Kai Wu, founder and chief investment officer of Sparkline Capital, joined Phil Rosen on Full Signal to discuss the K-shaped stock market, the AI infrastructure trade and whether AI labs like Anthropic and OpenAI have real moats.
His core takeaway is that the AI build-out looks like a classic capital cycle, and the best opportunity may sit with AI early adopters that trade at laggard valuations rather than with the picks-and-shovels names.
Why the stock market is so K-shaped right now
Wu said the entire stock market is being "held up by the AI trade." He noted the Magnificent Seven are now 33% of the S&P 500, while many software names are down 50% to 80%.
By his estimate, Nvidia and Micron explained about one-third of the earnings contribution last quarter. I noted that the share of individual stocks outperforming the index has dropped to 30%, one of the lowest readings since 1990.
Wu sees that as largely downstream of "a single macro effect, which is this AI capital cycle."
The AI capital cycle and the circular financing risk
Wu, who worked under Jeremy Grantham at GMO, said transformative technologies from railroads to 1990s telecom tend to spark an investment boom that becomes a prisoner's dilemma, where every company keeps spending because its rivals do.
He added that the free cash flows of the hyperscalers are "basically zero now," so the build-out is increasingly financed with equity, debt and SPVs.
On circular financing, his worry is correlation, not fraud: "it ties everybody to the same ship," so trouble at one company like OpenAI, CoreWeave or Oracle could spread through the ecosystem.
Why Wu is bearish on AI lab moats
"I'm pretty bearish on their moats," Wu said of the frontier labs. ChatGPT launched far ahead of rivals, but now several labs sit near the frontier and Chinese open-source models trail by three to six months at a fraction of the cost.
Switching costs are low, and casual users don't need the top model. "Who would really want to pay an additional $200 a month to be able to post better memes?" he asked.
In his view, the winner is "going to be the guy who has the most compute," which makes raising capital critical for the labs. For a different view on chips, see VanEck's Nick Frasse on why AI chips aren't repeating 1999.
Are chip stocks still cyclical?
Wu noted that semiconductors have historically run on boom-bust cycles: prices spike, everyone adds capacity, supply overshoots and prices collapse. I noted that many chip and memory insiders now call this a secular change.
His biggest risk is efficiency. "Einstein didn't need $100 billion," he said, and a leap in model efficiency would mean far less compute per query.
He pointed to the dot-com era, when the telecom index fell more than 90% and the fiber builders "ironically did not capture the gains." Companies like Netflix and Google that used the cheap capacity did.
Where Wu sees opportunity: AI early adopters
Wu splits the market into three groups: AI infrastructure, AI early adopters and laggards. Infrastructure now makes up almost half of the S&P 500, so he thinks investors should be a bit more cautious there.
"I think that the middle category is probably the most interesting," he said. Early adopters trade at roughly laggard valuations and aren't funding the build-out, yet could benefit if AI delivers.
He described that as a "free call option." If there's an overbuild, he expects users of AI, not its builders, to capture most of the value.
I brought up two names from my Best Ideas Club interviews, ADP and FactSet, as businesses with sticky customers that are adding AI. Wu didn't comment on specific securities but said companies with complementary intangible assets like brand, customer relationships and switching costs tend to survive disruption.
Sparkline's two intangible value ETFs, ITAN for U.S. markets and DTAN for non-U.S. markets, look for companies that are cheap relative to their IP, brand, human capital and network effects.
Tune in to the full conversation with Kai Wu on YouTube, Spotify and Apple Podcasts.
For more on the AI cycle, read Bloomberg Macro Strategist Tatiana Darie on the biggest risks and opportunities in the market today, and subscribe to the free Opening Bell Daily newsletter for a daily briefing on stocks, macro and Wall Street.
This article is for informational purposes only and is not investment advice. Views and stock opinions are those of Kai Wu and market figures are as stated by the guest at the time of recording.