Test report DSG-7678 · Rev C · tested October 10, 2026

Processors & AcceleratorsDevice under test

SpaceX in $40B debt talks for Nvidia AI chips: report

SpaceX is in talks to borrow up to $40 billion to purchase Nvidia AI accelerators, per The Next Web, in one of the largest disclosed AI chip deals to date.

Read
3 min
Words
502
Node
65nm
Operator
Amara Osei

Spec summary

  1. SpaceX is in negotiations to borrow $40 billion, per The Next Web
  2. Funds would be directed at Nvidia AI accelerators
  3. Neither SpaceX nor Nvidia has publicly commented on the talks
  4. At current list pricing, $40B would buy roughly one million top-tier GPUs
  5. Prior SpaceX debt facilities ran in the high hundreds of millions, not tens of billions
SpaceX reportedly in talks to borrow $40B to buy Nvidia AI chips - The Next Web
Fig. ASpaceX reportedly in talks to borrow $40B to buy Nvidia AI chips - The Next Web — AI-generated

SpaceX is negotiating with lenders to secure up to $40 billion in debt financing earmarked for the purchase of Nvidia AI accelerators, according to a report from The Next Web.

The reported figure would rank among the largest debt-financed hardware purchases publicly disclosed for a single artificial intelligence buildout. The proceeds, if the facility closes, would go directly to Nvidia chips — the graphics processing units that power training and operation of large language models and other generative AI workloads.

Neither SpaceX nor Nvidia has commented on the talks. SpaceX has not named the lenders involved, the structure of the proposed facility, or the hardware delivery timeline.

What does $40 billion buy in Nvidia silicon?

Nvidia's top-tier data-center GPUs list at roughly $25,000–$40,000 per unit, depending on the SKU. A $40 billion commitment would translate into a hardware order measured in the high six figures to roughly one million accelerators, before any volume discounts.

Reference prices for Nvidia's current data-center lineup:

  • Blackwell B200 / GB200 — flagship generation, $30,000–$40,000 per GPU
  • H100 — prior-generation workhorse, $25,000–$30,000 per unit
  • H200 — H100 successor with higher memory bandwidth

Analyst consensus forecasts Nvidia's data-center revenue for fiscal 2026 at above $150 billion. A single $40 billion order from one buyer would represent a meaningful share of that total.

Why is SpaceX in the AI chip market?

SpaceX has historically focused on launch services, the Starlink broadband constellation, and the Starship vehicle program. The company has not publicly stated why it would buy AI chips at this scale.

The Next Web ties the order to SpaceX directly. Elon Musk separately operates the social platform X and the AI laboratory xAI, which runs one of the largest publicly disclosed training facilities — the Colossus cluster in Memphis. Neither company has commented on whether the SpaceX-purchased chips would supply those operations.

How does the deal compare to prior SpaceX borrowing?

SpaceX has previously raised debt at much smaller scale:

  • $750 million facility reportedly arranged in 2023
  • $1.5 billion reportedly raised later that year
  • Additional smaller rounds reported in 2024

A $40 billion package would mark a step-change in the company's leverage profile. It would also test the syndicated loan market's appetite for AI-linked paper, a segment that has expanded through 2024 and 2025 as capital allocators chase exposure to the AI buildout.

What is the broader AI-financing context?

Public commitments to AI infrastructure across the largest technology companies have exceeded $300 billion cumulatively for 2025, according to analyst estimates tracked by major investment banks. Major disclosed buyers include Microsoft, Amazon, Google, Meta, and Oracle — each with multi-billion-dollar Nvidia allocations.

A debt-financed $40 billion purchase by a non-hyperscaler would be unusual. Most AI hardware purchases to date have been funded through operating cash flow or equity issuance, not standalone debt facilities.

The Next Web did not name sources in the publicly available portion of its report.

via Google News: AI chip (Source)

Filed under

  • spacex
  • nvidia
  • ai-accelerators
  • debt-financing
  • ai-infrastructure
Share this article:

More from Amara Osei

Amara Osei

Show full bio

Staff writer covering business strategy at Die Signal.

249 articles

Same lot · LOT-C1C6

« Previous articleNext article »