June 8, 2026
Bubble, toil, and trouble
AI Is Slowing Down
Critics say the AI money party is wobbling — commenters are in full meltdown mode
TLDR: The article argues the AI boom needs enormous profits and mountains of funding to survive, or the whole thing could stumble badly. Commenters immediately split into warring camps: some say the warning is valid but overheated, while others mocked the skepticism and insisted AI is already too useful to dismiss.
The article itself is a full-on alarm bell: AI companies, chip makers, and giant cloud firms may need mind-bending amounts of money to keep the boom alive, with the author arguing the whole thing only works if revenue explodes by the end of the decade. In plain English, his claim is simple: this trend can’t just be cool or useful — it has to become wildly profitable at a scale that sounds almost unreal, or the bill could come due in a very ugly way.
But honestly? The real fireworks were in the comments. One camp basically said, “Sure, some of the numbers are scary, but can we please talk about the writer sounding like he’s rooting for disaster?” That was the mood behind several replies, with readers complaining the piece’s angry tone made it harder to trust, even when some arguments seemed solid. Another commenter went full meme mode, comparing the whole debate to arguing that cheese isn’t real — a gloriously absurd way of saying AI is already everywhere, so denial feels pointless.
Then came the split-screen drama: skeptics demanded hard proof, saying we really need public financial filings from companies like OpenAI and Anthropic before declaring either victory or doom. Meanwhile believers pushed the “I use this every day and it saves me time” argument, insisting the doomsayers are missing obvious real-world value. Translation: the crowd is not debating whether AI matters — they’re fighting over whether it’s a revolution, a bubble, or both at the same time.
Key Points
- •The article argues AI infrastructure economics require continued rapid growth and says the sector would need $3 trillion or more in revenue by the end of 2030 to sustain itself.
- •It cites Sightline Climate data showing 190GW of planned data centers and Jensen Huang’s estimate of $80 billion to $100 billion per gigawatt.
- •Based on those figures, the article estimates total planned data center buildout costs at roughly $9.5 trillion to $15 trillion.
- •The article cites a Financial Times report on bank concerns about data center debt and argues annual debt issuance would need to rise from about $250 billion to $500 billion-$1 trillion.
- •It says NVIDIA projected $1 trillion in revenue through 2027 and that 54% of NVIDIA revenue comes from three clients, highlighting concentrated demand among major hyperscalers.