4.2%target
2026-09-01Industry basket0.6%target
2026-09-02Industry basket1.3%target
2026-09-03Industry basket5.1%target
2026-09-04Industry basket12.1%target
2026-09-08Industry basket22.8%target
2026-09-09Industry basket19.1%target
2026-09-10Industry basket12.5%target
2026-09-11Industry basket11.4%target
2026-09-14Industry basket4.7%target
2026-09-15Industry basket2.2%target
2026-09-16Industry basket4.1%target
2026-09-17Industry basket4.5%target
2026-09-18Industry basket6.8%target
2026-09-21Industry basket11.7%target
2026-09-22Industry basket13.3%target
2026-09-23Industry basket10.8%target
2026-09-24Industry basket17.2%target
2026-09-25Industry basket14.1%target
2026-09-28Industry basket11.2%target
2026-09-29Industry basket13.2%target
2026-09-30Industry basket13.5%target
2026-10-01Industry basket13.3%target
2026-10-02Industry basket16.8%target
2026-10-05Industry basket12.7%target
2026-10-06Industry basket19.9%target
2026-10-07Industry basket14.6%target
2026-10-08Industry basket5.9%target
2026-10-09Industry basket6.6%target
2026-10-12Industry basket2.6%target
2026-10-13Industry basket0.3%target
2026-10-14Industry basket-12.2%target
2026-10-15Industry basket-9.3%target
2026-10-16Industry basket-20.3%target
2026-10-19Industry basket-19.1%target
2026-10-20Industry basket-18.2%target
2026-10-21Industry basket-18.4%target
2026-10-22Industry basket-25.3%target
2026-10-23Industry basket-26.1%target
2026-08-10Market benchmark0.0%benchmark
2026-08-11Market benchmark-0.3%benchmark
2026-08-12Market benchmark-0.1%benchmark
2026-08-13Market benchmark0.6%benchmark
2026-08-14Market benchmark0.4%benchmark
2026-08-17Market benchmark-0.0%benchmark
2026-08-18Market benchmark-0.7%benchmark
2026-08-19Market benchmark-0.5%benchmark
2026-08-20Market benchmark-1.3%benchmark
2026-08-21Market benchmark-0.9%benchmark
2026-08-24Market benchmark-1.2%benchmark
2026-08-25Market benchmark-0.9%benchmark
2026-08-26Market benchmark-0.9%benchmark
2026-08-27Market benchmark-0.3%benchmark
2026-08-28Market benchmark-0.5%benchmark
2026-08-31Market benchmark-0.8%benchmark
2026-09-01Market benchmark-1.5%benchmark
2026-09-02Market benchmark-1.0%benchmark
2026-09-03Market benchmark0.0%benchmark
2026-09-04Market benchmark-0.4%benchmark
2026-09-08Market benchmark-0.9%benchmark
2026-09-09Market benchmark-1.4%benchmark
2026-09-10Market benchmark-2.0%benchmark
2026-09-11Market benchmark-1.1%benchmark
2026-09-14Market benchmark-1.6%benchmark
2026-09-15Market benchmark-2.0%benchmark
2026-09-16Market benchmark-2.5%benchmark
2026-09-17Market benchmark-1.3%benchmark
2026-09-18Market benchmark-1.2%benchmark
2026-09-21Market benchmark0.3%benchmark
2026-09-22Market benchmark0.3%benchmark
2026-09-23Market benchmark-0.4%benchmark
2026-09-24Market benchmark-0.5%benchmark
2026-09-25Market benchmark0.0%benchmark
2026-09-28Market benchmark-0.7%benchmark
2026-09-29Market benchmark-0.9%benchmark
2026-09-30Market benchmark-1.1%benchmark
2026-10-01Market benchmark-0.9%benchmark
2026-10-02Market benchmark-0.2%benchmark
2026-10-05Market benchmark0.5%benchmark
2026-10-06Market benchmark1.0%benchmark
2026-10-07Market benchmark0.8%benchmark
2026-10-08Market benchmark0.4%benchmark
2026-10-09Market benchmark1.0%benchmark
2026-10-12Market benchmark0.9%benchmark
2026-10-13Market benchmark0.9%benchmark
2026-10-14Market benchmark-0.2%benchmark
2026-10-15Market benchmark0.1%benchmark
2026-10-16Market benchmark-1.0%benchmark
2026-10-19Market benchmark-0.6%benchmark
2026-10-20Market benchmark-0.3%benchmark
2026-10-21Market benchmark0.3%benchmark
2026-10-22Market benchmark-0.9%benchmark
2026-10-23Market benchmark-1.4%benchmark

That gap is the thing to hold on to. The market was close to flat. The group was not. My reading is that when a group moves this far away from the benchmark, investors are repricing something specific to that group, and the talk in this case has been about borrowing. Prices on their own cannot prove that.

Inside the group, the losses were not shared equally

Zoom out to the longer window and the picture softens a little. The basket returned -21.2% over the longer window. The benchmark returned 2.7% over the same stretch. So the group handed back ground while the market inched ahead.

How far did CoreWeave fall over the longer window?

Show answer

-26.6% CoreWeave returned -26.6%. The surprise is the distance to its neighbour: the same headlines hit both names, yet one lost noticeably more ground than the other.

The members did not fall together, though. Nebius Group returned -15.8% over the longer window. CoreWeave returned -26.6%. Same business idea, very different numbers. The spread between members, a cross-member standard deviation, was 5.4% over the longer window, which is a plain way of saying the names drifted apart.

Quick quiz

Over the longer window, which member held up better?

Show answer

Nebius Group Nebius Group returned -15.8% over the longer window, while CoreWeave returned -26.6%.

We also wanted the building side of the story, how much of each company's revenue goes into new data centres, but those company financials are not available here. So this note stays with prices.

Around the news day, both names trailed the market

Now narrow the lens to the days around the news. Abnormal return is the leftover part of a stock's move once you subtract what the market did on the same day. Only 2 stocks are measured here. The median abnormal return from day zero was -6.6%.

Who moved after the news day
Both members trailed the market after day zero, and CoreWeave's shortfall was the larger of the pair.
Data table
namecar_postseries_role
Nebius Group-3.7%peer
CoreWeave-9.5%peer

Nebius Group's abnormal return was -3.7%. CoreWeave's was -9.5%. Both trailed the market in the same window as the debt headlines. That is co-movement and nothing more; no figure here tests whether the news moved the prices. Note too that these price series run through the last available trading day, 23 October 2026, so a later quarterly report is not in them.

On the business itself, Nebius' founder and chief executive told an investor conference that the company uses 「those big contracts to help finance our main part and to build our independent part, our cloud part」.

Risks

  • An equal-weight basket of a few names is not the whole industry.
  • One member can dominate a basket this small.
  • A shared move can come from the whole market, not this group.

Sources

  1. FinMind USStockPrice SPY — data as of 2026-10-23; fetched 2026-10-24 18:20 UTC, 2026-10-24 20:11 UTC; License note: FinMind open data; original source TWSE/TPEx or US exchanges
  2. FinMind USStockPrice CRWV — data as of 2026-10-23; fetched 2026-10-24 18:20 UTC, 2026-10-24 20:11 UTC; License note: FinMind open data; original source TWSE/TPEx or US exchanges
  3. FinMind USStockPrice NBIS — data as of 2026-10-23; fetched 2026-10-24 18:20 UTC, 2026-10-24 20:11 UTC; License note: FinMind open data; original source TWSE/TPEx or US exchanges