Did this major player just signal an end to the ‘picks and shovels’ stage of the AI gold rush?
Nvidia became the most valuable company in the world by a near-perfect execution of the principle that during a gold rush, you should sell shovels. The chipmaker’s GPUs have become the shovels, picks, roads, bridges, or any other hard infrastructure metaphor, for the ongoing generative AI theme.
Massive tech companies have invested hundreds of billions of dollars into the development of generative AI models and AI software, and a significant portion of that investment has flowed into Nvidia’s accounts. As Nvidia printed money on AI chip sales, investor capital flowed into its stock. All the while the companies buying its chips have grown significantly, but on the promise of future AI business rather than real AI revenue. That’s why many were struck when Softbank announced it would exit its entire $5.83 billion stake in Nvidia to invest that money in other AI companies.
“There's a reason that this is a big story. There's a big signal that comes from Softbank from what they do because they're such a massive investor,” says Elliot Johnson, Chief Investment Officer at Evolve ETFs. “Softbank making this change is causing now everybody to rethink their framework for how they think about AI investing, which is moving from picks and shovels to software services and other elements of the value chain.”
Johnson explains that the AI story has been so dominated by Nvidia for so long, and that the other names most associated with AI have been the ‘hyperscalers’ building data centres and AI software. Those companies have been more rewarded by investors for their investments rather than their cash flow. He notes, though, that the move by a player as large and culturally significant as Softbank might signal a shift towards more downstream revenues from the use of AI.
Likening this moment to the development of fibre optic cable networks at the start of the internet boom in the ‘90s and early 2000s, Johnson compares Nvidia to Cisco. Cisco was the investor darling of that early boom as capital flowed to cable networks, the ‘picks and shovels’ of that moment. Eventually, however, the investment value of the internet has been realized more through downstream names like Amazon and Meta, rather than Cisco or its Canadian equivalents in Nortel and JDS Uniphase. Johnson argues that Softbank’s repositioning is being done with a view to investing in downstream value additive names.
While a major player like Softbank might seek to identify specific names that can add alpha to its AI investment portfolio, Johnson argues that retail investors should continue to take a total portfolio approach to the theme. He argues that exposure to Nvidia and other AI infrastructure names is still important, as demand for these chips will continue to grow as the chips themselves become more advanced. However, he notes that much like the adoption of the internet, the software developers and eventual end users of AI will realize value there as well. Rather than trying to select the winning point on the value chain, or the winning names within the theme, broad allocations can give retail investors the exposure they need.
“I think this is a classic example of time in the market beats timing the markets and having a diversified approach to the investment in the theme is going to stop you from guessing and getting it wrong,” Johnson says.
While softbank appears set to back the hyperscalers, there has been some noise made on corporate bond markets about the financing deals and sheer quantity of spending associated with AI infrastructure. Johnson again likens the question of misallocated capital to the 1990s telecom and internet boom, noting that much of the fibre first laid in that boom was dark for years. He argues, though, that even if AI capacity slightly exceeds demand in the short-term, the nature of AI’s computing and energy requirements, as well as its myriad applications, means that demand will likely fill excess supply quickly.
For advisors fielding questions about thematic shifts in AI and what major figures and names do to grab headlines each week, Johnson says that paying too much attention to those moves can prove distracting. Instead he says that they and their clients need to remain disciplined, and not chase the latest big move, because they’re probably already too late.
“Advisors and clients need to be invested across the theme, hardware, software, services,” Johnson says. “You can do that yourself or you can use a fund like ours to get diversified exposure you need to stay invested. The value accretion is going to happen over time and if you're trying to time when it happens this month versus next month you're going to miss because, inevitably, we can't predict those things.”
