Could the AI boom be turning into a bubble?
John Borthwick: I think there may be a bubble right now in LLMs (Large Language Models) specifically, as they are reaching the capacity of what they can actually do. However, what we are currently witnessing is an intelligence wave that is a driving a fundamental transformation. We have not seen the influx of "tourist investors" that would signal we’ve hit the peak.
Dave Golob: The original Internet wave ran for six years, from the Netscape IPO in 1994 to the peak in 2000. Since we are only three years post the launch of ChatGPT, this could keep running for a while. A possible yellow flag would be potential reciprocal transactions between the hyper-scalers and other layers of the investment stack, which can be hard to identify due to a lack of transparency around some transactions.
John Borthwick: External factors are a major concern, as this technology is of geopolitical importance. Issues like rising power costs due to data centers are already gaining political traction and could quickly become a yellow or red light.
Dave Golob: Red flags would be seeing companies with much lower defensibility, negative cash flow and huge multiples starting to go public – that’s where you could have a "Worldcom” type of situation.
How transformative will AI be?
John Borthwick: While bubbles can only be seen in hindsight, I believe this moment is fundamentally different from the dot-com bubble. Firstly, as fast as we're laying the infrastructure, we're using it, not waiting in anticipation. This is because we are undergoing an intelligence upgrade in every corner of compute, which is far more fundamental than just building a network. Customers are responding positively to these changes as AI platforms seek to be "everything machines" that offer us a much more comprehensive version of search.
"We are undergoing an intelligence upgrade in every corner of compute, which is far more fundamental than just building a network."
John Borthwick, Founder and CEO, Betaworks
Greg Tuorto: We are now moving into a utilization phase – if you're not using AI at your job, you're behind. For example, the job of a junior analyst at an investment banking firm isn't obsolete, but it has changed.
Dave Golob: It is astonishing how some small AI companies are able to achieve significant revenue scale with very, very small teams, demonstrating incredible productivity gains. The idea that voice could takeover from typing as the dominant way we interact with devices also has a lot of transformative potential.
"It is astonishing how some small AI companies are able to achieve significant revenue scale with very, very small teams, demonstrating incredible productivity gains."
Dave Golob, Chief Investment Officer, Francisco Partners
What mindset shift is needed to properly assess the viability of AI stocks?
Greg Tuorto: We have to re-evaluate the old unit economic model that was based on things like the number of salespeople or workers. Instead we need to take a step back and look at what companies can become, as opposed to what they once were.
"We need to re-evaluate the old unit economic model and take a step back to look at what companies can become, as opposed to what they once were."
Greg Tuorto, Portfolio Manager, Goldman Sachs
Dave Golob: As a private equity firm, we generally view technology disruption as a risk factor as opposed to an opportunity. We do not buy companies with a high risk of disruption from an AI-type risk or that have a fundamental product problem.
John Borthwick: The moat is shifting, with companies seeking defensible positions. For example, AI models are hungry for data, so companies that own highly specific, local data that isn’t available elsewhere have a strong advantage.
How are valuations and pricings changing?
Greg Tuorto: Outside of infrastructure, everything is clumped close together in terms of valuation; making it hard to differentiate growth potential. If investors lose faith in a company's predictability, the valuation can drop significantly and they may never break back into higher tiers.
John Borthwick: For early-stage companies building "native AI applications" a big shift is toward outcome-based pricing, which aligns much more obviously with customer interests. This has significant potential to change the market outlook.
Greg Tuorto: The size of public floats in the capital markets right now is falling, with small IPO floats becoming increasingly common. This is leading to tightly grouped allocations, which is creating investor uncertainty. This is partly because last-round valuations are often too high, and nobody wants to take a down-mark to achieve a more active public trading environment.
Dave Golob: Interestingly, in the last few months infrastructure stocks have done well and application stocks have underperformed; fueled be a narrative that applications are going to get disrupted by AI. However, historically infrastructure tends to underperform applications due to lower customer retention rates. So this is a surprising development.
