Listen to the latest episode of Credit Exchange with Lisa Lee
Published in London & New York
10 Queen Street Place, London
1345 Avenue of the Americas, New York
Creditflux is an
company
© Creditflux Ltd 2026. All rights reserved. Available by subscription only.
Opinion Credit
What does Elon Musk mean by ‘guppy-level death and consolidation’?
by Duncan Sankey
Duncan Sankey
Partner and head of
credit research
Cheyne Capital
AI small fry may not make it, as happened in the tech bubbles of the 1920s and late 1990s
Elon Musk’s recent tweet that the “200+ neoclouds are going to see guppy-level death and consolidation” had us reaching out to our new best friend, Claude. What did he mean by the gnomic “guppy-level death”? It seems he was referring to the rainbow fish’s capacity to reproduce in huge numbers under the right conditions, only to have most of its offspring perish — in some cases at the hands (well, teeth) of adult guppies.
By analogy, many of the neoclouds and other assorted small fry of the AI ecosystem, hatched in warm waters of easy liquidity, will not make it, especially as credit cools; others will simply be gobbled up by bigger fish. There’s nothing particularly oracular about this — most paradigm-breaking new technologies experience the same fate.
Consider two 20th-century tech bubbles — the car industry in the 1920s and the dotcom sector in the late 1990s — and consider the similarities to our current AI experience.
Both were spawned in a period of monetary accommodation and plentiful liquidity and both benefited from the evolution of a supportive financing ecosystem — equity markets and investment banks in the 1920s and VC, the IPO and HY markets in the 1990s. Both availed themselves of creative financing technology — consumer instalment credit and floorplan financing in the 1920s (GM created GMAC in 1919) and vendor financing for telcos in the 1990s.
AI has thrived among monetary easing
AI investment has also proliferated in a period of unprecedented monetary easing following Covid. It has been able to turn to burgeoning private credit/equity in addition to the public capital markets and is relying on an incestuous network of contingent liability financing — leases, SPVs, vendor financing, residual value support — to simultaneously facilitate growth and obfuscate risk (unsuccessfully, we would argue, given the rating agencies’ capitalisation of contingent and off-balance sheet liabilities).
The only real difference is scale: capital spend as a proportion of GDP topped out at about 3% for the buildout of the US auto sector, 6% for dotcoms, but 8%+ for AI.
There will inevitably be casualties as credit constricts. Rickenbacker, Haynes, Apperson, Velie, Moon, Jordan, Gardner, Elcar and Kissel were all consigned to the museum of car bonnet ornaments in the 1920s and 1930s. Pets.com, Webvan, Urbanfetch, Kozmo.com, theGlobe.com and Pseudo.com lit up computer screens brightly, but all too briefly. AI, too, will have its share of failures as easy credit dries up.
And yet there are success stories, too. Ford and GM consolidated and thrived and are still with us (the latter with a little public assistance). Microsoft, Intel, Cisco, Dell, Amazon, Google, Netflix and Nvidia have also flourished. Survivors typically had two advantages: a focus on cashflow generation and scale.
The latter is important for creditors. It’s a truism, but while equity holders have no defined limit on their potential upside, the best creditors can hope for is to be paid back; risk/reward is asymmetric. Scale usually confers market access and alternative liquidity — that startups do not enjoy — with which to facilitate repayment and debt service.
Synthetic credit of AI giants is compelling
The good news is that value in the synthetic credit of the giants of the AI universe is currently highly compelling, since investors have used CDS to hedge not merely the tsunami of public AI-related debt issuance (nearing USD 500bn YTD), but also private debt and equity exposure across the wider market.
As a result, spreads for five-year protection on double A-rated Meta and Nvidia currently trade around 110bps and 120bps, respectively. As this is a largely technical phenomenon driven by hedging, such value is not evident in those companies’ cash bonds.
Moreover, investors anticipating a rout would be well placed accessing credit exposure synthetically; cash bonds, as funded instruments, would be sold into a falling market to meet redemptions, prompting their material underperformance versus CDS.
In addition, as North American CDS trades “without restructuring”, in the highly unlikely event that an AI major would become embroiled in a liability management exercise/distressed exchange/amend-and-extend, CDS would not face a credit event even though bondholders would be compromised.
So, yes, Elon Musk is right to voice concerns about the guppies. But it’s a good time for credit investors to be long the whales.
