Is AI About to Pull a 2008 on Us?

Okay, I need to talk about something that's been keeping me up at night. And before you say "Jo, this isn't your usual AI-meets-shopping content," hear me out. I generally write about very niche topics specifically around the intersection of AI in e-commerce and/or marketing, but I wanted to break down what's happening in terms of the wider economic piece as it will affect consumer spending, the economy, and of course the e-commerce operators reading this newsletter.

Because here's the thing: when the guy who predicted the 2008 financial crisis drops a billion-dollar bet against AI stocks, we should probably pay attention.

The Big Short 2: Electric Boogaloo

Michael Burry just went public with short positions against Nvidia and Palantir, and the market immediately had what I can only describe as a collective panic attack. AI stocks dropped 10-20% faster than my motivation to exercise on a Monday morning.

But wait, it gets better (or worse, depending on your portfolio). Right as Burry made his move, OpenAI's CFO Sarah Friar casually floated the idea of the US government guaranteeing their debt at some California event. Sam Altman had to scramble to deny it faster than a teenager caught with their parent's credit card.

Government. Debt. Guarantees. For an AI company.

[Stress-eating leftover Halloween candy while typing this]

If that doesn't give you 2008 flashbacks, let me paint you the full picture.

The Numbers Are Genuinely Insane (And Not in a Good Way)

UK analyst Julien Garran dropped some numbers that made me literally spit out my coffee (RIP keyboard #47):

• This AI bubble? 17 times bigger than the dot-com crash

• 4 times bigger than the 2008 housing crisis

• 10 AI startups with ZERO profits have gained nearly $1 trillion in market value

• OpenAI is spending $3 for every $1 it earns

I had to read that last one three times. They're burning cash at a rate that would make a Vegas high-roller blush. We're talking potential $20 billion annual losses here.

And it's not just the startups. Microsoft's cloud margins are getting crushed. Alphabet committed $85 billion in capital expenditure for 2025. Meta just secured $30 billion in off-balance-sheet debt (the largest deal of its kind ever, because of course it is).

"But Jo, AI Is Revolutionary!"

Look, I get it. I don’t write about AI 3x a week for nothin’. The bulls have a point. Marc Andreessen calls it "computer industry version two." ChatGPT hit 800 million users faster than any technology in history. Jensen Huang from Nvidia says we need to replace literally every computer on Earth with AI infrastructure, creating a $10 trillion annual market.

[Currently at 2 AM wondering if I should buy more Nvidia stock or short it]

The thesis is seductive: AI is the first fundamental reinvention of computing in 80 years. It'll unlock productivity gains comparable to the steam engine. We're witnessing the birth of a new era of human civilization, etc., etc.

But then I look at the actual numbers and... yikes.

The Unit Economics Are Absolutely Cooked

Ed Zitron uncovered what might be the most circular funding scheme since... well, 2008. Here's the simplified version:

1. Nvidia invests in data center operators

2. Those operators use the money (plus massive debt) to buy Nvidia's chips

3. Nvidia becomes their biggest customer by renting back the capacity

4. Everyone pretends this is normal business

Outside of Big Tech and OpenAI, these data center operators have less than $1 billion in combined REAL revenue. Combined!

Here's my favorite stat that made me stress-order unnecessary kitchen gadgets at 3 AM: For a data center to make a reasonable 10% return on Nvidia's new Blackwell chip, they'd need to rent it for $6.31 per hour. Current market rate? $3.79 per hour. That's a guaranteed 25% loss on every transaction.

The only one making money in this entire ecosystem is... drumroll please... Nvidia. Everyone else is bleeding capital like they're trying to win some twisted Silicon Valley endurance contest.

The Consumer Economy Is Already Wobbling

Here's where it gets scary for us e-commerce folks. The University of Michigan's consumer sentiment just hit 50.3—near a three-and-a-half-year low. 62% of households expect unemployment to rise, the highest since 1980.

But here's the weird part: it's a K-shaped recovery on steroids. Rich people with stocks? Feeling great! Everyone else? Not so much.

The top 20% of households drive 40% of consumer spending. If AI stocks crater, that wealth effect reverses overnight. Your premium product sales? Gone. Your luxury category growth? Evaporated. Meanwhile, lower-income consumers are already maxed out.

We could see spending compression across ALL price points simultaneously. That's... that's not a scenario I've planned for in my disaster prep spreadsheets (yes, I have disaster prep spreadsheets, don't judge).

What This Actually Means for Your E-commerce Business

Let me get practical for a second, because I know you're sitting there thinking "Cool story Jo, but what do I DO with this information?"

First, about those AI tools you're using: Many are operating at subsidized rates that are about as sustainable as my New Year's resolution to quit coffee. If you're heavily dependent on AI customer service, content generation, or analytics tools, maybe have a Plan B. These prices could 10x overnight when the venture capital music stops.

