- Generative AI is booming, but signs of risk are growing.
- Investment vs. reality may burst the bubble.
In the years ahead, Silicon Valley might look back on recent events as the moment when the generative AI craze went too far.
Investors spent the summer questioning whether top AI stocks could maintain their high valuations, given the lack of immediate returns from hefty AI investments. Yet, as September turns into October, fresh enthusiasm for key AI sectors — chips, large language models (LLMs), and devices — suggests the frenzy isn’t over. However, there are growing reasons for caution.
Cerebras: A chip contender with one big customer
Cerebras, a chip startup, was founded in 2015. The company aims to challenge Nvidia’s dominance by developing chips to train smarter LLMs. Nvidia, a major player in the AI boom, saw its market capitalisation grow from US$364 billion at the start of 2023 to more than US$3 trillion.
Cerebras, on the other hand, is highly reliant on a single customer: the AI firm G42, based in Abu Dhabi. The company generated 83% of Cerebras’ revenue in 2023 and 87% in the first half of 2024. Although G42 is a significant player, funded by Microsoft and Silver Lake, relying solely on one customer is risky. Even though Cerebras has sought to diversify its customer base, including securing a deal with Saudi Aramco, investors will likely scrutinise this concentration as the company pursues a US$7 to US$8 billion valuation in its IPO, according to Bloomberg.
OpenAI’s record-breaking funding round
Meanwhile, OpenAI set a new benchmark in Silicon Valley by closing the largest funding round in history, raising US$6.6 billion and reaching a valuation of US$157 billion. But there’s more to this story.
According to the Financial Times, OpenAI has requested that its investors avoid backing rivals like Anthropic or Elon Musk’s xAI. It is a rare request, given venture investors typically make a variety of investments. However, critics such as Gary Marcus and David Sacks have called OpenAI’s decision “shady” and said the company is feeling at least the potential for pressure from competitors.
Another concern is that OpenAI is backed by “bubble chasers” such as SoftBank and Tiger Global. These firms have a lengthy history of funding ventures at the peak of their success, which frequently leads to significant losses. Furthermore, OpenAI has encountered internal issues; since the resignation of Mira Murati, the company’s CTO, OpenAI has been in turmoil, attempting to reduce losses. The corporation is expected to lose US$5 billion this year, despite monthly revenue of US$300 million.
Meta’s push into AI wearables
Adding to the AI frenzy, Meta recently unveiled its ambitious Orion augmented reality glasses at its Connect conference. These wearables promise to integrate AI into everyday life, offering an immersive experience that “understands the world around you,” according to Meta.
Nvidia CEO Jensen Huang has backed the device, causing a positive market reaction. However, the product is quite expensive – each unit is said to cost US$10,000. Meta is attempting to save costs, but getting consumers to adopt AI-powered wearables remains a difficult task.
Snapchat, for example, tried something similar in the 2010s and suffered a US$40 million loss. It’s giving AI glasses another shot by collaborating with OpenAI, but it remains to be seen whether these devices will take off. Other AI form factors, such as the Humane AI pin and the pocketable rabbit R1, have failed as well.
What’s next?
Where does all of this leave the AI sector? On one side, OpenAI must prove it can justify its US$157 billion valuation while operating at a loss. On the other hand, Cerebras must reassure investors that its dependence on a single customer isn’t a deal breaker. Meta has a problem selling a new type of AI interface to a sceptical consumer base.
If these companies succeed, this moment could mark a turning point where AI truly reshapes industries. But the history of the tech world shows that high-stakes ventures often encounter turbulence along the way.
