AI bubble trouble
The signs of an impending crash, the impact on AI stocks like Nvidia, and the future of AI.
This year, AI helped the stock market hit some all-time highs. For a few days in June, microchip maker Nvidia become the world’s most valuable company, seemingly out of nowhere. But stocks have tumbled in the past few weeks, and Nvidia’s stock has dropped by over 20% since late June.
Amid growing concerns about the sustainability of AI investments, people are asking: when’s this AI bubble going to burst?
No one could possibly know the answer to that question, but what seems to be certain is that it will happen (is happening?), it’ll be bad (but how bad?), and despite all that, AI is here to stay. The past few economic bubbles (like the dot-com crash of the late 90s and the housing crash in 2008) didn’t mean the internet was a passing fad or that no one would buy houses ever again. It meant that websites and houses were overvalued, like AI is now.
What’s a financial bubble?
A bubble is when prices of things like stocks or houses quickly increase and then suddenly crash. Usually, we can only tell a bubble happened after it bursts. To see where we are with AI, we can look at the five stages of a typical bubble:
Displacement: This is when investors get excited about a new, game-changing idea. For AI, this started with the launch of ChatGPT, which showed people how advanced AI could be in a user-friendly interface, similar to when the first web browsers came out in the early 1990s.
Boom: After ChatGPT, there was a huge rush of money into AI technology. Tech companies, startups, and venture capitalists all wanted a piece of the action, similar to the dot-com boom when lots of internet companies went public. For example, some AI startups saw their funding increase by ten times in just a few months.
Euphoria: We’re probably in this stage now. Tons of money is being invested in AI, with companies like OpenAI getting huge amounts of funding for big projects. This stage is marked by a lot of excitement and risky bets, where people aren’t being very cautious. Some companies are valued very highly even though they aren’t making much money yet.
Profit-taking: Right now, most AI companies aren’t making big profits. Some investors are starting to sell their shares and take their profits because they see signs that the market might turn. For example, some early investors are selling their stakes in AI startups even though the hype is still high.
Panic: This is the stage where the bubble bursts. Prices drop quickly as everyone tries to sell. We haven’t reached this point with AI yet, but there are signs it could happen. People are getting more skeptical, and the market is becoming more unstable. Watch out for falling investment rates, bad earnings reports, and big sell-offs by major investors as signs that a crash might be coming
Why are AI stocks falling?
Major AI stocks like Nvidia and Super Micro Computer have dropped significantly in the past few weeks. Nvidia fell more than 30% from its highest price, and Super Micro Computer lost over half its value. This drop affected other big tech companies too, like Alphabet (Google), Meta (Facebook), and Microsoft. People are becoming more doubtful about AI investments because AI has yet to make much money. Nvidia's delay in releasing its latest AI chip made things worse. The sell-off shows that investors are getting impatient with how long it takes for AI to be profitable.
Jim Covello from Goldman Sachs says that spending $1 trillion on AI needs to be justified by solving $1 trillion problems. Since many AI applications are still being developed, there are worries that the benefits of AI aren't worth the high costs. Also, global shortages of computer chips and power are expected to slow down AI's growth.
Is Goldman Sachs right? No one really knows. Right now, any predictions about AI's impact are just guesses.
Why isn’t AI stock doing better?
The name “artificial intelligence” implies that machines are intelligent. Right now, they aren’t. Large language models (LLMs), the AI du jour, are just machines that find the median answer to your query (whether it’s words or numbers). You, the user, can lead and cajole AI to give you more or less what you’re looking for, but you are the guide.
The next AI milestone is artificial general intelligence, or AGI. As I wrote in another post, “AGI typically means artificial intelligence that matches (or outmatches) humans on a range of tasks.” AGI will inject new ideas into its outputs. AI companies are in an arms race to find a way to AGI.
If and when AGI happens, it will change the course of history. That’s no bubble.
The global AI chip shortage
One of the reasons AI is surging is because of the advancements in microchip technology, particularly chips produced by Nvidia. These chips are essential for building systems like ChatGPT. But there’s a growing concern about the sustainability of this high-volume production. Many of Nvidia’s significant clients are developing their own competing chips, and the market’s reliance on materials like aluminum and copper is leading to shortages and increased costs.
AI ethics could help profitability
History shows that bubbles can be catalysts for growth, leading to technological advances and more sustainable practices. With some baseline principles, we can turn an AI bubble into an opportunity to create a more advanced and equitable society. This includes:
Eco-friendly data centres: Investing in renewable energy sources and advanced cooling technologies to reduce AI’s carbon footprint.
Circular economy for tech: Implementing recycling programs for electronic waste, ensuring that materials like copper and aluminium are reused.
Robust regulatory standards: Creating international agreements on AI development standards to prevent misuse and bolster safety.
Public-private partnerships: Encouraging collaboration between governments, academia, and private companies to drive responsible AI innovation.
Education and workforce transition: Preparing the workforce for changes brought by AI through education and retraining programs.
Ethical AI development: Focusing on creating AI systems that are not only powerful but also ethical, transparent, and beneficial to society as a whole.
Thanks so much for reading! I’m on vacation for the next couple weeks, so expect a new Human+AI newsletter in September.
Weekly disruptions
Scientists suggest sending artificial intelligence to aliens for real-time communication (Mashable) Two scientists suggest using AI to communicate with aliens in real-time. They propose sending AI models into space as part of "messaging extraterrestrial intelligence" (METI) to answer questions about humanity and interact with extraterrestrial beings.
Meta courts celebs like Awkwafina to voice AI assistants ahead of Meta Connect (The Verge) Meta is reportedly recruiting celebrities like Awkwafina, Judi Dench, and Keegan-Michael Key to voice its AI assistants. The company aims to finalize these deals before its Meta Connect conference in September, where it plans to showcase the new voices as part of its digital assistant product, MetaAI.
World’s first major AI law enters into force — here’s what it means for U.S. tech giants (CNBC)The EU's landmark AI Act, the first major law regulating artificial intelligence, officially came into effect last Thursday. The regulation imposes strict requirements on how companies develop and deploy AI. The Act introduces a risk-based approach to regulation, with severe penalties for non-compliance, and aims to ensure safe and ethical AI practices, setting a global standard for AI governance.