The trillion-dollar question
Goldman Sachs (GS) released the report on June 25th titled: Gen AI – Too much spend, too little benefit. The abstract of the report starts with:
Tech giants and beyond are set to spend over $1tn on AI capex in coming years, with so far little to show for it. So, will this large spend ever pay off?
So, that’s the trillion-dollar question. Will the Gen AI capex prove to be profitable and add value to the AI-related companies, and overall, to the US economy? Or, alternatively, will the trillion-dollar Gen AI capex prove to be a wasteful investment?
In this report, Daron Acemoglu from MIT “estimates that only a quarter of AI exposed tasks will be cost-effective to automate within the next 10 years, implying that AI will impact less than 5% of all tasks”.
Overall, the Goldman Sachs report is very skeptical of the Gen AI capex benefits, as the title of the report suggests.
Just recently, a Bank of America report noted that “AI hype days are over as the market shifts into assessing investment pays offs.” This is essentially the same concern shared by Goldman Sachs. The Bank of America report notes that:
Expected capital expenditures in AI by so-called hyperscalers such as Microsoft (MSFT), Meta Platforms (META) and Amazon (AMZN) have been upwardly revised by $18B to compared with only a $2B increase in sales forecasts since March, Bank of America said in a note. Expressing it another way, AI capex has been revised up $9 for every +$1 sales.
The key implication of the statement above is that a $9 of Gen AI capex is required for a $1 in sales, which does not appear to be a very efficient investment.
The Alphabet bomb
As I noted, Alphabet (GOOG) (GOOGL) was asked during the Q2 earnings call what was the ROR on the Gen AI capex. Sundar Pichai of Alphabet did not answer the question, and only said:
I think the one way I think about it is when we go through a curve like this, the risk of under-investing is dramatically greater than the risk of over-investing for us here, even in scenarios where if it turns out that we are over investing.
It appears that Sundar Pichai does not know what the ROR on Gen Capex is, but it fears that the risk of underinvesting is greater than the risk of overinvesting.
It’s important to understand the context behind Alphabet’s fear of underinvesting in Gen AI, even if the project is unprofitable. The entire Gen AI capex boom started when Microsoft invested in OpenAI in late 2022 to directly challenge Google Search.
Naturally, Alphabet was forced to develop its own Gen AI LLM product to defend its core business – Search. Meta joined in the Gen AI capex race, together with Amazon, in fear of not being left out – which led to a massive Gen AI capex boom.
Further, this massive Gen AI capex boom possibly caused the Gen AI bubble, as the stocks associated with Gen AI soared, in what seems to be very similar to the dot-com bubble of the late 1990s.
Obviously, if the Gen AI capex is a wasteful investment, which is what Goldman Sachs and Bank of America seem to imply, then I believe we are facing a potential massive bubble burst, with the S&P 500 (SP500) falling sharply, similarly to the bear market of 2000-2003 – especially given the fact that the market cap-based index such as the S&P500 is heavily concentrated with the Gen AI related tech mega-caps.
So, what’s the ROR on Gen AI capex?
So, the key question is what is the ROR on Gen AI capex – because if the ROR is high enough, the Gen AI is possibly not a bubble.
Sundar Pichai of Alphabet apparently does not know what the ROR on Gen AI capex is. We will find out this week when Microsoft, Meta, Amazon and Apple (AAPL) report their earnings if other executives are able to answer the same question. The ROR question will undoubtedly be asked at the earnings conferences, given Alphabet’s omission.
The capital budgeting process
But let’s try to understand what exactly the rate of return on capex investment is, how to calculate it, and how to interpret it. In corporate finance, the process of evaluating the potential new projects and related capex investments is called the capital budgeting process.
One of the most widely used methods in capital budgeting is the simple IRR method – or the internal rate of return.
The capital budgeting process starts with the estimation of the initial (and subsequent) investment in a potential new project – for example, a Gen AI chatbot. This is the actual capex – it’s the amount invested in the project. For Gen AI, this is estimated to be at over $1T for the entire industry over the next few years. The capex investment is considered a cash outflow.
The next stage of the capital budgeting process is the estimation of the cash inflows from the project over the specific period of time. These are the actual profits generated from the investment. Specifically, in this stage the corporation needs to estimate the number of units sold over the specific period per year, and the price of the unit sold (that’s the revenue) minus the relevant project expenses. The corporation also needs to estimate the project’s terminal value at the end of the project.
The next stage of the process involves the estimation of the project’s Internal Rate of Return – this is the rate of return at which the value of the capex investment equals the value of the project’s cash inflows.
Finally, the corporation needs to estimate the weighted average cost of capital. The new project is good only if the project’s IRR is greater than the cost of capital.
The point is that the Gen-AI related companies must address the capex ROR question within the capital budgeting process, as outlined above. They need to specify the timeline and the expected cash inflows, to be able to discuss the IRR with respect to the cost of capital.
Implications
The Gen AI tech mega-caps likely did not even go through the process of capital budgeting, that’s why I believe they are unable to answer the ROR question. The Gen AI capex has likely been an impulse investment out of fear of being left behind, in my opinion.
The Gen AI capex could turn out to be very profitable – eventually. But the problem is that the Gen AI hype is now over, as Bank of America notes. Investors cannot wait for that eventual profit, and investors need to see the results now – because it has been almost two years since the ChatGPT revelation.
Thus, in my opinion, the Gen AI is a bubble based on valuations. For example, Nvidia trades at a PS ratio of 35. Given that the top six Gen AI mega-caps represent around 30% of the S&P500, the S&P500 is likely in a bubble as well. The Shiller PE ratio for S&P500 is at 34, which is a bubble-like valuation, based on the chart below. I believe this bubble is in the process of bursting (for example, Nvidia is already down 20% from the top), and the mega-cap tech earnings this week could trigger another leg lower.
All ears will be at the earnings conference when the ROR question is likely to be asked.
Read the full article here