While many investors argue that tech giants’ strong profitability proves there’s no AI bubble, legendary investor Michael Burry has fired back with a sharp rebuttal—drawing eerie parallels to the 1999 dot-com bubble.
Burry’s Counterargument: History Rhymes
Burry points out that the 1999 Nasdaq peak wasn’t driven by money-losing startups, but by high-margin giants—what he calls the “four horsemen”: Microsoft, Intel, Dell, and Cisco.“Contrary to popular belief, it wasn’t unprofitable internet companies fueling the boom—1999’s soaring Nasdaq was propelled into the new millennium by then-high-margin blue chips…”Today, he argues, the AI craze is similarly dominated by “the five public horsemen”—Microsoft, Google, Meta, Amazon, and Oracle—plus startups like OpenAI. These companies plan to spend nearly $3 trillion on AI infrastructure over the next three years.At the center of it all? Nvidia, which Burry compares to Cisco—the dot-com era’s infrastructure provider whose stock plunged over 75% after the bubble burst.“Once again, there’s a Cisco at the heart of it all—one providing the ‘picks and shovels’ with grand visions. Its name is NVIDIA.”
The Real Bubble: Supply-Side Gluttony
In a November 11 teaser, Burry accused tech firms of artificially inflating profits by extending depreciation schedules.
- •AI chips have a real lifespan of 2-3 years, but some companies depreciate them over 6 years.
- •He estimates this could lead to $176 billion in overstated profits by 2028—with Oracle’s earnings potentially inflated by 26.9% and Meta’s by 20.8%.
In his latest piece, “The Primary Signs of a Bubble: Supply-Side Gluttony,” Burry deepens this critique, framing the issue as “catastrophically overbuilt supply and nowhere near enough demand.”
- •Tech giants are engaged in an unsustainable CapEx binge, pouring billions into data centers and chips—but actual downstream revenue can’t justify the costs.
- •He warns investors not to fall for the “this time is different”fallacy:“No matter how many try to prove it, this time isn’t different.”
Nvidia Strikes Back: “No Systemic Fraud”
Facing Burry’s accusations—and broader AI bubble fears—Nvidia fought back aggressively.
- •Memo to analysts denied claims of 610billionincircularfinancing∗∗,clarifyingthatits∗∗Q3strategicinvestmentswerejust3.7B (total YTD: ~$4.7B).
- •Depreciation & margins? Nvidia blamed higher warranty costs (due to Blackwell’s complexity) and insisted its accounting is proper.
- •No SEC investigation, and crypto volatility doesn’t affect its books.
Some analysts back Nvidia:
- •Raymond James’ Simon Leopold called the “systemic fraud”narrative “inconsistent with fundamentals.”
CEO Jensen Huang dismissed bubble talk:“There’s a lot of discussion about an AI bubble… but from our perspective, we see something very different.”
Burry’s New Business Model: “Knowledge Paywall”
Burry published this scathing analysis in his new paid newsletter, Cassandra Unchained($379/year).
- •The first article, “The Primary Signs of a Bubble: Supply-Side Gluttony,”marks his return to full-time commentary after shuttering his fund.
- •In the “About”section, he explained his shift:“Managing client money came with regulatory constraints that ‘effectively muzzled’ my voice. Those limits are gone.”“Cassandra is unchained.”(Referring to the Greek myth of the prophet ignored.)
Fun fact: After media falsely reported he had 912MshortingNvidia/Palantir∗∗(actualstake:∗∗9.2M), Burry quietly closed Scion Asset Management and teased a November 25 launch of his new project.
The Unpopular Prophet?
Burry ends his piece with a Charlie Munger quote:“If you go around popping a lot of balloons, you are not going to be the most popular fellow in the room.
Test the product over more than one session
A first conversation can show personality, but long-term quality appears only after the system has accumulated memory, preferences and notifications. Return later, change a preference and see whether the companion adapts without losing its core identity.
Longitudinal testing also exposes repetition and stale context that a short demo cannot reveal.
Use controls as part of the quality test
Try the memory, notification, privacy and reset controls before you need them. A strong companion should make important settings understandable without forcing the user to delete the entire relationship to repair one problem.
The best experience becomes more relevant over time while remaining easy to correct.