Contrary to all the doom and gloom and popular headlines about the imminent bursting of an AI bubble, investment in AI is greater than ever. Two tech podcasts have addressed this topic, and both conclude that there is no evidence to suggest an AI bubble or the imminent end of the AI hype.

The Hard Fork podcast by Kevin Roose and Casey Newton lists the investments made and announced this year by companies such as Nvidia and Microsoft. Nvidia has just announced that it will invest $100 billion in OpenAI, the developer of ChatGPT. To date, OpenAI has received investments totaling $71.4 billion. OpenAI is also building five new data centers in the US, three with Oracle and two with Softbank. These are part of the US government’s Stargate project and will bring in $400 billion in investments over the next three years. Nvidia CEO Jensen Huang estimates that around $600 billion will be invested in AI data centers in 2025 alone.

The two hosts call this the Great AI Buildup and compare it to the expansion of the US Interstate Highway System. The highways built over a 36-year period involved investments of $300 billion, adjusted for inflation. Now, twice as much money is being invested in AI infrastructure in a single year.

The scale of investment is enormous, and overall, OpenAI CEO Sam Altman’s statements that he wants to allocate $7 trillion to AI seem to be within reach, and no one finds these sums ridiculous anymore. The companies themselves are willing to invest come hell or high water because they see themselves in a competition to be the first to achieve superintelligence (AGI). And this goal justifies spending such sums.

The companies that are shouldering these expenses have one advantage: they are all highly profitable and have tens of billions in cash in the bank. They are also all still run by their founders, who are taking this big gamble and linking the future of their companies to it.

In addition, superintelligence could be one of those zero-to-one technologies where the first to achieve it monopolizes the market and thus obtains a license to print money. For the founders, this prospect seems to more than justify the expenditure.

Azeem Azhar takes a more methodical approach in his podcast Exponential View. He created five gauges to assess whether AI is in a bubble.

• Gauge 1 – Economic burden: Are investments currently large enough to influence the economy?
• Gauge 2 – Burden on the industry: Are industry revenues in reasonable proportion to the investments made?
• Gauge 3 – Revenue growth: Are revenues growing/expanding fast enough to catch up?
• Gauge 4 – Valuation heat: How hot are valuations? Are stocks excessively overvalued in historical terms?
• Gauge 5 – Financing quality: What type of money is being used to finance this? Is it strong balance sheets or fragile, volatile capital?

In his detailed analysis, he also refers to previous bubbles, including those in the railway and telecommunications sectors. But back then, just like today, infrastructure was created that we are still building on today.

In any case, with the exception of the second gauge, industrial strain, all of his indicators are in the green and therefore safe zone. Even the second indicator is still in the yellow zone.

Morgan Stanley expects $3 trillion to be invested in AI infrastructure by 2029. Comparing these investments to gross domestic product provides an estimate of how infrastructure projects have performed in the past. AI infrastructure investments would thus range between 0.9% and 1.7% between 2025 and 2030. Railway investment was almost 4%, while telecoms investment was 1.2%.

One problem here is that railway infrastructure is durable. It can be assumed to have a service life of 30 years. Telecommunications investments also have a minimum service life of 12 years. The situation is different with AI, where the service life is only five years, as chip designs quickly become obsolete and have to be replaced by new, more powerful chips.

None of these indicators currently point to an AI bubble. The companies involved in the race can afford it.

It is also important to note what countries such as Germany or France would have to invest in order to remain on a par with the US. Germany would have to invest between €36 and €68 billion annually, France between €24 and €46 billion. However, these amounts are not seen in these countries.