The Risks of Excessive AI Chip Funding: A Historical Comparison

When a CEO asks for trillions in funding, not billions, you know the sector could get very hot.

In the long term, generative AI will transform many industries and the way people work. However, the report that OpenAI CEO, Sam Altman, is talking to investors about the artificial intelligence chip project, raised many questions.

A person familiar with the talks said the project could require raising as much as $7 trillion. Even a fraction of that figure, which is larger than the combined GDP of the United Kingdom and France, may seem, to put it mildly, excessive.

However, it reflects what has become of the great interest in artificial intelligence and the segments that occupy it. The historical parallel this situation invokes is that the record high valuation of AI-related stocks and funding targets resembles the boom and bust in telecom stocks during the dot-com bubble era.

Back then, investors expected that the Internet would transform the world, and telecom companies and hardware suppliers would be the biggest winners. But the problem was that sector valuations believed that this transformation would happen almost overnight. Now, a similar level of optimism is driving investment in AI-related companies.

When the Internet first began to be widely used, networking devices were the most in demand. Servers had to be built and connected using routers. Companies began building and purchasing hardware with the understanding that the intense demand for servers would continue indefinitely. Shares of telecommunications equipment companies such as Cisco rose more than 30-fold in the years they reached their peak in 2000.

But the collapse of the telecommunications industry came earlier than expected, as it took only four years to go from boom to recession, much faster than the Internet changed our lives. Oversupply pushed more than 20 telecommunications groups into bankruptcy by 2002, and their shares tumbled.

Now, in the world of artificial intelligence, chips are the most in demand. Therefore, the rush by AI companies to own more of the chip industry supply chain is understandable. As AI models get larger, more chips are needed. The ongoing shortage in supply increases the urgency of demand.

However, how long this supply shortage will last is debatable. It was only two years ago that the global auto industry came to a halt due to a severe shortage of car chips. It took less than a year to alleviate this impasse. Today, the supply of automotive chips is not only insufficient, but in abundance.

The biggest risk of spending too much money too quickly on AI chips is excess production capacity. This is already a problem for older generation chips. As the current downturn in the sector continued for a longer period than expected, Samsung was forced last year to reduce production to deal with the glut of chips, and its Japanese counterpart, Kioxia, incurred a record loss of $1.7 billion during the three fiscal quarters until December. In addition, more than 70 new manufacturing plants are being built.

Meanwhile, global silicon chip shipments fell 14.3% last year, in part due to a cyclical downturn in the chip sector and lower demand for consumer electronics. But the decline in global chipmaking equipment bills, which fell more than 10% in the third quarter, suggests that future chip sector growth will remain at a more normal level than the AI ​​boom led us to believe.

The other problem is that chips are quickly becoming a commodity. Take, for example, the old 40-nanometer chips used in home appliances. These chips are hardly in short supply today, but they were also a scarce and cutting-edge resource when they were introduced in 2008. As the value of the equipment declines, Capitalism, prices of older generation chips fall.

The speed of chips increases and the efficiency of programs improves every year. It only took two years for the chips to evolve from the 7-nanometer technology to the advanced 5-nanometer technology used in the latest Nvidia chips. This rapid technological advance means that companies may end up spending much less on chips in the future than companies predict today.

It is true that there are clear differences between the dot-com era and the artificial intelligence boom. For example, OpenAI’s revenues have already exceeded two billion dollars on an annual basis, catching up with the fastest growing technology platforms in history months after its launch, as is available to companies today. More ways to make profits.

But as was the case in the early days of the Internet, broader adoption of AI in enterprises is still some way off, and the transformation caused by AI may take many years longer than current stock prices and funding forecasts indicate today.

Excessive hype and overinvestment are a dangerous combination. The way to avoid a similar fate to their overrated peers in the 1990s is to remember that history repeats itself.

2024-02-17 22:03:04
#hype #revolution #similar #boom #bust #period #telecommunications #sector