Listen to the audio version of this article (generated by AI).
Meta’s Muse becomes the fastest AI app since ChatGPT… the “agentic bank run” threat… plus, the latest headwind to AI companies
Since launching on Sept. 8, Meta’s (META) Muse AI agent has shot to No. 1 on both the U.S. App Store and Google Play – and is still there.
By late last week, third-party trackers at Sensor Tower pegged downloads north of 3.4 million, outpacing even ChatGPT’s blistering early adoption.
Here’s JPMorgan:
While it is still early, we believe that Muse has the potential to become the most widely used consumer AI application since ChatGPT.
So, what’s the big deal here?
In a word: agency. Muse doesn’t just answer questions. It runs continuously in the background, operates its own virtual computer and browser, coordinates multiple sub-agents across a task, and works across your devices.
It already plugs into Gmail, PayPal, Shopify, Walmart, Best Buy, and Expedia – and connects to your bank through Plaid, the same plumbing your budgeting app uses.
In AI Revolution Portfolio – the AI-focused investing service helmed by Louis Navellier, Luke Lango, and Eric Fry – the team put it this way in last Thursday’s issue:
The distinguisher here isn’t just intelligence – it’s access and persistence.
The more access Muse gets to your email, calendar, shopping, computer, and physical surroundings, the more tasks it could potentially handle on your behalf.
And this is where this stops being a tech story and becomes a disruption story.
AI agents as a commoditization machine
If an agent can comparison-shop as well as a person – or better – it could reshape how consumers pick financial products, insurance, travel, retailers, you name it. So, instead of sticking with the bank or brokerage you’ve always used because finding a better/cheaper option is a pain and switching is a hassle, you could hand the chore to an agent with no loyalty (or search fatigue) at all.
Think about the potential fallout…
Entire industries have been built on the idea that customers are loyal, inattentive, or simply too busy or unwilling to shop around. Meanwhile, businesses spend billions cultivating brand recognition, loyalty programs, slick apps, and manipulative design tricks, all meant to nudge a distracted human toward a choice that isn’t always the best-value one. It works because humans are emotional, forgetful, and easily anchored.
An agent is none of those things.
An agent doesn’t recognize a logo, care about a funny Super Bowl ad, fall for a “limited-time only” banner, or forget to cancel the free trial. It just optimizes on cost and value, full stop.
This means that the agentic era could become a powerful commoditizing force – stripping the premium from any product whose advantage rests on branding and inertia rather than genuine, provable value.
The early data suggests consumers are ready to let it. A recent Checkout.com survey found 57% of consumers would let an AI shopping agent switch brands if it found a better-value option. And in a Deloitte survey, 44% of retail executives said they expect AI to weaken brand loyalty by pushing choice toward value and fit over brand recognition.
Here’s Bain & Company:
Shopping agents could relegate retailers to drop shippers or commoditized fulfillment mechanisms.
Now, let’s be clear: even if this is what’s coming, we’re in the earliest possible phase of this shift.
Agents haven’t yet proven they shop better than people at scale, and companies won’t go quietly. Some will fight back by blocking agents outright, as Amazon.com Inc. (AMZN) has already done with Muse. Others will race to game the machines with a new kind of marketing built for algorithms instead of eyeballs. So, this won’t be a clean, overnight rout.
But the direction is hard to argue with. And it points toward a surprisingly long list of industries whose moats are built on friction, which leaves them vulnerable to agentic disruption:
- Insurance – Millions of drivers and homeowners overpay for years out of pure inertia. An agent re-shops every renewal.
- Credit cards – Rewards, teaser rates, and balances that never get moved. An agent automatically chases the best terms.
- Wireless, broadband, and streaming – The home of the “loyalty penalty,” where existing customers pay higher rates than new ones. An agent renegotiates or cancels without a second thought.
- Airlines, hotels, and online travel – Loyalty programs vs. an agent that sees only the best itinerary at the best price.
- Consumer brands and retail – The brand premium erodes the moment an agent weighs it against a cheaper equivalent on identical specs.
- Banking and brokerage – The cheap-deposit model, which I’ll dive deeper into in a second.
Bottom line: The businesses most exposed won’t be the ones offering the best deal – they’ll be the ones profiting from the fact that you’ve never gone looking for a better one.
Driving it home: the “agentic bank run”
Let’s look at an example that’s already made waves in millions of investor portfolios.
The banking industry has had a rough month. In yesterday’s Digest, we walked through how the fastest jump in the 10-year Treasury yield in years is squeezing the financial system. As of last week, regional banks, as measured by the SPDR S&P Regional Banking ETF (KRE), had fallen nearly 10% from their recent highs.
This agentic fear is the sharp, new worry on top of woes about interest rates.
Take last Tuesday, when the S&P 500 Financials Index fell more than 2%, while Charles Schwab Corp. (SCHW) and several brokerage names dropped more than 5%. That selloff wasn’t due to rates – it was because of Muse.
