Capitalism is a long succession of booms and busts stretching back hundreds of years. We're now at the peak of another boom; that means a crash is inevitable. It's just a question of when. But there are other questions to ask too.
If many of the current AI players are destined to crash and burn, what does this mean for the type of AI we will end up with in the 2030s?
Is AGI destined to be created by an as-yet-unknown post-crash company?
Will open-source AI become the bedrock of global AI during the crash & post-crash period?
Crashes mean recessions, which means cost-cutting. Is this when AI will make a big impact on employment?
AI Bubble Warns: Sløk Raises Concerns Over Market Valuations
Wall Street’s AI Bubble Is Worse Than the 1999 Dot-com Bubble: This means when it crashes, the AI that arises from the ashes will be different. What will it be?
byu/lughnasadh inFuturology
23 Comments
Why would the AI be different?
The article doesn’t really say much other than that “AI companies’ market values have become too large.”
During the dot-com bubble, many dot-com companies inflated their market values based solely on expectations, despite having minimal actual revenue.
However, current AI companies are already generating billions to tens of billions of dollars in revenue. For example, OpenAI’s projected revenue for 2025 is over $12 billion.
PS – Of course, there is one possibility. AI could eliminate all jobs and destroy the market itself.
I assume you’re taking highly leveraged positions to put your money where your mouth is? You have NVDA shorts, for instance? Or are you just yapping?
less LLM and more other types of more specialized models
I actually think AI is under valued to various extents. In the next 10yrs AI will become a utility everyone has in their homes. No different than electricity or water. AI will replace streaming services. AI will be able to produce and or reproduce any movie, show, podcast, song, etc one wants. AI will operate/replace Ring, Nest, and all the other smart features currently in homes.
Just the domestic home uses will be a $100 billion a year business in North America alone.The industrial uses will be even bigger. AI will enable small businesses to scale up quickly by performing basic administrative, accounting, and legal functions.
Today AI is just helping students with their homework and making memes. In the future it will be a much more utilized tool.
It won’t be different. Just all the AI tools that deliver no value will disappear. Just like what happened with the DotCom bubble.
This is actually a bit overstated. AI is already producing large returns in both efficiency, cost-savings, and even driving growth in key profit sectors for companies like Microsoft, Meta, Google, and so on.
Yea the valuations are ridiculous but they are being driven by actual capital expenditures. This is tangible stuff being built all across the world. These are real products and tools that are rapidly transforming the way we live our lives. Whether you like it or not, those are the facts. To compare this to the dotcom boom is false equivalence.
From what we have been seeing it seems the incumbents in the tech sector will keep all the spoils of the LLM, generative-AI and GPU compute boom.
I don’t call it the AI boom because AI takes many forms, what we are seeing is just a few forms being expanded massively.
Of note I think LLM and gen-AI are going to get better and integrated more deeply into products/platforms/services we already use today, but not much more out of it will come out of it in the near future. I don’t think we will see new products/services using these techniques that massively change how we do things today.
The GPU compute boom though is another matter, once the hype dies down a bit the GPU compute is going to get lot cheaper I imagine and people will find new forms to apply that compute beyond LLMs and gen-AI. I imagine the simulation space is going to expand massively with the compute available will drive a lot of optimization in everything, from materials, pharmaceuticals, cars, engines, planes, rockets, gene-editing, etc.
People will call it AI but it will be more like linear regression which is already widely used. Just applying those techniques to more things and in a far larger scale through the extra compute available for simulations.
The current LLM algorithms and math behind how current AI works will need a complete overhaul. Current AI needs to be able to run locally on a phone for it to be profitable. I agree we will see or are starting to see companies that are fake it till you make companies and don’t have a real market. Also the fact that mentioning AI in earnings calls still causes volatility means investors are still on cloud nine that AI will solve all problems. I think in the 2030s it needs to be completely different. It’s actually not efficient at all to communicate the way humans do. It was just done that way because how else can you show off its ability to generate more curated responses than current search engines. I’m still of the opinion that ChatGPT is just the next evolution of search and other adjacent abilities are showing potential areas where it’s also useful like big data and code generation. Even though code generation is very simplistic. Still a long way to go to get to actual AGI
Betting on crashes is all about timing and I can guarantee you wont have the timing right or the money to keep betting
One of the big things will be the glut of datacenters. Hundreds of billions of dollars are going into datacenters that will not necessarily be needed when the crash happens
Same thing happened in the dot com crash, there was a TON of fiber laid that didn’t get used. In some areas we are just now finally absorbing all of that dark fiber
Now, datacenters are different. You can’t turn off servers and have them be viable 20 years later, but cloud computing will get cheap and there will be a glut of buildings that will get repurposed or sit fallow for years until their capacity can be absorbed
Integration. You don’t go to it, it’s already there, embedded in everything.
Bold of you to assume we’ve already reached the peak.
AI will ultimately have a similar effect to the ubiquitous networking that came with the internet and cause most software to be replaced or rewritten
It’s not worse than 1999. There were many worthless companies pretending to be something they’re not. The AI companies today, at least some of them, are growing at 50-75% a year.
If it can replace people companies will pay huge for it.
There are so many AI startups. My guess is AI will become a commodity with low barriers to entry, meaning nobody is going to be making a lot of money off of it.
Think of it like Uber. Yeah, a great idea destined to make huge profits. But it’s not hard for competitors to pop up in that space. Uber ended up losing gobs of money for over a decade before they managed to make profits.
The AI bust can’t get here soon enough… Problem will be how badly will it hurt the market
Niche efficiency tools, unstructured data processing, and simple code lookups- as God intended.
It’s not crashing. That’s the difference.
I’m in VC now. Many of the partners in my for
Had been working the Bay Area and Silicon Valley in 1999. They said the current vibe they get from AI is more 1997/1998, and they think there is time for the hype to keep going.
I personally think as we approach the singularity we will eventually have a Kondratiev cycle overlap. This would permanently end the boom / bust cycle
thinking about an AI ‘phoenix’ rising from the ashes of a bubble burst is seriously fascinating, makes you wonder what kind of tech will truly last.
If the bubble pops there will be so much distracting news you won’t be thinking about what comes next.
Instead of AI winter were going to get a damn AI Ice Age.