Key Takeaways

  • NScale expanded its pipeline to hundreds of billions of dollars despite starting with almost zero physical capacity.
  • Neoclouds bypass upfront capital expenditure by securing five-year capacity commitments from buyers like Anthropic and OpenAI before purchasing servers.
  • Founders take those signed enterprise agreements directly to Wall Street lenders to raise debt against committed revenue figures as high as $10 billion.
  • Hardware suppliers like Nvidia, Dell, and Super Micro fulfill server orders funded entirely by this debt, leaving neoclouds exposed if chip prices fall.
  • This debt-fueled growth engine forms The NeoCloud Pre-Sold GPU Financing Playbook.

The NeoCloud Pre-Sold GPU Financing Playbook

Specialized cloud providers expand at speeds that look impossible from the outside. Erik Bernhardsson points to NScale as the clearest example: “How can they expand so quickly? The most crazy example is NScale, which went from nowhere to having hundreds of billions of dollars in their pipeline despite barely having any capacity.” The entire model runs on a four-step financial chain.

Step 1: Anchor Offtake Commitment

Negotiate long-term compute commitments with major foundation model labs before touching hardware. Bernhardsson describes the initial pitch: “They go out to Anthropic or OpenAI, some big buyer, and basically say, 'Do you want to buy 5 years of chips? Sign this contract, we give you a good rate.'” Locking in an anchor customer establishes guaranteed future revenue.

Step 2: Debt Financing Against the Signed Contract

Use the signed enterprise contract as revenue collateral to secure institutional debt financing. Founders do not rely on dilution or cash reserves. As Bernhardsson outlines, “They take that contract and then they go to Wall Street and say, 'Look, we got this big buyer. They committed to $10 billion worth of chips. We just need the money right now.' And then they go and raise the money.”

Step 3: Hardware Procurement and Installation

Deploy the borrowed capital to purchase servers and chips directly from suppliers. Bernhardsson traces the handoff: “They go to Super Micro and Dell or Nvidia and get the hardware, and then they install it.” The supplier ships physical racks funded by bank debt rather than equity.

Step 4: Capacity Delivery and Debt Servicing

Turn over the newly commissioned clusters to the contracted customer, creating predictable cash flow to service the loan. Bernhardsson summarizes the outcome: “Now they deliver it to the customer. You have this financial construct where you can basically pre-sell things that you haven't even bought.”

When This Works (and When It Doesn't)

This playbook works when demand outstrips supply and frontier AI labs rush to guarantee multi-year compute blocks. Pre-selling capacity lets founders build massive infrastructure businesses with minimal equity dilution.

It breaks down when market conditions shift. Modal operates differently, focusing on serverless aggregation and product-led growth. As Bernhardsson notes, “Modal, because we've always had a little bit more of a PLG focus and not the ability to lock in long-term contracts in the same way, it's a little bit more challenging from a financial perspective to scale that fast.” If chip performance doubles quickly, hardware prices mean-revert, or utilization drops, operators with fixed balance-sheet commitments face catastrophic tail risk. A firm holding billions in debt on depreciating hardware cannot easily renegotiate with lenders if buyers walk away.

What to Do With This

If you run an AI infrastructure company, test whether you can pre-sell compute capacity before buying server racks this week:

1. Identify two enterprise clients running steady inference workloads. Pitch them a 24-month reserved capacity contract at a 25 percent discount in exchange for guaranteed uptime.

2. Take that signed commitment to an equipment financing firm or venture debt fund. Ask for an asset-backed loan secured against the contract cash flow.

3. Order the exact chip clusters from your hardware distributor only after debt approval clears.

4. Route your client traffic to the machines on day one. Use their monthly contract payments to pay off the principal and interest.