Verdict: Mixed

Key finding: The +42% day is driven by a $15M buyback with no funding source named. Management says data center financing and delivery come first. Behind that sits a 45.4M-share resale registration (S-3), which is larger than the 26.4M Class A shares outstanding. The real asset is a 15-year, ~$1.25B take-or-pay lease on a 43 MW Oklahoma site, against a ~$72M basic market cap.

Scorecard

Ticker Rev last Q ($M) Rev YoY last Q Rev YoY prior Q Gross margin (last Q, trend) Op margin EPS last Q vs yr ago P/S Mkt cap 1-mo price %
HOST — (post-merger quarter not yet reported; legacy grocery division) — — — — — — ~$72M basic / ~$126M incl. 19.9M pre-funded warrants (at $2.72) -76.0% (vs $11.33 on 9/17)
APLD 341.9 +432.4% +406.6% 15.3% (down from 42.5% two Qs ago) -18.3% -0.82 vs -0.07 6.8x $5.85B -31.5%
CORZ 164.2 +108.8% +44.9% 42.7% (up) -43.9% -3.32 vs -0.04 10.1x $4.45B -23.7%
IREN 137.2 -26.8% -0.0% 75.7% (up) -452% -1.98 vs +0.69 14.7x $10.43B -40.6%
WULF 44.8 -6.0% -1.1% 72.3% -314% -1.94 vs -0.05 36.5x $6.04B -29.1%

HOST is the pre-revenue laggard of the group. It has no data center revenue yet: the first tenant delivery is targeted for H1 2027. Its 1-month drop (-76%) is about twice that of any peer.
The whole AI-infrastructure group fell 24-41% over the month, so part of HOST's drop is sector-wide. HOST's market cap is small next to its contracted rent (~$83M/yr base) because the financing and the build are both still ahead of it.

Data note: FMP lists HOST at ~$2.5M market cap and 927K shares. That is stale and wrong. The S-3 shows 26,355,685 Class A shares as of Oct 1, 2026.

Move

Catalyst

Sector

AI data center and neocloud names have been in a broad selloff. Over the month APLD fell -31.5%, CORZ -23.7%, IREN -40.6% and WULF -29.1% (Sep 18 to Oct 19 closes). Rates and AI-credit worries hit the group in mid-October. HOST is the smallest and least-funded name in the group, so it fell hardest.

Risks

Judgement

Mixed. The bounce comes after an ~87% collapse in a month. Against $1.25B of contracted rent, some recovery is understandable. But the catalyst named for today cannot carry +42% on its own. The buyback has no stated funding, ranks behind project spending, and comes from a company that still needs project financing. Meanwhile a resale registration larger than the current share count overhangs the stock. The move reads as a short-covering squeeze on a headline (inference). It is not new information about the business. The lease is the thing that matters. Closing project financing, or naming the backstop and tenant, would be real news. This buyback is not.

Sources