Here’s Apollo Global Management’s chief economist, Torsten Slok, from his piece titled “Is an Agentic Bank Run Coming?”:
If every household used AI agents to optimize the return on their cash balances, banks could lose a large share of the cheap deposits they rely on to make loans, which would be a problem for the entire financial system.
To make sure we’re all on the same page, banks earn money in part by paying you almost nothing on the cash in your checking account (the national average is a laughable 0.1%) while lending out your money at a much higher rate.
Now, new fintech companies already pay far more than traditional banks. Slok notes SoFi Technologies Inc. (SOFI) offers 4.5% (restrictions apply). What’s kept the money parked at traditional banks is friction. Moving it is a chore that most people never get around to.
An AI agent erases that friction.
Back to Slok:
Muse and similar agentic AI assistants could soon sweep household cash automatically into accounts paying 3.3% to 5.0%, instead of the 0.1% national average on checking accounts.
How willing would you be to spend, say, 10 seconds prompting Muse to move your money if it earned you 5% instead of 0.1%?
As I wrote earlier, this isn’t just a tech story – it’s a disruption story.
Now, before we bury the American bank: this threat, while real, is still theoretical. While it’s a genuine risk to banks’ margins, agents won’t flip the entire deposit base overnight, giving banks time to respond. They’ll roll out their own AI assistants to keep customers (and their cash) where they are.
What this means for the broader AI trade
In yesterday’s Digest, we highlighted Luke Lango’s “5% line in the sand” related to the 10-year Treasury yield and its eventual impact on the AI trade.
In short, the AI sector – and stocks in general – ultimately depends on a healthy consumer whose spending funds Big Tech’s revenues, which in turn helps fund the enormous AI buildout. Break the consumer, and eventually, it impacts the capex.
Muse is the other side of that coin. Mass adoption is precisely what turns all that AI spending into actual revenue.
Here’s Luke:
Muse may not be advancing the frontier in the same way as Claude Fable or ChatGPT Astra, but it is advancing mass adoption – a critical ingredient in turning AI investment into revenue and sustaining the infrastructure spending cycle…
The infrastructure implications extend well beyond the launch downloads…
If Muse sustains its engagement, it could become a durable, growing source of demand that was absent from many forecasts a month ago.
The market has spent two years worried about whether the AI infrastructure boom would ever pay for itself. An app that hundreds of thousands of people reach for every single day is the beginning of an answer – one that’s bullish for AI.
But – and in this market there’s always a “but” – that same buildout just drew fresh fire in Washington…
The latest political fallout
On Sunday night, Sen. Elizabeth Warren, D-Mass., joined by Sens. Tina Smith, D-Minn., Jeff Merkley, D-Ore., and several colleagues, sent letters to the CEOs of Meta, Amazon, Alphabet (GOOGL), and Microsoft (MSFT) demanding an accounting of the tax breaks they’ve claimed on AI and data-center spending thanks to 2025’s “One Big Beautiful Bill.”
The numbers they flag do catch the eye…
Meta paid $2.8 billion in federal income tax in 2025 – down from $9.6 billion the year before – while earning roughly the same profit. Microsoft’s current federal income tax expense fell more than $11 billion from fiscal 2025 to 2026. Amazon’s tax expense dropped nearly $8 billion, and Alphabet’s fell more than $7 billion.
All told, corporate tax payments are running about 25% lower this year, even as revenues climb.
Here’s the heart of Warren’s letter to Mark Zuckerberg:
This enormous tax cut appears to have been driven in significant part by President Trump and Republicans’ tax breaks subsidizing your spending on AI…
Your company has spent lavishly to stay on the good side of President Trump, and it appears that you are now seeing your investment bear fruit.
Remember – the AI buildout hasn’t just been powered by demand and cheap-ish capital. It’s also been powered by generous tax treatment – the immediate deductions that let Meta write off much of its $72 billion in capital spending last year.
The Joint Committee on Taxation estimates roughly $67 billion in retroactive breaks are available to companies in 2026 alone. Strip those incentives away – or even threaten to – and some of the math behind the AI buildout changes. And with data centers a flashpoint heading into the 2026 midterms, this is unlikely to be the last shot fired.
So, we end today where we’ve ended so many Digests in recent months – with genuine crosswinds
Muse shows the AI flywheel finally spinning the way the bulls promised – real adoption, usage, and recurring demand for compute. That’s a huge, encouraging tailwind.
But much of what consumers like about AI – savings and value – is a headwind for the profit margins of companies that have grown fat and happy on consumer inertia. Meanwhile, Washington is beginning to eye the subsidies propping up the buildout.
This is the kind of market where it pays to know which companies sit on the right side of the disruption – the ones collecting the inference tolls – versus the ones, like the banks, whose moats face new challenges. For Luke’s latest research and stock recommendations to navigate this challenge, .
In the meantime, we’ll keep tracking all of it – Muse’s momentum, the deposit-flight threat, and Warren’s tax investigation – and will report back.
Have a good evening,
Jeff Remsburg
(Disclosure: I own MSFT, AMZN, WMT, GOOGL)
