In November 2022, the launch of the popular artificial intelligence (“AI”) app, ChatGPT, made generative AI applications available to a large public audience. There are multiple applications of AI in the real world, with varying degrees of capability. Machines may never get to human-like self-awareness, but generative AIs are getting better and better at cognition, which is the ability to incorporate past actions and learning experiences into future practices. AIs are now generating text, poems, research papers, drawings, and videos. In order to do all this, these models require to be fed vast amounts of data, which entails using a significant amount of computing power and thus energy. Digiconomist estimates that by 2027, worldwide AI-related electricity consumption could increase by 85 to 134 TWh annually. This would be comparable to the annual electricity needs of countries like the Netherlands, Argentina and Sweden. The potential growth highlights that we need to be very mindful about what we use AI for. It’s energy intensive, so we don’t want to put it in all kinds of things where we don’t actually need it.

Government efforts to encourage and develop renewable energy resources have intensified in recent years. But multiple factors have simultaneously driven up demand and prices for traditional petroleumbased energy resources. These include OPEC production cuts, geopolitical turbulence in Eastern Europe and the sharp rise in interest rates which has driven up the cost of financing new renewables projects. To the surprise of economists who expected a recession based on rising interest rates, most major economies have avoided steep economic downturns. Indeed, many economies are growing faster than expected, resulting in rising demand for energy from consumers and businesses. Global efforts to get work-from home employees back to the office are contributing further to rising oil demand. This all adds up to surging petroleum demand in what was supposed to be by now a renewables world.

August has seen the unwinding of the seven-month rally in technology and digital stocks that, to varying degrees, lifted equities across most sectors this year. Stocks often struggle in the summer; but usually that means they are lethargic, not heading notably lower. What caused stocks to give up big parts of their year-to-date gains? AI mania drove much of the stock rally in the first half of the year. And in any stock-market mania, unbridled enthusiasm gives way to second-guessing at some point. Consumers and businesses show few signs of spending fatigue, contributing to inflation’s persistence. We think the difficulty in getting inflation down to the 2% target level, along with some buyers’ remorse on AI mania, has prompted investors to take profits.

Data on NYSE margin debt (borrowed funds from brokerage firms) has been an excellent indicator confirming bull and bear markets. It is based on monthly data, so there is a time lag, but it is impossible to catch the top or the bottom in any case. Margin debt bottomed at the end of 2022 and has crossed above its 10-month average for an intermediate-term to long-term buy signal. This reverses the sell signal from late 2021 (when its uptrend broke) and a confirming signal when it broke the 10-month at the end of January 2022. Prior buy signals for the S&P 500 were given in June 2020, July 2016, September 2012, July 2009, April 2003, May 1995, May 1991, December 1982, and June 1975. These were all good signals. Sell signals were produced in October 2018, August 2015, July 2011, Sept. 2007, October 2000, November 1987, November 1981, and April 1973. These were good as well, but the reading was late for the 1987 crash.