Best Laid Plans: The AI Build Super Cycle
Clients of the Firm,
Robert Burns the Scottish Poet famously wrote, "The best laid schemes o' mice an' men gang aft agley, An’ lea’e us nought but grief an’ pain, For promis'd joy”. This line from To a Mouse conveys that even the most carefully made plans can fail or go poorly.
Burns’ lyrical prose seems appropriate to consider in this moment of the AI build super cycle. The euphoric rush of capital towards AI scalers continues to produce an elevated market level overall with P/E levels reaching the high end of normalized ranges for technology companies. Market structure remains heavily weighted to these firms and thus the major U.S. indices are reaching new highs to reflect this multiple expansion as well. At 6,700 the S&P 500 is currently trading at 25 times projected 2025 and 22 times projected 2026 earnings. The justification for this multiple expansion thesis has some complex competing forces including productivity expansion in the face of labor contraction with efficiency leading to higher profits.
Four major AI hyper-scalers, Microsoft, Alphabet, Amazon and Meta, have committed a combined $364 billion of capital investment to build and upgrade AI data centers. Much discussion has been heard on Wall Street around the need for AI capacity, the enormous spend on plant and equipment, the imbalance of demand versus supply and the winners providing the “picks and shovels” of the AI server racks and requisite chip sets. Further excitement has been leveled at the need for increased energy infrastructure and commodity access for generation resources to power these new AI data centers.
More ink has been spilled on the application layer of AI which will likely enhance productivity, especially at very large companies. Here, labor forces will be reduced and replaced particularly in the area of knowledge work, coding and entry level administrative, customer service and business development roles. It is not impossible to imagine productivity gains manifested in cost structure reductions in excess of 5-10%. Similar to the extended corporate tax cuts, these productivity gains could result in permanently compounding cost benefits that will be corporate earnings accretive.
Longer term, it is also feasible to see AI enabled robotics replacing front line service workers in the retail and QSR industries. Material capex will be required to replace the workforce, but also the management need to navigate the significant societal upheaval this shift would cause. Manufacturing is an early adopter of this innovation and the transition is already underway in many major manufacturers.
One economic offset to this potential workforce reduction is the challenge of how to deal with the unemployment and re-employment of workers in both white and blue collar jobs displaced by AI. It remains unknown what the net economic and policy impact of this transition will be. Most agree the displacement will result in at least a temporary headwind for the economy as new training and retooling of workers and innovation jobs take time to emerge.
Corollaries to the dot com boom abound. These include the elevated P/E ratios of tech leaders and the clear build super cycle. The results of the internet emergence cycle were the bust of over capacity followed by the birth of the web innovation economy and the eventual transition to Web 2.0, mobility and cloud infrastructure. As you may recall from our last letter the arc of innovation tends to lead up and to the right, though the timing is always uncertain.
Investment strategies to deal with such an environment vary from sitting it out in investment grade fixed income and clipping coupons, to fully embracing the risk and volatility associated with a technology super cycle and riding it where it may lead. As usual, our approach is somewhere in the middle of these two extremes. Tail risk hedging, for example, can allow one to ride the market rollercoaster while protecting against catastrophic loss. Traditional asset allocation is also a valuable tool to balance risk allowing an investor to seek upside returns and absorb volatility with a portion of invested capital, while protecting another portion in interest bearing investments.
While it is far from certain that we will be left with “naught but grief and pain”, it is worth noting that technology cycles do not always play out as planned and unexpected winners and losers can emerge from the best laid plans. Our task remains to navigate risk and reward according to our clients’ goals and tolerances. We appreciate the opportunity to continue to serve in this capacity with your funds.