OpenAI Is About To Make The AI Bubble A You Problem
OpenAI Is About To Make The AI Bubble A You Problem
The absurdity is palpable.
OpenAI is set to implode. An investigation by The Information recently found that the company has burnt over $8.5 billion on staffing and AI training and is set to make an operational loss of $5 billion by the end of the year. As such, bankruptcy is a severe possibility for the AI giant. For reasons we will get into in a minute, I thought OpenAI had no hope in hell of stopping this inevitable flop. They have no viable way of drastically increasing their revenue, and if investors take even a customary glance at their business fundamentals and basics of AI, there is no way they would pile billions more into them. But it looks like I have egg on my face; as it turns out, OpenAI is in talks to raise $6.5 billion from investors, and secure $5 billion in credit from banks, paying for another six months of AI development, and their upcoming gargantuan loss. In effect, OpenAI is bringing the banks on board their sinking ship and making the AI bubble a problem for us all.
So, why is the OpenAI ship sinking? It is still a young company, and its AI is predicted to get better and better; surely, it will only see colossal growth over the coming years.
Well, while this is a widely held notion, it is far from true.
I know I have referred to this a million times by now, but the truth of the matter is that AI development is hitting a hard ceiling as diminishing returns take their toll (read more here).
You see, to keep the same AI development progress, its training data needs to grow exponentially. For example, to make a simple AI 5% more accurate with each generation, each successive generation might need to double the training data size of the previous generation. Sourcing and preparing this data is far from cheap, so this represents a rapidly expanding cost.
However, training the AI on this data is also expensive, and this expense also increases abruptly. The AI has to compare each individual point of data to every other data point in the set to find connections and trends. This means that for each bit of data you add to an AI training database, the amount of computational work it takes to train that AI on that database increases exponentially.
There are also numerous other restrictions, such as a finite supply of GPUs needed to build the giant supercomputer clusters used to train these AIs (whose size also needs to dramatically expand as training data size increases) and a limit to the energy available to power these behemoths.
As such, OpenAI’s operational costs are only set to skyrocket, and the progress of its AI development is set to grind to a halt. This is beautifully demonstrated by their upcoming $5 billion loss and the fact that ChatGPT 4o is barely better than ChatGPT 4.
This is a massive problem for OpenAI, as its current hype and valuation are based on the fact that its AI models will get significantly better in the coming years and that the added reach this will give them will produce a vast profit. But this future simply isn’t possible with the limitations of AI. That isn’t just me saying that; OpenAI CEO Sam Altman has admitted that next-generation AI models will require an energy revolution, like fusion power, to become viable.
When I say OpenAI is being overvalued, I mean it. This current funding round places the value of OpenAI at $150 billion, just under $3 more than Uber. Uber is very similar to OpenAI, being a tech-lead industry disrupter. However, unlike OpenAI, it posted a profit of $1.1 billion in 2023 from an annual revenue of $37.28 Billion. In comparison, OpenAI’s predicted $5 billion loss and $3.4 billion annual revenue for 2024 is utterly pitiful. The only metric OpenAI beats Uber is active users, with 200 million compared to Uber’s 150 million. However, Uber customers spend an average of around $95 monthly, while OpenAI customers typically only spend $14 monthly with them. So, in reality, Uber’s customer base is way more valuable.
This alone makes OpenAI look wildly overvalued. However, when you find out that industry experts calculate Uber is overvalued by 27%, the absurdity of OpenAI’s value is utterly ridiculous and quite frankly dangerous.
Why is it dangerous?
Well, we know that OpenAI is going to get nowhere near profitable any time soon, thanks to the huge limitations of AI. But it also can’t continue raising investment and debt forever to keep the walls from the door. Eventually, its hype will fade, its value will crash, and it will default on debt.
Who do you think pays for that debt? Not OpenAI, as they don’t have assets anywhere near valuable enough. As such, the banks will look to fill the hole in their books from you and me.
Now, this would be fine if it were only OpenAI in this situation. $5 billion going missing is bad, but it won’t impact the wider economy much. But that isn’t the case. OpenAI is only one of thousands of AI startups, and securing this funding and debt will set a precedent for lending billions of dollars to near-destitute AI companies. As banks and investors are desperate to catch the next big thing, this precedent will open the proverbial floodgates and inundate the AI industry with more debt. At that sort of scale, the impact of the AI bubble bursting could trigger an economic crash.
I’m not making some wild prediction here. This has happened before with the dot com bubble, which caused a recession. The only difference is that the dot com bubble occurred off the back of the booming 90s economy. Our economy is nowhere near as strong, so an AI bubble-triggered recession would hurt us far more, especially if this industry is allowed to take on billions upon billions of dollars of debt.
Thanks for reading! Content like this doesn’t happen without your support. So, if you want to see more like this, don’t forget to Subscribe and help get the word out by hitting the share button below.
Sources: Reuters, Yahoo, Yahoo, CMC, The Verge, Macrotrends, Business Wire, Vice, Reuters, Will Lockett, Will Lockett, Alpha Spread