Scott's Latest Blog

Demand Growth: Is it time for a Contrarian View?

Ever heard of the phrase that timing the stock market is like “trying to catch a falling knife?” I thought of that when I looked at my own investment portfolio thinking “there must be a correction around the corner.”

But then I remembered the former Federal Reserve Chair Alan Greenspan’s “irrational exuberance” comment about valuation of the market in 1996. The NASDAQ closed that day at 1,300 points, and would continue to climb until the dot com bubble burst in 2000 when it reached over 5,000 points. However, when it burst it fell back  to 1300 points, where it closed a year earlier.

I thought about his after hearing a guest on Bloomberg discussing the massive investments in AI stocks, or anything “data”, but especially investments necessary to increase electric capacity and transmission.

Our time to shine?

We know that we work in the most essential industry of our modern economy: electricity. Yet, whenever we hear the media discuss “energy” we are used to something regarding oil, or perhaps natural gas. This has been true even for those of us who have been working for years on making electricity more efficient through competition.   

By my reckoning, that changed in the first quarter of last year when the Artificial Intelligence (AI) and Data Center boom exploded. After that moment, everybody was suddenly aware that demand for electricity would likely grow 7% or more per year for several years. That “belief” would evolve in the remaining months of 2024 to what seemed a “certainty”.

Consequently, media, politicians, regulators, even your barista became obsessed with the need for more and more electric capacity. This reaction was rational. To a point. Have we passed that point?

I don’t know. There is a reason it’s so hard to predict macro-economic trends. The variables in an economy the size of ours are so large and varied.  A colleague, who is a PhD in Micro-economics,  referred to Macro predictions as “witchcraft”.

However, when one considers the longish run of the current Bull Market, roughly 16 years if you remove the Covid blip, one begins to look more closely at what is driving valuations? The “normal” economy (Main Street, not Wall Street) appears to be cooling. Unemployment is beginning to creep up as the level of job creation appears to be slowing. I’ve heard more than a couple of commentators say that if one took out investment related to AI and Data Centers (including power), the real economy represented by GDP growth would be running flat. In other words, zero growth rather than the estimated 2-3% growth of the last quarter. When one sector of the economy represents that much growth, that is a lot of money.

No doubt, a great deal of investment in this area is warranted and will represent real economic value in the future. Even if one says the “dot com” bubble was excessive – it was – the result did represent infrastructure that was needed in the digital economy of the last 25 years. But as our industry has become a secondary focal point of investment, parts of our industry have come in for criticism for not anticipating the expected load growth as well as the increased cost of electricity. The frequently declared expectations of electric demand might require some examination.

The “lumpiness” of power supply & demand

Famed power market expert Bill Hogan probably coined the term “lumpiness” to describe electric power supply. It tends to come in big batches that sometimes exceed average demand. This phenomenon can sometimes put the regulatory community “to sleep” on power procurement issues until a crisis erupts. It can also lead to procurement that can appear to be excessive[1].

As our industry rushes to procure new power resources of all kinds - thermal, renewable, storage, nuclear, not to mention the price for transmission equipment - the price of everything in the supply chain is going up. Like many of you, I have heard the stories about the time it takes to get new natural gas turbines (2030?) and how much they will cost. Uncertainties about tariffs and global demand have pushed up prices for everything from copper to transformers.

Efforts to get large customers that are driving power demand for AI and data centers are correctly in my opinion, being asked to bear more of the costs. However, as they do this, prices for procurement to meet other customers’ demand will naturally go up (whether in commodity prices or finance costs). This forms a “Hobson’s choice” for regulators and policy makers[2].

As a lifelong competitive market advocate, I am not going to argue against legitimate market prices – whether high or low. Higher power prices have helped investment in keeping thermal assets needed for reliability and giving life to the possibility of new nuclear investment. Solar and storage resources – the fastest assets to procure – will continue to help meet new demand. But will the tidal wave of demand growth continue for the next decade as forecast by so many commentators? History suggests it might not.

While the AI industry needs to acquire more power than traditional digital demand, the experience of Moore’s Law[3] might be instructive in moderating expectations of high growth for years to come. Secondly, an economic slowdown will occur at some point. While it would be hazardous to suggest when such an economic event will happen, we know that economic growth is not linear.

So, what if  AI/Data Center growth isn’t  as huge, for as long as some have suggested, what does one do? Certainly, new investment is necessary. However, competition in procurement is always needed, but especially when everyone appears to be buying. Hedging tools in procurement such as “safe harbor” pricing seems prudent. Like all forms of “insurance” these can sometimes be seen as expensive, but it can contain volatility. Bottom line, I’m not suggesting I’ve got answers or even that my caution will prove correct. But every now and then, some contrarian thinking is a good thing.

As General George Patton once said; “If everyone is thinking the same thing, somebody’s not thinking”.