It’s impossible to cover AI and not talk about Meta Muse. This is not an endorsement, just some of my observations and information. Just yesterday Meta has announced Meta Enterprise Platform.
This at a time when AI has literally disrupted media and creatives1.
AI and the Fall of the Creative Class
Young people might be deskilled in how to code and how to write simultaneously, within just a micro generation. That’s not by the way, the empowering AI we were promised.
Will Agentic Swarms Disrupt Core Skills of Human Literacy and Intelligence?
ChatGPT Era is official Over
Whether we like it or not, agentic swarms are here. In terms of AI Supremacy on the consumer (B2C) front, the last four years have been the Chatbot era where ChatGPT of OpenAI was the dominant leader2 (late 2022 to late 2026). This weekend + all Monday I’ve been going over Meta’s Muse product and I can say with some confidence that the ChatGPT era is now really over. We’re about to embark for likely the next few years into a journey of Personal Super-Intelligence Agents (PSIA). Meta is SOTA in PSIA, let me explain:
Muse’s Rapid Ascent and a new Personal AI Race
I predict Muse will have around 1 million users by November, 2026. In 2027 it could continue to grow quite fast. The AI industry is a bit confused about the branding, not a “companion”, but maybe more like a teammate or casual personal assistant that is supposed to help you optimize your personal life that will become increasingly capable to actually accomplishing tasks for you.
When do Personal Super-Intelligent Agents Arrive?
The new era of AI begins when two things occur in Meta’s timeline: the arrival of its Watermelon model (due in late October) and the finishing of its Prometheus (near Columbus in New Albany, Ohio) datacenter coming fully online in the next few months. By then adoption of Meta’s Muse agent will be even more impressive than its first month. So this new era arrives circa or around December, 2026. Right after Anthropic’s IPO a new race in consumer AI kicks off (that’s right after U.S. midterms). Simultaneously ChatGPT will be losing marketshare to open-source models that are taking more token marketshare especially in the last six months3. OpenAI is thus in deep trouble in 2027 for a few major reasons:
Signs of Cracks in OpenAI Dominance are starting to show
Pressure in Consumer AI from Meta and SpaceXAI.
Pressure from Chinese open-source models on cost and token marketshare.
Late to IPO, and markedly over-taken in ARR and growth by Anthropic in Enterprise AI.
With Google late to deliver their next frontier model, among BigTech incumbents only Meta has made a serious challenge to be a frontier player in both LLMs and consumer AI products. The Meta AI app even before Muse has gotten very decent traction the past nearly 1.5 years. Muse is currently available only in the U.S. and Canada.
Meta’s Scale Advantages Up-sell its First Comer Personal Super-Intelligence Move in Muse
Muse was launched on September 8, 2026 or around 3 weeks ago. By the time its Watermelon 🍉 model drops it will be about seven weeks and where more countries will likely have been rolled out to try the first iteration of Muse. Meta has two major advantages over SpaceXAI, OpenAI and others trying to be players in the next personal AI market. These are huge numbers of daily active users and a warchest of Ad revenue. Meta reports an average of 3.60 billion “family daily active people” (DAP). Meta is also the internet’s dominant digital advertizer.
Meta’s full-year total revenue is projected to be in the area of $245 billion and Muse could materially accelerate their advertising spend for small businesses at scale. Meta also has resources to spend on marketing their new AI products that most AI startups don’t have. Meta is widely expected to have significant negative FCF by the end of 2026. The compute spend on Muse will make META 0.00%↑ a difficult company to value, and likely far more risky to invest in. All of this inspite of Meta’s impressive adoption rates and the performance of the app on leaderboards.
Token Spend and Machine-to-Machine Inference (M2M Internet)
The PSIA (personal super-intelligent agents) market will radically accelerate token use and only truly profitable companies will be able to compete due to the high compute (costs to subsidize), talent, operations, training and RSI costs. These systems will be designed to keep getting better and more personalized to the user including up-selling them to new more specialized AI products in the next few years and nudging them to use Meta’s digital advertising to grow their side hustles, entrepreneurial projects or existing small businesses.
