Licensed financial lessors

Sale-and-leaseback

You take legal title, not a claim. During the lease the GPU assets belong to the lessor; on default you repossess rather than litigate a mortgage. The assets are purchased and inspectable: 278 eight-GPU chassis / 2,224 cards; value on tax documents. Downstream payment is checkable: MiniMax is in the token-supply commercial-signing process; a formal contract has not been signed.

We hope to close a near-par sale-and-leaseback funding loop. Advance rate, tenor and pricing are yours to quote. We do not disclose our floor.

Value to you

The first value to a lessor is the asset form: legal title sits with you during the lease. It is not a claim that needs a separate mortgage registration. On default you repossess.
The collateral is a data-center GPU chassis with an active secondary market. Serial numbers can be counted; inference is metered in tokens and can be audited. The exit path is clearer than generic IT kit.
The same GPU budget yields a verifiable extra unit of effective capacity: acceleration hardware is our own FX line, not a resale. Signed result: throughput +29–40%, TTFT −26–32% versus a local-NVMe single-drive baseline (reports R2/R3).
The rent repayment source can be locked by you and Xinyuanyi together — after a deal, downstream token-supply / offtake contracts may be co-signed as joint suppliers, so you see the cash-in end. This is a prospective condition; a formal contract has not been signed.
Once the purchased batch is running, about 600 chassis in the prospect pipeline (intent, not held) can reuse the same process. You get a repeatable channel, not a one-off ticket.

Agenda for a first formal meeting

Structure: sale-and-leaseback or direct lease; asset schedule, delivery and acceptance.
Advance rate, tenor, repayment calendar and price — you quote under your policy; we supply a re-runnable model and do not bid our floor first.
Credit support: title, insurance, collection account; whether the offtake is co-signed as joint supply (a formal contract has not been signed).
Lease administration: serial-number counts, token-usage reconciliation, residual value, early termination / purchase option.
Staged drawdown and priority on the mid-term expansion toward the 5,000-card-class plan.

What this means in a term sheet

Purchased: 278 × 8-GPU AMD Instinct MI308X chassis (2,224 cards). Equipment value will be shown on tax documents.
Three checks: asset (serial count), load (token meter), exit (standard chassis, residual path).
After NDA, the full costing model is delivered as runnable Python — you can rerun it with your own parameters.
Downstream token-supply / offtake contracts may be co-signed by you and Xinyuanyi as joint suppliers — a prospective condition; a formal contract has not been signed.

Why we can deliver this

Five reasons this is not another deck. Figures carry a denominator or a report ID. Economic numbers are engineering-model estimates, delivered as runnable Python after NDA.

Purchased kit, countable
278 eight-GPU chassis / 2,224 cards on the books; value on tax documents. About 600 chassis are a prospect pipeline (intent, not held) and are not added into stock.
Acceleration is our line
FX arrays are in production; KV tiering is source-adapted by us, not a resale. Signed throughput +29–40%, TTFT −26–32% (R2/R3 vs local NVMe).
Pre-production acceptance is done
8×MI308X, 1T-parameter INT4 MoE, 3,210 samples across four benches, official delta inside the 2% red line.
Downstream progress is not overstated
Talks and joint tests with Zhipu, Kimi and MiniMax; MiniMax is in token-supply signing, and a formal contract has not been signed. Communication is not collection.
Verify first, stop-loss built in
Joint test first, decisions second: gate-based acceptance with built-in stop-loss. G1 Week 2 / G2 Week 4 / G3 Week 7 / G4 Week 10. Full model after NDA.

On this track specifically

The assets are on the books: 278 × 8-GPU AMD Instinct MI308X chassis (2,224 cards) purchased; value on tax documents; a site count can be booked. This is not a slide-deck order.
Load is not a verbal promise: pre-production acceptance on a single 8×MI308X node for a 1T-parameter INT4 MoE; 3,210 samples across four accuracy benches all passed.
Acceleration is our product line: FX all-flash arrays are in production; KV tiering is source-adapted by us; report IDs are in the evidence library.
Downstream talks are checkable: commercial talks and joint tests with Zhipu, Kimi and MiniMax; MiniMax is in the token-supply signing process, and a formal contract has not been signed.
Diligence follows your committee habit: the full cash-flow model, including adverse cases, is delivered as runnable Python after NDA — you change the parameters.

Next actions and dates

D0 (today)
Tell us your license type and whether you prefer a direct lease or a sale-and-leaseback, via the form or email.
48 hours after NDA
We send the diligence-pack index, the asset-list basis, and the reproducible model.
This week or next
A 30-minute call on structure and lease admin only. No price on that call.
5–10 business days after the call
Site count of the 278 chassis; serial numbers, insurance and exit-path files.
After your internal mandate
Term sheet on your paper; start with a batch, then priority on the mid-term expansion.

Verifiable progress

Purchased: 278 × 8-GPU AMD Instinct MI308X chassis (2,224 cards); equipment value to be shown on tax documents; mid-term plan 5,000-card class.
Pre-production acceptance on a single 8×MI308X node for a 1T-parameter INT4 MoE model: 3,210 samples across four accuracy benches all passed (official delta −0.8% to −1.5%, inside the 2% red line).
Deployment adaptation and performance tests with MiniMax are complete; token supply is in the commercial-signing process; a formal contract has not been signed.
Signed storage-acceleration result: throughput +29–40%, TTFT −26–32% (Qwen3-Coder-480B-FP8 long-context cold recovery vs local NVMe; reports R2/R3).

Evidence library →

Send an inquiry All five tracks

Contact Wang Limin · +86 13161818898 · mingxin@agentmail.to

Other counterparties

This page is a public description of one-to-one B2B cooperation. It does not constitute an offer, an investment offer, or any return promise. Figures are the 2026-09-15 external baseline; economic numbers are engineering-model estimates, not signed purchase or sales prices.