# SpaceX borrows $40 billion to buy AI chips as Musk predicts failures among smaller cloud rivals

The debt would more than double SpaceX's borrowing from the end of 2025, and the scramble for chips is already pushing up phone prices in markets such as Pakistan.

- Published: 2026-10-26T05:15:30.644Z
- Canonical: https://polylog.news/2026-10-26/spacex-borrows-40-billion-to-buy-ai-chips-as-musk-predicts-f
- Publisher: Polylog (Global desk)
- Section: tech
- Sources: [CoinDesk](https://www.coindesk.com/markets/2026/10/26/elon-musk-confirms-usd40-billion-spacex-debt-deal-for-ai-chips-warns-of-neocloud-shakeout), [MarketWatch](https://www.marketwatch.com/story/musk-says-spacex-landed-a-40-billion-debt-deal-and-forecasts-doom-for-some-ai-rivals-a0a4991e?mod=mw_rss_topstories), [The Japan Times (Business)](https://www.japantimes.co.jp/business/2026/10/26/softbank-malaysia-data-center-investments/), [Dawn](https://www.dawn.com/news/2039859/sharp-rise-in-prices-of-smartphones-burdens-consumers-in-karachi)

Elon Musk said SpaceX has [secured a $40 billion debt deal](https://www.marketwatch.com/story/musk-says-spacex-landed-a-40-billion-debt-deal-and-forecasts-doom-for-some-ai-rivals-a0a4991e?mod=mw_rss_topstories) to fund its fast-growing business of renting out access to advanced chips. [CoinDesk reported](https://www.coindesk.com/markets/2026/10/26/elon-musk-confirms-usd40-billion-spacex-debt-deal-for-ai-chips-warns-of-neocloud-shakeout) that the money will buy graphics processors and that Musk predicted a major shakeout among smaller AI cloud providers, specialist firms that rent out chips and are known in the industry as "neoclouds." The Financial Times reported earlier this month that Apollo Global Management is leading the package, split into [about $10 billion of bank loans and $30 billion of investment-grade debt](https://investinglive.com/stocks/spacex-seeks-40-billion-in-debt-led-by-apollo-to-fund-nvidia-chip-order-ft-says/). SpaceX ended 2025 with [$24.5 billion of total debt](https://www.tikr.com/blog/spacex-the-largest-global-neocloud-per-jim-chanos-as-we-gear-up-for-earnings-tomorrow), so the new borrowing would more than double it.

Asia is seeing the same rush to build. Malaysian Prime Minister Anwar Ibrahim said SoftBank founder Masayoshi Son [plans a $50 billion technology investment in Malaysia](https://www.japantimes.co.jp/business/2026/10/26/softbank-malaysia-data-center-investments/), according to The Japan Times. Southeast Asia is becoming a hub for data centers and other AI infrastructure.

Consumers are paying part of the cost. In Karachi, [Dawn reported](https://www.dawn.com/news/2039859/sharp-rise-in-prices-of-smartphones-burdens-consumers-in-karachi) that the average price of a phone with 4 gigabytes of memory and 64 gigabytes of storage has risen to about 40,000 rupees from 22,000 rupees a few months ago. Dealers told the paper that government taxes have not risen much and that higher chip costs explain the increase.

## What this means

Credit is financing a large, concentrated purchase of a single asset class, AI chips, whose rental value depends on demand that has not yet been proven. Consumer electronics buyers in poorer markets absorb the cost through higher memory prices. Lenders, including private-credit funds, take on the duration risk. This is the pattern Austrian business-cycle theory describes: cheap or abundant credit pulls resources into long-dated capital projects ahead of real consumer demand. When credit tightens, it ends in consolidation. Musk's forecast of a shakeout is consistent with that, and it also serves his interest, because SpaceX gains if leveraged rivals fail and their chips sell cheaply. The precedent is the telecom fiber buildout of 1998 to 2001, when debt-financed capacity ran ahead of traffic and the weakest carriers failed while the strongest bought assets. The likelier outcome is consolidation among smaller GPU-rental firms within months if the Fed stays restrictive and credit spreads widen. Larger buyers with cheaper funding would come out stronger. Sustained growth in paid AI usage that keeps rental rates high would prove this wrong.

## What to watch

- Rental prices for high-end AI chips. Falling rates while capacity keeps growing would be the earliest sign that the predicted shakeout among smaller providers has started.
- Credit spreads on debt issued by data-center and AI chip-rental companies. Widening spreads would show lenders repricing the risk Musk described.
- Memory and smartphone prices in import-dependent emerging markets. Further increases would show the compute buildout is moving costs onto consumers outside the AI economy.