Poolin’s Bankruptcy Sale: What Hut 8’s Terms for Two Texas Sites Show

Poolin’s Bankruptcy Sale: What Hut 8’s Terms for Two Texas Sites Show

Poolin’s Chapter 11 sale of two West Texas sites shows, in a public court docket, what buyers will pay for power capacity once mining stops. The filings tell a more complicated story than the headline that Hut 8 won at $140 million.

Key Takeaways

  • Poolin, a former bitcoin mining pool operator, filed for Chapter 11 on July 22, 2026 and is selling two West Texas sites, Pyote and Tarbush, against about $173.1 million of debts.
  • Hut 8’s affiliates are the designated buyers under amended agreements worth up to $180 million: $100 million at closing and $80 million only if a data-center lease or qualifying transfer follows. The court hearing to approve the sale was set for October 2.
  • Our reading: the terms show buyers paying for secured power capacity, with part of the price tied to AI or HPC demand actually arriving, rather than paying for mining machines.
$173M — Debt Poolin's Chapter 11 filing aims to resolve

What happened

BidderRoleAmount in court filings
Thor CALAP LLCStalking horse, July 22$15M Pyote + $37M Tarbush = $52M; raised to $180M aggregate at the Sept. 10 auction
Hut 8 affiliates (Raptor Basin, Mesquite Falcon)Backup aggregate bidder at auction; designated winner Sept. 22$179M at auction; $140M cash designated; amended Sept. 29 to $100M + up to $80M
Pecos Industrial Development (Fluidstack designee)Tarbush backup bidder$100.5M
DigiPower X Inc.Pyote backup bidder$36.5M

Poolin, once one of the world’s largest bitcoin mining pools, filed for Chapter 11 on July 22, 2026 in the US Bankruptcy Court for the District of New Jersey (Case No. 26-18325). Its Texas mining and hosting operations had stopped on July 10.

The case is a sale and wind-down, not a reorganization (first-day declaration).

Update, October 2, 2026 (docket as of the evening of October 1, Eastern). The sale was not approved on September 29. The court ordered a further open-court auction and sale hearing for October 2 at 10:00 a.m., with bids due October 1 at noon.

The debtors’ October 1 letter says no bids arrived.

How the price moved: $52 million, $180 million, $140 million, then $100 million plus $80 million

The $140 million headline is one step in the docket. Thor CALAP’s stalking-horse offer was $52 million. The September 10 auction, which ran over twelve hours, opened from a $55.48 million aggregate baseline. Thor finished as high aggregate bidder at $180 million, with Hut 8 as backup at $179 million.

The debtors then chose Hut 8. The CRO’s declaration says Thor declined to raise its deposit from 1.5% to 5%, drop “sole and absolute discretion” closing language, or move its closing date. The debtors judged Hut 8’s $140 million cash bid more certain to close.

Thor objected, and that objection was pending on October 1.

On September 29 the agreements changed again, with the creditors’ committee consulted. Hut 8’s affiliates would now pay up to $180 million: $100 million in cash at closing ($74 million Tarbush, $26 million Pyote) and up to $80 million in milestones ($59.2 million Tarbush, $20.8 million Pyote).

Sources: Docket 240, Docket 272, Docket 277.

The claim that Hut 8 bid “nearly three times the $52 million stalking-horse offer” holds for the $140 million cash bid ($140M / $52M = 2.7x). But the auction itself ended at $180 million, and today only $100 million is certain at closing. The other $80 million is conditional.

Most of the roughly $173.1 million owed, about $163.7 million, traces back to IOUs issued after Poolin Wallet suspended withdrawals in September 2022. Around 11,700 wallet holders had balances above $100 when the IOUs were created. Court approval, closing and the milestone payments will determine what creditors recover.

The sale process targeted AI buyers by design. The CRO’s first-day declaration says marketing reached over 335 parties, with emphasis on AI and HPC data-center players, and produced 28 NDAs and seven letters of intent.

Management said it believed selling the energy capabilities to AI and HPC users would yield the highest value.

What the $173 million in debt represents

The debt at the center of this case is not a fresh 2026 liability. It is an obligation dating from the September 2022 withdrawal freeze. The first-day declaration lists $163.7 million of unsecured wallet IOUs, $4.5 million of other unsecured debt and a $4.3 million shareholder loan, among smaller items.

Against the $173.1 million of claims, $100 million of closing cash is 58%, the original $140 million cash bid was 81%, and the full $180 million would be 104%. These are gross ratios, not recovery forecasts: they ignore fees, administrative and priority claims, and the conditional $80 million.

The creditors’ committee asked that no sale proceeds be distributed without further court order. The sale price alone does not say what each IOU holder receives, and the docket gives no distribution estimate.

Two Lenses

Reading one: a legacy failure being cleaned up

The debtors’ own declaration blames China’s crypto ban, bitcoin price volatility, the pandemic, start-up cost overruns and cross-border legal claims. It reports $45.9 million of cumulative losses at the Texas sites since formation.

On this reading the sale says little about AI. A buyer is picking up two sites whose previous owner never received the power allocation it planned for.

Reading two: power capacity is being repriced for AI use

Buyers are paying for grid interconnection, cooling and land, and $80 million of Hut 8’s price depends on data-center leasing. A Fluidstack designee bid $100.5 million for Tarbush alone at the auction.

The evidence is mixed. The Fluidstack designee ended as a Tarbush backup bidder and did not bid in the October 1 round. The winner, Hut 8, describes its platform as covering AI, HPC and ASIC compute, so its win does not isolate AI demand. Source: Irwin declaration.