Second, consumer behavior: Start scenario planning NOW. What happens if your high-value customers suddenly feel 30% poorer? What if your budget customers get hit with a recession? Can your unit economics survive a simultaneous compression across all customer segments? (If you just said "yes" without checking, go check. I'll wait.)

The Three Ways This Could Still Work Out

Because I'm not a complete doomer (despite what this newsletter might suggest), here are the scenarios where the bulls end up being right:

1. Inference costs collapse by 90%+: The price to achieve fixed AI performance is falling between 9x and 900x per year. If this continues, current business models suddenly work.

2. The killer app emerges: Something that justifies way higher prices. Maybe something that actually replaces the $2 trillion global market for developer wages.

3. Architectural breakthrough: Someone figures out how to sidestep the "scaling wall" without exponential cost increases.

The probability of one of these happening before the money runs out? [Shrugs while eating stress chocolate]

OpenAI's Trillion-Dollar Math Problem

Let me leave you with this nugget that made me question reality: OpenAI needs over $1 trillion in the next four years. That's:

• $300 billion committed to Oracle

• $325 billion to build data centers

• $115 billion in operational costs

The top 10 private equity firms combined have roughly $477 billion available.

The math literally doesn't math.

The Bottom Line: It's Complicated (As Always)

Look, I think we're seeing both things at once: a genuine technological revolution being financed by an unsustainable bubble. The technology IS useful—800 million people don't use garbage products. But can the value delivered justify burning hundreds of billions in capital?

The current structure where only Nvidia makes money while everyone else loses it? That's got an expiration date, and it might be sooner than we think.

For us e-commerce operators, the prudent move is to prepare for multiple scenarios. Keep using AI where it makes sense, but don't bet the farm on tools that might not exist (or might cost 10x more) next year. Watch your customer segments like a hawk. And maybe, just maybe, keep some powder dry for picking up the pieces if this whole thing goes sideways.

Because if there's one thing I've learned from watching tech cycles, it's that the infrastructure survives even when the bubble doesn't. The dot-com crash left us cheap fiber optic cables that enabled the next revolution. An AI crash might leave us computational resources at fire-sale prices.

The winners won't be the ones who predicted the bubble. They'll be the ones who prepared for what comes after.

P.S. Burry has been mostly silent for YEARS. The fact that he's going public with a $1.1 billion bet against AI stocks? That's not just a trade. That's a warning siren.

P.P.S. If you're reading this and thinking "Jo's being dramatic," please note that Meta's stock dropped 20% after announcing they're doubling down on AI spending. Twenty. Percent. That's not normal market behavior, that's the market having an existential crisis.

P.P.P.S. Former FCA chair Adair Turner says banking regulations should prevent a 2008-style collapse. Which is reassuring, except that's exactly the kind of thing people say right before discovering a new way for everything to collapse.

• AI Overviews and GPT style search pull from long tail sites that never rank on page one, compress answers and sometimes miss fresh nuance, rewiring how you think about authority and SEO.

• Semrush AI Visibility Index shows evergreen hubs like Wikipedia are losing value while additive UGC and data rich stories win

• The 7 Phases of the Internet A map to where the Web goes next

• If you can’t afford a vacation, an AI app will sell you pictures of one

• You cannot fully automate a content audit yet, but wiring crawlers analytics and Claude through MCP turns AI into a sharp audit assistant that surfaces underused pages and fast win improvements.

• Ahrefs SEO pricing study racks up LLM citations simply by being deep structured original research, proof that timeless best in class pages remain the easiest path to durable AI visibility.

• Get inspired with what people are actually putting their Claude skills up to!

  

• Sora 2 now lets you turn pets toys or stylized personas into reusable video cameos and stitched clips

• Scales Remote Labor Index finds top AI agents can fully complete only about 2.5 percent of paid freelance jobs

• Shopify is seeing AI driven traffic and purchases surge 

• And on the flip side, we have this. Everyone has something to say about AI traffic lol.

• Snap and Perplexity just signed a 400 million deal to bake AI search into Snapchat chat

• OpenAI readies GPT-5.1 Thinking model ahead of Gemini 3 Pro

📳 Papermark AI: Revolutionize your document sharing and collaboration. Share pitch decks, sales proposals and other docs securely with real-time analytics and white-labeling options.

🤖 ChatHub - Engage with six leading chatbots, including GPT-4 & Claude, simultaneously on a single screen.

📝 Synclabs - Translate any video to any language, with perfectly matched lip movements.

📹 Video interviews by Outset - Conduct user surveys with video and AI.

🎙️ Narrativ AI - A marketplace for advertisers to buy talent likeness.

🛠️ Spiritime - Create personalized videos with digital avatars.

😍It was great, all I needed

🙂Good, not great

🙄It sucked