The artificial intelligence revolution has reached a fever pitch that should make anyone who lived through the dotcom bubble extremely nervous. With $1.6 trillion in projected spending by 2031 and the “Magnificent Seven” tech giants now comprising nearly half of the S&P 500’s market value, we’re witnessing an unprecedented concentration of capital chasing returns that remain largely theoretical.
The Numbers Tell a Stark Story
The scale of AI investment defies historical precedent. Goldman Sachs projects spending will more than double from $765 billion this year to $1.6 trillion by 2031. To put this in perspective, that’s roughly equivalent to the entire GDP of Canada being poured into AI infrastructure, chips, and datacenters over the next seven years.
This isn’t gradual adoption—this is a financial stampede.
The market concentration is equally stunning. Alphabet, Amazon, Apple, Meta, Microsoft, Nvidia, and Tesla—the so-called Magnificent Seven—have driven the S&P 500 up nearly 80% over five years. According to Bianco Research, 41 AI-related stocks now account for nearly half the index’s market value.
“The entire market has become one giant AI edifice. The danger is a repeat of the dotcom bubble – a huge crash, and years of lost returns.” — @cnfinancewatch
The Dotcom Parallel: History Doesn’t Repeat, But It Rhymes
The parallels to 1999-2000 are impossible to ignore. During the dotcom boom, internet stocks commanded astronomical valuations based on “eyeballs” and “mindshare” rather than profits. Companies like Pets.com and Webvan burned through billions while promising to revolutionize commerce.
Today’s AI mania shows similar characteristics:
- Massive capital commitment with uncertain returns
- Concentration risk in a handful of tech giants
- Adoption metrics (like ChatGPT’s 1 billion monthly users) treated as revenue proxies
- Token costs rising even as companies push “tokenmaxx” strategies
The key difference? The dotcom bubble was built on dial-up infrastructure that couldn’t support the promised applications. Today’s AI boom faces the opposite problem—demand that may outstrip physical infrastructure capacity.
The Infrastructure Reality Check
JLL predicts 100 gigawatts of datacenter capacity will be added between 2026-2030, equivalent to 1,200 datacenters. That’s doubling current global capacity in four years. The engineering and financial logistics are staggering:
- Power grid expansion requiring government coordination
- Land acquisition in suitable locations
- Skilled workforce for construction and operation
- Environmental impact assessments and approvals
Cecilia Rikap from University College London raises the critical question: “Has the government calculated whether such an expansion is feasible? Do they have the money to do it?”
If even modest delays occur in datacenter construction, Goldman Sachs warns it will “invite real scrutiny around the demand assumptions” justifying these investments.
The Productivity Paradox
Here’s where the AI bubble diverges most dangerously from historical precedent. McKinsey reports 80% of companies now use AI, up from 33% in 2023. OpenAI’s ChatGPT has reached 1 billion monthly users—faster adoption than any app in history.
Yet productivity gains remain elusive. Companies are encouraging “tokenmaxx” usage while token costs spiral upward. Liam Betsworth from AI startup Pendra reports costs are “getting completely out of control,” with developers rapidly maxing out expensive subscription tiers.
The fundamental promise—that AI spending generates more than proportional productivity gains—remains unproven at scale. Anthropic’s research shows AI “could perform a host of jobs from computing to legal work, but has yet to do so in any great force.”
The Competition Intensifies
Anthropic’s Claude is rapidly gaining ground on ChatGPT, with Kentik projecting it could overtake OpenAI’s flagship by summer. This competition should theoretically benefit users through innovation and pricing pressure.
Instead, we’re seeing the opposite. AI companies aren’t charging enough to cover infrastructure costs, while users face rising token prices. It’s a classic unsustainable growth model reminiscent of ride-sharing companies burning investor cash to subsidize rides.
Key Warning Signs
Several indicators suggest we’re approaching bubble territory:
- Valuation concentration: Nearly half the S&P 500’s value in AI-related stocks
- Infrastructure bottlenecks: Datacenter construction may not meet demand
- Rising costs: Token pricing increasing despite scale economics
- Productivity lag: Massive adoption without commensurate economic impact
- Speculative projects: $500 million monthly Claude licensing deals
- Government dependency: Infrastructure expansion requiring political commitments
The Path Forward
Unlike the dotcom crash, an AI correction wouldn’t necessarily invalidate the underlying technology. The internet survived the 2000-2002 crash and eventually delivered on many early promises. Amazon and Google emerged stronger after their stock prices collapsed 80-90%.
But timing matters enormously for investors and the broader economy. Neil Wilson from Saxo UK warns of “a huge crash, and years of lost returns” if current valuations prove unsustainable.
The AI revolution is real, but revolutions rarely follow smooth, predictable paths. Smart money prepares for volatility while maintaining long-term conviction in transformative technologies. The question isn’t whether AI will reshape the economy—it’s whether today’s prices reflect realistic timelines and returns.
With $1.6 trillion in commitments chasing hypothetical productivity gains, we’re about to find out if this time is truly different—or if the laws of financial gravity still apply.
Published in Stream · Dispatch #429 · June 7, 2026 · 4 min read.
Reply to paolo@mont3.ch - every email gets a human answer within 24h.