Lightnews — Scholar-powered news

Matthew Hughes

@matthewhughes.bsky.social

1.7K followers 1.6K following 1K posts

Journalist (ex The Register, The Next Web, HowToGeek). Writer. Software developer. Dog owner (x3). Scouser. I have a newsletter about how tech companies are ruining our lives. https://whatwelost.substack.com/

whatwelost.substack.com

Posts Media Videos Starter Packs

Pinned

Matthew Hughes @matthewhughes.bsky.social · 3d

My latest newsletter is about how generative AI doesn't just need money to survive, but faith.

Every time we see another big genAI foul-up, that faith will waver, and its end becomes all the more likely.

whatwelost.substack.com/p/generative...

Generative AI’s Impending Death By A Thousand Rake-Smacks

Give it enough rope...

whatwelost.substack.com

5 40 260

Matthew Hughes @matthewhughes.bsky.social · 1h

Given that it can’t count the number of letters in a word… I commend your optimism.

1 1

Matthew Hughes @matthewhughes.bsky.social · 3h

The math doesn’t math.

whatwelost.substack.com/p/generative...

Generative AI’s Impending Death By A Thousand Rake-Smacks

Give it enough rope...

whatwelost.substack.com

1 5

Matthew Hughes @matthewhughes.bsky.social · 3h

Assuming OpenAI’s obligations to Oracle are spread equally each month (they aren’t, but for the sake of argument), it would need to pay $5bn a month, each month, for five years.

Assuming subscriptions remain at 70% of revenue, that means making the same amount of money as Netflix, but from ChatGPT.

1 1 6

Reposted by Matthew Hughes

Matthew Terrill @matthewterrill.bsky.social · 3h

I was already a hard AI-skeptic but this cements my long suspicion that there is no feasible path to anything close to return on invested capital for these data centers. Tech would need 15 to 25 times current AI revenues within the next 2-3 years just to break even. Not financially viable.

"I clearly hit a nerve in the industry, when judging by the number of individuals who reached out to chat," he wrote in an followup blog post. "In total, l've spoken with over two-dozen rather senior people in the datacenter universe, and there was an interesting and overriding theme to our conversations: no one understands how the financial math is supposed to work. They are as baffled as I am, and they do this for a living."
Kupperman's original skepticism was built on a guess that the components in an average Al data center would take ten years to depreciate, requiring costly replacements. That was bad enough: "I don't see how there can ever be any return on investment given the current math," he wrote at the time.
But ten years, he now understands, is way too generous.
" had previously assumed a 10-year depreciation curve, which I now recognize as quite unrealistic based upon the speed with which Al datacenter technology is advancing," Kupperman wrote. "Based on my conversations over the past month, the physical data centers last for three to ten years, at most."
In his previous analysis, Kupperman assumed it would take the tech industry $160 billion of revenue to break even on data center spending in 2025 alone. And that's assuming an incredibly generous 25 percent gross margin - not to mention the fact that the industry's actual Al revenue is closer to $20 billion annually, as the investment manager noted in his previous blog.

"In reality, the industry probably needs a revenue range that is closer to the $320 billion to $480 billion range, just to break even on the capex to be spent this year," Kupperman posited in his updated essay. "No wonder my new contacts in the industry shoulder a heavy burden - heavier than I could ever imagine. They know the truth."
Kupperman called that gulf between tech industry spending and actual revenue in 2025 "astonishing."
However, it doesn't even begin to scratch the surface. For example, how does it all shake out when we account for 2026, when hundreds of new data centers are expected to pop up?
"Adding the two years together, and using the math from my prior post, you'd need approximately $1 trillion in revenue to hit break even, and many trillions more to earn an acceptable return on this spend," he writes.
"If the economics don't work, doing it at massive scale doesn't make the economics work any better
- it just takes an industry crisis and makes it into a national economic crisis," he concludes.
Overall, the pessimists broadly agree: it's no longer a matter of if Al is massively overhyped, but when the whole thing comes crashing down.
More on Al hype: Data Shows That Al Use Is Now Declining at Large Companies

9 140 380

Matthew Hughes @matthewhughes.bsky.social · 23h

Disappointed you didn’t write:

“Went to a craft brewery in Barcelona. Nek minnit…”

Matthew Hughes @matthewhughes.bsky.social · 1d

I guarantee those managers are lying, too, because “AI” is more palatable than the alternative (making existing workers do more, or outsourcing).

Fun fact, in their last financial years, four out of five WITCH outsourcing companies had positive growth. Some had faster growth than Apple!

Matthew Hughes @matthewhughes.bsky.social · 1d

Oakland!

Matthew Hughes @matthewhughes.bsky.social · 1d

Facebook reminded me that eight years ago today, I met @edzitron.com for the first time.

I don't think I took any pictures of us (I was fatter and had more hair), but I did capture this cool snap of the Ask.com headquarters on my Blackberry KeyOne, in part because I wasn't sure what year it was.

1 12

Matthew Hughes @matthewhughes.bsky.social · 2d

You can actually see the moment he realizes his life is over.

