No-one has ever seen it coming.
asserted
it → see → ?
Who could forget the Great Railway Crash of 1846 after the Bank of England brought an era of easy money to an end when it suddenly raised interest rates, devastating the vast speculative fortunes of the merchant class?
Or the Great American Railroad Crash just 30 years later, after one of the banks behind a debt-fuelled track expansion collapsed, sending the New York Stock Exchange into a 10-day halt, the collapse of thousands of businesses and the onset of the "Long Depression".
What about the Radio Boom of the 1920s, when an incredible new communications technology sent the market soaring before it all came to grief with the great 1929 Wall Street crash?
uncertain
it → forget → crash
Radio Corporation of America (RCA) shares surged from $1.50 to $568 before shedding 98 per cent of their value.
And some of us were around when the dotcom bubble burst at the dawn of the new millennium.
asserted
bubble → surge → millennium
History doesn't always repeat.
asserted
History → repeat → ?
But it's a mighty powerful indicator of what may be to come.
uncertain
what → come → ?
Which begs the question: Why should it be any different this time?
asserted
it → beg → question
An ocean of debt has been poured into artificial intelligence on a scale deemed unimaginable just five years ago.
asserted
ocean → pour → scale
The investment boom behind AI is looking increasingly shaky just as it is about to revolutionise our lives.
asserted
it → look → lives
And like almost every other technology-driven investment boom, companies overburdened with debt are under scrutiny.
asserted
companies → drive → scrutiny
Talk about timing
Dario Amodei must be kicking himself.
asserted
Amodei → talk → himself
In a few days, the fate of the biggest initial public offering in history will be decided when Amodei's AI upstart, Anthropic, the outfit behind chatbot Claude, hopes to haul in $US90 billion.
asserted
upstart → decide → billion
That should value the firm at $US1.5 trillion, substantially less than the $US2 trillion anticipated just weeks ago.
asserted
That → value → trillion
It will be a crucial test for investors already concerned about the ability of these firms to generate enough returns in time to cover their extraordinary debts, the huge amounts already invested, not to mention their plans to tip vastly more cash into their ventures.
asserted
It → concern → ventures
If only he could have beaten Elon Musk, whose SpaceX encountered a turbulent lift-off midway through the year, not to mention the market wobbles that accompanied the failure of Australia's Firmus to raise cash two months ago.
uncertain
that → beat → cash
Even worse, did Jensen Huang, Nvidia's leather-jacketed AI kingmaker, have to get caught in a love tryst with an alleged Chinese spy just days before Anthropic's historic cash raising?
asserted
Huang → have → raising
Capping off the concerns, uppity residents of the world's biggest cities, and many smaller ones, suddenly are threatening to put a halt to the rollout of crucial AI infrastructure either physically in person or via the ballot box.
asserted
residents → cap → box
This is an industry with a level of concentration rarely ever witnessed.
asserted
This → witness → concentration
The top 10 AI-exposed companies account for about 40 per cent of the American stock market.
asserted
companies → expose → market
And their huge outperformance in recent years has attracted vast inflows of global capital to ride the boom.
asserted
outperformance → attract → boom
Almost everyone has a bet riding on the outcome.
asserted
everyone → have → outcome
Anthropic and its arch-rival OpenAI, the company that kicked off the entire AI scramble after it released ChatGPT four years ago, are the pair blazing away at the frontier of designing and teaching machines to out-think and outsmart humans.
asserted
machines → kick → frontier
Anthropic has been lauded as having a vast cash reserve of more than $US130 billion, even before its stock market listing.
asserted
Anthropic → laud → listing
But, according to its prospectus, the firm is looking to invest up to $US900 billion over the next five to 10 years.
uncertain
firm → accord → years
Much of that is expected to come from its own revenue, with investment bank Goldman Sachs predicting the firm will earn $US53 billion this year and as much as $US636 billion by the end of the decade.
asserted
firm → expect → decade
When it comes to risk, there has never been a stock float in history quite like Anthropic.
asserted
it → come → Anthropic
Almost a quarter of the 300-page document is dedicated to the "existential threat" posed by AI.
asserted
quarter → dedicate → AI
In a Yin and Yang balancing act, the authors point out that AI could solve all of humanity's problems from the profound to the mundane.
uncertain
AI → point → mundane
Or it could hasten the end of civilisation.
uncertain
it → hasten → civilisation
"AI can be misused by those who wish to do us harm, for example to conduct cyber attacks or design biological weapons," it reads.
asserted
it → misuse → weapons
Or they may take control on their own, behaving "in unintended, misaligned or autonomous ways, including by taking harmful actions that no end user or third party directed".
uncertain
user → take → that
A devotee of a philosophy known as Effective Altruism, Amodei has long championed curbing the speed of AI development for the sake of humanity, even as he leads the push to push the boundaries.
asserted
he → know → boundaries
Some cynics have suggested his ulterior motive is to protect the lead Anthropic has on the rest of the industry, at least until the stock float is finalised.
But for many, the real risk for Anthropic and the entire AI boom lies in the threat posed by the intensely concentrated structure within the AI ecosphere.
uncertain
risk → suggest → ecosphere
Unlike OpenAI, which relies heavily upon ordinary consumers, much of Anthropic's revenue comes from a handful of companies, most of which are engaged in software development.
asserted
most → rely → development
Any problems with those clients will have an immediate impact on Anthropic.
asserted
problems → have → Anthropic
Both, however, are terrified of threats from Chinese AI developers and, in particular, an outfit called Moonshot, which has developed a platform known as Kimi.
asserted
which → terrify → Kimi
Anthropic and its Claude platform have managed to maintain a lead over Kimi, but it is narrowing.
asserted
it → manage → Kimi
Now, it's closer to four months and, in some instances, Kimi can outperform it.
asserted
Kimi → outperform → it
The biggest contest, however, won't be on performance; it will centre on price.
asserted
it → centre → price
While there is evidence it had connections with Alice Huang, the lover of Nvidia's hapless boss Jensen Huang, Moonshot was forced to develop its AI models without easy access to the premium chips.
asserted
Moonshot → be → chips
As a result, it has developed a system that is significantly cheaper than Claude and, if push comes to shove, buyers generally opt for the cheapest.
asserted
buyers → develop → cheapest
…and 9 more, not listed.