Why the Texas sites were sold on power, not hash rate

Hut 8’s agreements make power the central closing condition. Pyote requires written confirmation of not less than 54 MW of firm capacity. Tarbush requires approvals for 176 MW at an adjacent property, following ERCOT’s Batch Zero eligibility review, with a possible ramp to 300 MW by December 31, 2029.

Poolin’s own history points the same way. The first-day declaration says early talks with the utility suggested upwards of 600 MW, but only 100 MW was initially allocated. Poolin had over-ordered mining equipment, and later reported $8.8 million of equipment-sale losses between fiscal 2023 and 2025.

The asset schedules attached to the sale agreements list mainly land, buildings, cooling systems, transformers and electrical equipment. ASIC miners are fixed-function chips for bitcoin hashing and cannot run AI models, so they are not what AI or HPC tenants need. Source: Docket 272, Exhibits A and B.

What the price implies per megawatt

Dividing Hut 8’s price by the megawatts named in its closing conditions gives a rough all-in price per MW. The denominator is 54 MW at Pyote plus 176 MW at Tarbush, or 230 MW. Prices include land, buildings and equipment, so this is not a price for power alone.

Hut 8 price scenarioAmountPer MW (230 MW)
Original all-cash APAs (Sept. 21)$140.0M$0.61M
Amended: cash at closing$100.0M$0.43M
Amended: maximum with milestones$180.0M$0.78M

By site, closing cash is $0.48M per MW at Pyote ($26M / 54 MW) and $0.42M per MW at Tarbush ($74M / 176 MW). The maximum is $0.87M and $0.76M. Method: price divided by contractual MW, rounded to two decimals.

Treat these as ranges, not market prices. The 54 MW is a minimum, and the 176 MW depends on utility and ERCOT approvals.

The first-day declaration says only 100 MW was initially allocated across the mining sites, so a different denominator moves every figure. No sourced per-MW price for a comparable sale is available here, so none is compared.

Why it matters

For site owners: in these agreements the utility approval, not the mining history, is the gating item. A site with unconfirmed power is priced and closed differently from one with confirmed load.

For AI developers: a converted mining site is not instant capacity. Closing is set for December 15, 2026, tied to an ERCOT approval expected around December 10, and at Tarbush the buyer bears the electrical build costs. Our Ohio piece covers how one state allocates grid-upgrade costs.

For creditors: the gross price is not a recovery rate. Fees, administrative claims and the milestone condition sit between the headline number and any distribution.

A five-point checklist for reading a distressed mining-site sale

These questions come from the terms in this case and can be applied to similar sales:

  1. Cash or conditional? Separate money paid at closing from payments that depend on later events.
  2. Which megawatts? Check whether capacity is confirmed, minimum or expected, and who must approve it.
  3. What counts as a data center? Poolin’s milestone definition covers AI, HPC, cloud, hosting or colocation, and excludes facilities operated mainly for bitcoin mining.
  4. Who bears build costs? At Tarbush the buyer pays substation, interconnection and other customer-side costs.
  5. How certain is closing? Look at deposits, termination rights and outside dates, not only the headline price.
Poolin's assets, from freeze to auction

What to watch next: the October 2 sale hearing; closing, set for December 15, 2026 and extendable by the sellers to as late as March 15, 2027; and whether a data-center lease or qualifying transfer triggers the $80 million milestone payments.

Hut 8 began as a bitcoin miner and now also builds data-center and AI infrastructure, so the winner does not cleanly settle the miner-versus-AI question.

Limits of this case

This is one bankruptcy with two sites, a sale run on a liquidation timetable, and a price that includes land, buildings and equipment. One docket cannot show a market-wide repricing of mining sites. The per-MW figures rest on contractual closing thresholds, not measured operating capacity.

The choice of Hut 8 over a higher auction bid is contested by Thor, and the court had not ruled when the docket was last checked. This post analyzes public filings and is not investment or legal advice.

FAQ

Q. Does Poolin’s bankruptcy affect bitcoin’s price or network security?

A. Nothing in the filings points to a direct effect. Poolin’s Texas mining and hosting operations stopped on July 10, 2026, and the filings do not quantify their hash rate, so we have no sourced estimate of any network impact. This is mainly a creditor and asset-sale event.

Q. Can bitcoin mining hardware actually be reused for AI computing?

A. No — ASIC miners are fixed-function chips for bitcoin hashing and cannot run AI workloads. What matters to AI buyers is the surrounding infrastructure: power rights, substations, cooling systems and land, which is also what the sale agreements’ asset schedules mostly list.

Q. Is $140 million the final price for Poolin’s Texas sites?

A. Not as the filings now stand. The $140 million all-cash structure was replaced on September 29 by up to $180 million: $100 million at closing and up to $80 million only if a data-center lease or qualifying transfer occurs. The sale still needed court approval on October 2.

What would change our view

Our reading that buyers are pricing power for AI use would weaken if the $80 million milestone is never triggered, if the court approves a different buyer, or if later distressed mining-site sales go mainly to bitcoin miners at similar per-MW prices.

Sources

Correction, October 2, 2026. An earlier version said Poolin’s Chapter 11 would take years to resolve and that the auction winner was still unknown.

The case moved from a July 22 filing to a September 10 auction and a sale hearing reset to October 2. Hut 8 was designated the winning bidder at $140 million, and its agreements were amended on September 29 to up to $180 million.

This post was expanded on October 2 using the court docket.

The post also described AI buyers’ interest more strongly than the sources support, and an unsourced hash-rate claim was removed.

Related from 2mind

Comments

Leave a Reply

Your email address will not be published. Required fields are marked *