Genuinely one of the most brutal things I've ever seen on the Internet.

Matthew Hughes @matthewhughes.bsky.social · 2d

You say that, but...

Yes. This video enjoyed a brief moment of virality a few years back. Californian guy flies over to the UK for a rap battle, dedicates the last round to disclosing his opponent's infidelity.

Said opponent hasn't been seen since.

www.youtube.com/watch?v=F4IR...

JEFFERSON PRICE VS CAUSTIC | Don't Flop Rap Battle

YouTube video by Don't Flop Entertainment

www.youtube.com

1 8

Matthew Hughes @matthewhughes.bsky.social · 2d

NGL, considering writing about this just so that I can use that headline.

Matthew Hughes @matthewhughes.bsky.social · 2d

Turds of a feather flock together.

www.bbc.co.uk/news/article...

Rishi Sunak hired by Microsoft and Anthropic as paid adviser

The former prime minister has been told not to lobby the UK government by the appointments watchdog.

www.bbc.co.uk

1 2 19

Matthew Hughes @matthewhughes.bsky.social · 2d

Tesla actually makes a profit though and doesn't require potentially more than a trillion dollars to meet its short-to-medium-term commitments, all while losing vastly more than it brings in.

I agree, Tesla's share price is removed from reality. But share price != business fundamentals.

1 1

Matthew Hughes @matthewhughes.bsky.social · 2d

Another fun opening to a Yelp review.

Matthew Hughes @matthewhughes.bsky.social · 2d

"Come for the food, stay for a poo" will forever be my favorite line I've ever written in my entire life.

And I once wrote a headline that made Danny Dyer laugh on national TV.

1 3

Matthew Hughes @matthewhughes.bsky.social · 2d

The one bit of advice I'd give to any writer is to treat any opportunity to write as an exercise, and to be weird and funny and creative, especially in low-stakes situations where it doesn't matter if you don't actually accomplish the thing you're writing for.

Yelp reviews are a good sandbox!

1 3

Matthew Hughes @matthewhughes.bsky.social · 2d

Telling @edzitron.com about the time I got kidnapped by a British Airways Boeing 787 toilet, and the subsequent letter of complaint I sent to BA, which for reasons I don't fully understand, they didn't take seriously...

1 1 15

Reposted by Matthew Hughes

Matthew Hughes @matthewhughes.bsky.social · 3d

/fin.

Short post today. By which, I mean only 2,000 words.

Subscribe if you haven't already.

whatwelost.substack.com/p/generative...

Generative AI’s Impending Death By A Thousand Rake-Smacks

Give it enough rope...

whatwelost.substack.com

2 2 4

Reposted by Matthew Hughes

Matthew Hughes @matthewhughes.bsky.social · 3d

Each fuck-up that catches headlines only serves to shake that enterprise faith further.

And makes OpenAI's long-term survival look even more distant.

whatwelost.substack.com/p/generative...

1 4 12

Reposted by Matthew Hughes

Matthew Hughes @matthewhughes.bsky.social · 3d

Every time that faith is shaken, OpenAI and genAI become more tenuous.

When Deloitte -- one of the top-four accounting firms in the world -- has to refund a customer because an AI hallucinated stuff, it's bad for OpenAI and GenAI.

whatwelost.substack.com/p/generative...

1 3 10

Reposted by Matthew Hughes

Matthew Hughes @matthewhughes.bsky.social · 3d

Enterprise adoption of genAI is driven by faith in it being more capable than it is, and that it'll improve over time.

It's a faith that, like any other faith, will only be broken by the thing that the person has faith in.

whatwelost.substack.com/p/generative...

1 3 6

Reposted by Matthew Hughes

Matthew Hughes @matthewhughes.bsky.social · 3d

In practical terms, this means having 125m subscribers on its $20-a-month plan.

That's not going to happen. So, we're left to conclude that for OpenAI (and genAI) to survive, it needs to massively grow its enterprise users.

whatwelost.substack.com/p/generative...

1 2 4

Reposted by Matthew Hughes

Matthew Hughes @matthewhughes.bsky.social · 3d

The math for genAI doesn't make sense. Especially OpenAI.

Take its $300bn deal with Oracle. To pay for that, it'd need to grow its subscriber business to bring in as much revenue as Netflix does each month, while growing its API business at the same scale.

whatwelost.substack.com/p/generative...

1 3 9

Reposted by Matthew Hughes

Matthew Hughes @matthewhughes.bsky.social · 3d

What was once said only by critics is now being echoed by central banks, and even the people running genAI companies. What if we're in a bubble?

There's been a real vibe shift.

whatwelost.substack.com/p/generative...

1 5 13

Reposted by Matthew Hughes

Matthew Hughes @matthewhughes.bsky.social · 3d

If you're tired of generative AI, you're probably wondering how it'll end, and when.

So far, the conversation has centered on the broken, putrid economics of companies like OpenAI.

whatwelost.substack.com/p/generative...

1 3 10