☁️

CIFR

πŸ’‘
Thesis. A datacentre landlord that never buys a GPU β€” roughly $11.4B contracted across 10–15 year leases with Google backstopping one tenant and AWS signed to another β€” now finally converting, with Black Pearl rent commencing two months early, but carrying $5.45B of long-term borrowings against barely any revenue.
Reviewed 19 Sep 2026, after Q2 2026 results on 4 Aug and the August Black Pearl rent commencement. Next review after Q3 2026 earnings on 5 Nov 2026.

What it does

Cipher is a datacentre landlord, not a GPU cloud. It builds powered buildings and leases them on 10–15 year terms; the tenant brings their own GPUs. No chip-obsolescence risk, no utilisation risk, much longer contracts β€” and a lower ceiling on returns. Structurally closer to a REIT than to
☁️
CRWV
. The company now trades as Cipher Digital. Legacy Bitcoin mining is what still produces most of the reported revenue.
A backstop is the mechanic that makes the Fluidstack lease investment-grade: Google has guaranteed $1.73B of Fluidstack's payment obligations, so Cipher's credit exposure is effectively to Google, not to a venture-funded startup.
The gap between $11.4B signed and $24.8M collected is not a warning sign by itself β€” it is what a landlord's income statement looks like while the buildings are still going up. The question is only whether they finish on time.

Bull case

  • ~$11.4B of contracted revenue across 10–15 year base lease terms, 907 MW operating and contracted, three signed campus leases with hyperscale tenants
  • Fluidstack: 168 MW / 10-year / ~$3B, plus a further 56 MW at ~$830M. Google backstops $1.73B of Fluidstack's obligations and holds ~5.4% of Cipher
  • Amazon Web Services: a 15-year datacentre campus lease
  • Black Pearl is delivering. Rent commenced at the start of August, two months ahead of schedule, with the tenant in occupation
  • Barber Lake: tenant has begun beneficial use, including partial building occupancy and network rack deployment
  • Stingray fully funded to substantial completion by an $810M bond offering, which also reimbursed $56.7M of prior spend
  • Apollo option acquired near San Antonio: ~288 acres supporting up to 900 MW. ~3.3 GW pipeline through 2029, including ~95% ownership of the 1 GW Colchis site
  • No GPU obsolescence risk whatsoever β€” the biggest structural advantage over the GPU operators here
  • Cash $831.8M at 30 June

Bear case

  • Q2 2026 revenue only $24.8M, all Bitcoin mining, down 43% YoY. Net loss $267.5M (βˆ’$0.65/sh). Adjusted EBITDA βˆ’$30.0M
  • Long-term borrowings $5.45B against $562.1M of stockholders' equity on $7.5B of total assets. The leverage taken on to build is now the dominant fact on this balance sheet
  • The market is valuing backlog, not income β€” which cuts both ways, and the stock has swung violently on prints in both directions
  • Construction and power-delivery risk is total. Everything depends on delivering buildings and energised capacity on schedule
  • Still carries bitcoin-price correlation from legacy mining, so the chart reflects crypto as well as AI
  • Convertible and bond financing creates future dilution and fixed obligations regardless of lease-up
  • Long lease terms lock in revenue and also lock out repricing if the market for powered capacity tightens further

Major customers

  • Fluidstack β€” 168 MW / 10-year / ~$3B, plus 56 MW / ~$830M. Backed by Google, which backstops $1.73B of the obligations and holds ~5.4% of Cipher
  • Amazon Web Services β€” 15-year datacentre campus lease
  • A third unnamed hyperscale campus tenant
  • Legacy: Bitcoin mining self-operations, still the source of reported revenue
πŸ”—
Read-through. Cipher and
☁️
KEEL
are the two landlords in this category and will not move with
☁️
CRWV
,
☁️
NBIS
or
☁️
IREN
on GPU-cycle news, because neither buys GPUs. Cipher is what Keel is trying to become β€” same model, but Cipher converted its pipeline into $11.4B of contracts with Google- and Amazon-grade counterparties while Keel has signed nothing. The Black Pearl rent commencement is the first hard evidence that a shell-landlord can actually hit its dates, which is the read-across that matters for Keel and for
☁️
WYFI
's NC-1. Its buildouts drive orders for
❄️
VRT
,
❄️
NVT
and the Grid Equipment names.

What would change the view

  1. Barber Lake, Stingray and Colchis delivery on schedule β€” construction execution is the whole risk, and Black Pearl set a favourable precedent
  2. Revenue beginning to convert visibly from the $11.4B contracted base, quarter over quarter
  3. Debt service against $5.45B of long-term borrowings as lease income ramps
  4. Any additional campus leases signed, and whether the Apollo option is exercised
  5. Bitcoin exposure as a share of the business β€” whether the crypto correlation fades
  6. Whether Google increases its stake or extends the backstop

Update log

19 Sep 2026 β€” Converted to the standard template and brought current through Q2 2026, which the page had been missing; the prior version still quoted Q1. Added the Black Pearl early rent commencement, the $810M Stingray bond, the Apollo option (up to 900 MW), and the balance sheet, where long-term borrowings of $5.45B are now the dominant bear point rather than construction risk alone. Company now trades as Cipher Digital. Backlog and Dominant Risk properties updated.
Aug 2026 β€” Black Pearl rent commenced, two months ahead of the original schedule.
4 Aug 2026 β€” Q2 2026 reported. Revenue $24.8M (mining, βˆ’43% YoY), net loss $267.5M (βˆ’$0.65/sh), adjusted EBITDA βˆ’$30.0M. Total assets $7.5B, long-term borrowings $5.45B, equity $562.1M, cash $831.8M. Stingray funded by an $810M bond offering.

Research and education only β€” not investment advice.