New AI technology is transforming scientific research, drug discovery, education, software development, weather forecasting, robotics, energy, agriculture, transportation, manufacturing, accessibility, translation and communication, astronomy, and elder care.
Those are important! Keep them in mind when you read the next couple of sections so you don’t think I’m just a Debbie Downer.
AI and the economy – layoffs and entry level jobs
So far this year, more than 150 technology companies have cut more than 115,000 employees, according to the New York Times. AI is almost always called out as the reason for tech industry layoffs this year, although analysts and economists say that’s frequently “a smoke screen for companies looking to beef up profits or patch over old mistakes.”
Similar layoffs have begun in non-tech sectors – finance, logistics, consulting, media, retail, manufacturing, law. These are the first raindrops while the monsoon waits offshore, biding its time.
The job market will be disrupted for several years as businesses figure out the appropriate role for AI. “Disruption” is a code word for layoffs and not hiring people. Finding a job is already hellish and it will get worse.
In the short term eager employers will follow the tech companies down the road of believing AI can replace employees instead of empowering them. Entry level jobs will disappear because it will appear that they can be done more cheaply by AI, until employers discover they have killed the talent pipeline and no one has gotten the training to move into more advanced positions.
In other words, the graduates booing commencement speakers were absolutely correct to express their anger. Their world is going to be very difficult.
No one has a clue what the long-term effect of AI will be on the job market.
AI and the economy – the bubble
We’re in the largest infrastructure and investment cycle in history. At the same time the stock market has become completely untethered from risk. This is a double-plus-ungood combination.
Amazon, Microsoft, Alphabet (Google), and Meta will invest almost $700 billion this year on AI infrastructure – massive data centers, chips to fill the racks inside, and electricity to run them. Goldman Sachs estimates global spending on AI infrastructure will reach $7.6 trillion in the next five years.
The AI companies are moving giant stacks of Monopoly money in circular deals with each other for access to AI computing power – more than $100 billion in just six months according to Bloomberg.
The vast majority of AI companies are either burning massive amounts of cash or operating on razor-thin margins because their revenue is entirely eclipsed by the cost of computing power. Nvidia is the one and only AI company that is massively, over-the-top hyper-profitable because everyone needs its chips. It has a near monopoly on selling shovels to the goldminers.
Both consumer- and enterprise-facing AI companies are struggling to find profitable business models to justify the AI hype.
After their public offerings this year, the combined market value of SpaceX, OpenAI, and Anthropic will be around 5% of the entire U.S. stock market. Think about that – three companies supposedly will fairly represent five percent of the value of all the companies in the US stock market. Their gravitational pull will bend the stock indexes around them. Rules are being changed so they can almost immediately affect your index mutual funds. This is unprecedented and weird and creates an absurdly unbalanced, volatile market dynamic.
Worth noting as a likely side effect of the public stock offerings by SpaceX, Anthropic, and OpenAI: California will likely enjoy another brief respite from its perennial budget crisis, as in 2004 when Google’s public offering provided Sacramento with a much-needed windfall.
The path is narrow for AI ever to be profitable enough to justify the infrastructure investment. It’s not impossible, it’s just very hard to imagine from where we sit today.
If the demand doesn’t materialize, companies like OpenAI and Anthropic and thousands of other AI and AI-adjacent companies may begin to fail. Much of the impact will spill over to the private credit markets and junk bonds that are increasingly being used to fund the buildout. There are echoes of the risks that led to the telecom crash in 2001. Meanwhile the AI gold rush is crowding out investments in all other non-tech sectors of the global economy.
One accounting trick to watch for: data centers are filled with Nvidia chips that have a useful life of 1-3 years before they suffer physical degradation or become obsolete. The companies are depreciating them on a 5-6 year schedule to keep quarterly earnings looking more profitable. Soon the companies may have to start taking massive write-downs for billions of dollars of hardware that goes out of service. Oh, and those same chips are serving as the collateral for billions of dollars in debt, so the write-downs could trigger severe losses rippling through the private credit and banking sectors.
Analysts disagree about whether this is a bubble, even though hello?, it’s obviously a bubble, I mean, come on, we’ve gone full tulip mania, tech giants are spending like sailors on leave.
The AI bubble is singlehandedly propping up our economy. Jeff Bezos says it might be a “good” kind of bubble. Professor and economic advisor Mohamed El-Erian asks: “What if we’re in a “rational bubble” that, unlike other big speculative manias in history, takes our economy to a fundamentally better place?”
Sure! Right! What if everything is great? How dangerous can it be? Bubbles are harmless, children chase them across the grass, and everyone points at the rainbow colors and pretends not to notice the fuse burning inside – look, it’s a bubble, ooh, pretty!