Anthropic is trying to raise billions of dollars in its Initial Public Offering of shares, to fund its projected growth in the artificial intelligence (AI) market. Karl Denninger retorts:
After hyping every single possible use for "AI" and firing people en-masse to replace them with computer "intelligence" firms are discovering that the code and processes produced don't provide an actual economic benefit all-in compared with skilled employees. Firing people of course saves money but then the compute you buy directly and indirectly costs money and what the machine produces had better not require much "spit and polish" or you're underwater.
Worse, and I've discussed this before, doing this makes the technical debt percentage on said generated stuff nearly 100%. That is, you have no humans that understand it at all because no human built it and if a human reviews it that is akin to reading a book as opposed to writing it. The reader might (in this case will) skim it and lacks deep understanding -- specifically they have no understanding at all why certain decisions were made when it comes to structure because no human made the decision. The alternatives were weighed (if at all) in a machine rather than a human and the machine may or may not retain enough context to help you in the future. Five or ten years hence it will almost-certainly be "not" in that regard.
Firms have already run into this and many who have realize the problem -- as this accelerates it becomes a deeper and deeper hole. When (not if) something goes wrong there is nobody who can quickly look at the code and figure out why it broke.
There's more at the link, all of it very ugly from Anthropic's point of view.
Lawrence Person provides a different, but still damning perspective.
Anthropic made just under $1 billion in profit for one quarter (it’s first profitable quarter), yet it has committed to spending an additional $1 trillion, so it only needs 250 years of similar profits or so to pay off its projected capital costs!
. . .
If a small number of giant tech firms are just trading IOUs back and forth, how much real revenue is actually being generated by any of them? It’s less a self-licking ice cream cone than a self-climbing rope untied at the top, with each of the tech giants acting the fakir in turn.
Then there’s the big risk of the whole AI smoke and mirrors edifice collapsing: “If revenue growth stalls and fresh capital dries up, the company could breach covenants under its credit facility and commercial agreements, triggering cross-defaults and accelerated debt repayment.”
AI has proven that you can actually do useful things with it, and Anthropic’s Claude was (the last time I checked) the leader in AI code generators. But the amount of revenue coming into AI companies is nowhere near the trillions of dollars of capex being shoved into the insatiable maw of the AI beast.
Again, more at the link.
Anthropic is, of course, only one of half a dozen or so giant AI corporations and conglomerates vying to dominate the market. Of them all, I think only two or three, including SpaceX, Google and possibly Apple, generate (or are likely to generate) enough revenue from their existing resources to cover the costs involved - yet most of the others are already committed to borrowing enormous sums of money that they have no realistic chance of repaying out of earnings.
What happens when the train wrecks?
It's going to be a wild ride . . .
Peter
5 comments:
Most in the C suite are accounts and finance. They don't understand the technology or services their company provides, let alone any technology from outside their company. They are focused on growth. Just like Pavlov's dog, ring the growth bell and they will begin to salivate, or in this case invest money, at the thought. No actual growth required. Employees cause problems. Get rid of those problematic buggers. The C suite jumps all in.
Which leads us to the law of unintended consequences. The most problematic employees tend to be the ones that understand the customers and Tech. They are old and experienced and were getting paid commensurate with their experience. Who shall help unravel the Gordian knot.
Wild ride would be an understatement and still will be except most things are already broken to some degree as corporations have put margin ahead of customers.
Large Language Models are inherently flawed.
"Nothing is better than God.
Warm beer is better than nothing.
Ergo: Warm beer is better than God"
A simple bit of doggerel that demonstrates the problem. In reality, it is infinitely worse.
WorldCom... ding!
NorthPoint... ding!
Global Crossing... ding!
All have joined the crew of the Edmund Fitzgerald, and they all await the newcomers...
Ed Zitron has been sounding the AI alarm in detail.
https://www.wheresyoured.at/running-dry/
I keep seeing nonsense about how, if we don't spend trillions of Monopoly money on AI, the Chinese will beat us and we will become a third world nation more quickly than we are.
Here's a suggestion that Trump could make; We cede all AI leadership and tech to the Chinese. De-list all US firms working on it, remove it from use in the US.
I'm becoming disillusioned with Elon Musk, who seems to be less interested in expanding humanity into the solar system and beyond. His increasing focus on AI is leading me to think that he is just another oligarch, as are Zuckerberg and Bozos.
In exchange China ends it's manned space program and maintains only unmanned systems. No Chinese moon base, space station, or colonization.
Post a Comment