Bloomberg Surveillance · Sep 23, 2026
With Aaron Kennon, CEO and Co-Founder of Clear Harbor Asset Management
Software stocks have already sold off more than once this year on fears that AI will replace them, and this past week the same fear reached travel stocks after Meta launched its Muse agent.
Aaron Kennon does not read the latest scare as a reason to turn cautious on AI. He called the story maturing, and said the companies building its physical supply chain still face more demand than they can meet.
"Those supply chain bottlenecks are there, but the demand is very, very high."
Aaron Kennon, CEO and co-founder of Clear Harbor Asset Management, on Bloomberg Surveillance. He sits on the firm's investment committee, and before co-founding it he managed multi-asset portfolios for institutions and wealthy individuals at Ingalls & Snyder.
The full segment is covered here so you can skip it.
Here are the 5 calls that matter.
Key Takeaways
AI is still more tailwind than headwind for networking, electrification and power infrastructure
Shopify signed on with Meta's agent; Amazon's worry is that Meta learns from its shoppers
Wealth, more than age, decides who lets AI manage their money
Younger investors adopt technology faster but have fewer dollars to put to work
Companies that hold and organize data look like winners however the agent fight ends
The question investors are asking is which frontier models become a commodity, and he expects the answer within a year or two
1. Still a Tailwind for Power
The segment opened with tech lifting stocks near all-time highs. The host read a line Kennon had written: "AI may have started as a technology story. It is increasingly becoming a power, infrastructure, productivity, and capital market story." The first question was whether the market is repeating with other sectors the disruption trade it ran on software earlier this year.
The software scare has already run its course more than once
I think that the SaaSpocalypse has come and gone and come and gone again.
Aaron Kennon
The SaaSpocalypse is the market's name for the sell-off in software-as-a-service stocks on fears that AI will displace them. Kennon said this past week's version was Meta's Muse agent and what it could do to travel stocks, and that the same agent could help companies such as Shopify, which signed an agreement to work with it.
He sees the story maturing rather than reversing
I'm not sure it's a headwind more than it is still a tailwind for things like networking, electrification, power infrastructure. Those supply chain bottlenecks are there, but the demand is very, very high.
Aaron Kennon
2. Shopify In, Amazon Out
The host asked him to draw the line between companies choosing to come along with Meta's agent, such as Shopify, and those putting up a wall, such as Amazon. Kennon said he could not speak for either management team, then gave his read of each.
Shopify is betting on Facebook's reach
Well, I can't speak for the management teams, but clearly Shopify is of the view that with 3.4 billion users at Facebook connected now to Shopify's platform, that could be a huge economic tailwind.
Aaron Kennon
Amazon's calculation, on his reading, is about what Meta could learn from its shoppers. The risk he described is that Meta studies consumer activity on Amazon, builds its own models from it, and eventually cuts Amazon out as the middleman.
Amazon is protecting the advantages it has built
I think Amazon's very concerned that perhaps Meta is extracting insights into consumer activity on there and creating their own models that could potentially disintermediate some important sort of moats that have existed at Amazon over the past several years.
Aaron Kennon
He closed the point with a hedge: "So there's probably a nuance there."
3. Wealth, Not Age
One of the hosts said a colleague had teased her about the age of the people likely to hand their money to Anthropic's Claude, and asked Kennon whether investors beyond young college entrants are really letting AI manage their money.
Kennon compared it with the arrival of robo-advisors about 15 years ago. He recalled checking the average account at Betterment, then one of the leaders, and finding it held about $30,000 some 10 to 12 years ago.
The dividing line is money, not birth year
So that sort of speaks to maybe not an age demographic, but perhaps sort of a wealth quotient. And I think it's probably correlated as well.
Aaron Kennon
The correlation he meant runs both ways: younger investors are more inclined to use technology, and they also have fewer dollars to put to work. He expects that gap to narrow. A decade ago most people hesitated to give their email address to a shopping site, he said, and few think about it now.
Older investors are coming around
Today, I don't think we think twice about doing that. So I think there's probably a growing acceptance of this among even the older age cohort.
Aaron Kennon
4. Data Firms Win Either Way
Asked whether the rotating AI scares create a dislocation he wants to buy into, Kennon said each has to be judged on its own, since travel and software face different technologies. In software, he said, the market is learning how many companies control and segment data. He named Databricks, which is private, and said public companies compete with it.
His test is compute and agent adoption
As long as compute power is expanding, as long as AI agentic adoption is accelerating, you would think companies that are sort of in that data segment would be well positioned.
Aaron Kennon
His words for them were that they "seem to be in some sense like winners regardless" of how the theme develops. The host added cybersecurity to the list. Kennon placed it "clearly at the epicenter of all of this" and said "there are huge opportunities here in the midst of a period of uncertainty for sure with AI growing."
5. Which Models Commoditize
Another host noted that analysts see Muse as consumer-facing, which could mean faster adoption, and asked whether agents that scour the market for the best deal will force companies to cut prices. Kennon did not answer the pricing question directly. He moved it to the models themselves.
The debate investors are having
Well, I think there's a huge debate right now among the frontier models that investors are asking, how much of this model is a commodity and how much is it truly differentiating?
Aaron Kennon
His examples were Meta, because of what it knows about its users, and Google, with the consumer information that sits in Gmail and the rest of its ecosystem. The more unique that information, the more a company's model can set itself apart.
Not every model ends up the same
So one could argue that there's less commoditization among certain frontier models and more among others.
Aaron Kennon
He expects that to play out over the next year or two.
Bonus Insights
One host said he would never hand his travel to an agent
Prompted by the question of who trusts AI with their money, one of the hosts called himself an old man and said not even a human assistant would be allowed to choose his flights, because he needs the times, the airline and the seat.
Even if you gave me a world-class assistant, I would never allow anyone else to book my flights.
Bloomberg Surveillance
A second host said she simply gives a window she would like to fly in and lets someone else handle it, and the first predicted every detail-minded viewer would write in to agree with him.
Kennon's bottom line is that each AI scare so far has left networking, electrification and power infrastructure with more demand than supply, and that companies holding and organizing data look well placed whichever way the fight between Meta, Shopify and Amazon over AI shopping agents resolves.
Products, Companies & Tools Mentioned
Muse from Meta (Meta's new personal AI agent; this past week's disruption scare, and the reason travel stocks came into question)
Shopify (Signed an agreement to work with Muse; Kennon said it sees Facebook's users connected to its platform as a potential economic tailwind)
Amazon (Put up a wall against Meta's agent; his read is that it fears Meta will learn from its shoppers and build around it)
Databricks (The private company he named as an example of data businesses that look like winners either way)
Google and Gmail (His second example, alongside Meta, of consumer data that could keep a frontier model from becoming a commodity)
Betterment (One of the leading robo-advisors about 15 years ago; its average account size is his evidence that wealth, not age, drives early adoption)
Claude from Anthropic (The hosts' example of an AI people might let manage their money)
Listen to the full interview
If this was worth your time, send it to someone closer to the industry than you are.
Get the latest market chatter as it happens: