Stay informed with free updates
Simply sign up to the Artificial intelligence myFT Digest — delivered directly to your inbox.
Goldman Sachs wishes to make it clear that it’s not anti-gen-AI. Several of you may have got the wrong end of the stick after the bank’s publication in June of a report called “Gen AI: too much spend, too little benefit?” — but note the question mark! They’re just asking questions!
Goldman strategist Peter Oppenheimer and team have a sequel out this morning that seeks deeper understanding of what it calls “tech’s rational exuberance”. Here’s the summary:
The technology sector has generated 32% of the Global equity return and 40% of the US equity market return since 2010. This has reflected stronger fundamentals rather than irrational exuberance. The tech sector globally has seen EPS rise c. 400% while all other sectors together have achieved c. 25% from the peak pre-GFC.
So, not a bubble. But there are caveats.
The problem with any good story is that it can “amplify interest to the point of monopolising investor attention at the expense of other opportunities, and leading to unrealistic expectations about future profits and leaving companies vulnerable to a sharp derating,” say Oppenheimer et al.
Disruptive technologies nearly always go boom-bust-boom, with big players doing useful things only emerging after the initial bubble bursts under the weight of opportunistic me-too companies. The same risks and opportunities appear each cycle, Goldman says:
Historically, investors over-focus on the originators, understate the impact of competition and overstate the returns on capital invested by the early innovators. At the same time, investors tend to underestimate the growth of new entrants to the industry that can piggyback off the capex of others, enabling them to generate new products and services. Valuations often also understate the opportunities that can accrue in the non-technology industries that can leverage the technology to generate higher returns in existing, as well as in new, product categories.
Survivors drive their ploughs over the bones of the dead, says Goldman, though it phrases it this way: “the infrastructure left behind in the wake of the initial investor surge and capex leads to the emergence of new products and services”.
AI is not exactly like previous innovation bubbles (books, canals, radio, the telegram, radio, etc) because most of the dominant companies were the winners to emerge from the previous bubble, continues Goldman. Raking in money from advertising means they can burn capex — but even the current rates of spend won’t give them insurmountable monopolies over whatever emerges, the team argues:
While the protective ‘moats’ around the current AI winners are significant, and valuations are not bubble-like, the number of new patents in this area is growing rapidly, suggesting that new competitors will emerge and costs will come down. [ . . . ] While the hyper-scalers have huge scale and ability to invest in proprietary AI models, cheaper open source alternatives are emerging at a very rapid rate. The website Hugging Face, which is a network for enthusiasts, already has around 650,000 models, suggesting that the typical pattern of large-scale capital growth and competition is happening in the AI space, just as occurred in previous waves of technology.
Just as competition is often underestimated, the returns on innovation capex are typically overstated as the marginal cost of the technology falls and capacity increases over time, Goldman adds:
In the case of most major technological innovations throughout history, while the potential may be obvious, it is rarely clear in the early stages what business models will ultimately dominate to scale and commercialise the technology. This was evident in the early days of the internet. While there was widespread and broad speculation in any new company that offered potential exposure to the industry, the incumbent winners were generally seen to be the telecom companies. They were viewed as a relatively ‘safe’ route to the potential fortunes that the internet may generate compared to the more speculative unprofitable dot-com companies. [ . . . ]
As with other examples in history, the problem was not a miscalculation of the growth potential of the technology, but rather that investors had attributed too much future value to the companies that had built technology and infrastructure to provide it. In this case, like many others before, the ultimate winners were the companies that could ‘free ride’ off this spending and utilise the capacity to build business models that could leverage the technology and provide new products and services. Many of these winners did not emerge until the onset of the smart phone in 2006 and the onset of apps which then spawned a growing industry of platform companies, ride sharing, social media and so on.
To reiterate: not a bubble (relative to previous bubbles, which were definitely bubbles):
Nevertheless, “the AI winners of today are no longer capital-light businesses,” says Goldman. “Just as we saw with the networking companies of the internet, AI is driving a major capex boom and threatens to stifle the high rates of returns that have characterised the sector over the past 15 years and which current valuations imply will continue.”
And there’s not much evidence that technology capex extends the life of intellectual property assets, nor that first movers like ChatGPT have the staying power for commercial success:
Oppenheimer and team give a primer on concentration risk and the regularity with which market leadership changes, which we won’t repeat because for regular readers will be familiar territory. If you want to read Goldman’s section about the longevity of pop music artists and stock market winners, it’s in a screenshot here.
There’s also a bit about how tech is fundamentally deflationary. This is probably good for dodging antitrust scrutiny, but not so good for defending margins:
So in conclusion, investors should diversify. Rather than adding more Nvidia, a better way to play the gen-AI theme might be to buy . . . .
LVMH?
Because of Bake Off?
In past technology cycles, the second-round impacts on work and society often drive new areas of consumer growth. It is likely, for example, that more AI will mean more demand for fact-checking services. The ability to work more productively from home may mean the regentrification of shopping and entertainment in neighbourhoods close to large-density populations. The growth of artificial immersive entertainment may also boost demand for experiences in the real world. This might reflect the growing popularity of goods and services that are seen as ‘authentic’ or nostalgic. Retro ‘crafts’ are growing in popularity, whether it be the growth reality TV programmes where contestants compete in baking, spelling, sowing or even ballroom dancing competitions. [ . . . ]
In the 21st century, in a highly digitalised world where almost everyone is connected to the internet and the cutting edge of technology threatens to displace jobs and companies, it is meaningful that one of the biggest company in Europe is LVMH. This is a company that sells the value of heritage in historic brands. [ . . . ]
As the ubiquity of technology increases and individuals increase their reliance on technology as they communicate via networks, the value they place on ‘authenticity’ and human connectivity — which can evoke a nostalgic image of a simpler, pre-digital life — is likely to grow.
Righty ho.
For any second-wave AI company that wants to raise money in the current not-a-bubble (as well as for pretty much any other company doing nearly anything else), Goldman Sachs is available for all your investment banking needs.
Further reading:
— AI hype is not a replay of the dotcom bubble, but it’s a remix (FTAV)
Read the full article here