Products like Muse will contribute to a new definition of what it means to be compute constrained in the late 2020s. It’s going to be so expensive, few will be able to follow.
Why is Meta Muse the First Mover?
Meta’s Muse is the consumer culmination of Meta’s ambitions in the Metaverse and AI merged. According to my calculation Meta will have spent upwards of $180 Billion on this PSIA project by 2027. They will have gone into significant debt to do it.
That’s multiples more than Grok Bot of SpaceXAI or OpenAI’s Codex Bot (“o”) have to put into a persistent agent experience with dedicated memory and a cybersecurity hardened and encrypted Virtual Machine space for swarms of agents to work dedicated for you. Meta also has such a treasure trove of data to enable Muse to help us maintain and “show up” for our friends, family and personal connections. This is language Mark Zuckerberg specifically used in his multiple recent interviews and podcasts.
It’s actually fairly interesting to think of Muse as the first major project of Meta’s new Meta Superintelligence Labs founded around 15 months in.
Meta’s Muse Timeline
It only took AI to go from ChatGPT to Muse in 45 months. Not fast at all when you think about it. ChatGPT was released on November 30, 2022, and Meta's Muse was released on September 8, 2026. But for consumer B2C AI products, a lot happend in those nearly 1,400 days. For most people it hasn’t been that exciting to say the least. Meta’s Muse might have a great team, and the product might still not catch on like ChatGPT did four years ago. But how did we get here?
Meta’s Superintelligence Lab (MSL) was formed just around 14 months ago: Meta’s core Superintelligence frontier team has about 44 all-star members. They are supported by thousands of other machine learning researchers. Muse is the result of a long journey of capital, compute and talent concentration (that will be hard for Google, OpenAI or anyone else to emulate). The entire MSL division is around 3,000 AI researchers with a total of around 6,500 AI related employees in the applied AI support unit. This is what it takes to build something like Muse.
Some Context: The early prototype of OpenClaw was first launched on November 24th, 2025. OpenAI later acqui-hired the founder on February 14th, 2026 but failed to capitalize on the opportunity. The entire time Muse was in development.
April, 2025: The major setback for Meta's Llama series occurred with the release and aftermath of Llama 4 in April 2025, culminating in early 2026. This signalled Meta AI would need a restart.
April 29th, 2025. Meta releases its standalone Meta AI App on April 29, 2025
Late June, 2025. The Great talent poaching: The bulk of Meta's aggressive poaching to build its Superintelligence Labs took place during the summer of 2025, specifically kicking off in June and July.
Late June, 2025. Mark Zuckerberg announced the creation of Meta Superintelligence Labs (MSL) on June 30, 2025, with the official public announcements and restructuring details following in early July 2025.
Late July, 2025. Meta's blog post titled "Personal Superintelligence" (also published under the title "Personal Superintelligence for Everyone") was written and published on July 30, 2025, read it here.
Mark Zuckerberg's supposed Public Manifesto, a 6,000-word essay titled “The Future is for Everyone” was released on August 10, 2026. Read it here.
Meta’s Muse agent was launched on September 8, 2026
Meta’s next-generation frontier AI model, codenamed "Watermelon," is expected to be released in late October 2026 and to be at the level of GPT-5.5. approx.
Re Prometheus datacenter near Columbus Ohio, the marquee milestone for Prometheus to hit its initial 1-gigawatt (GW) target is scheduled for late 2026 around December, 2026. Major data halls and initial on-site gas-turbine power infrastructure (such as the Socrates sites) are turning online throughout late 2026 to reach this capacity. (Hyperion is approx 5x the size of Prometheus).
Meta’s biggest datacenter Hyperion, will likely finish its first initial stage in 2027. The first data halls and infrastructure zones are scheduled to complete commissioning and go online, delivering an interim capacity target of roughly 1.5 GW in Richland Parish, Louisiana. Full capacity will take at least until 2030.
September 28th, 2026 - Meta has just announced Meta Enterprise Platform.
Is Muse a ChatGPT Killer?
I’m curious if these persistent AI agents will essentially cannibalize vanilla chatbots. How about for you in the time you spend with and on AI?
What can you really do with Muse? (Use Cases)
While a lot of normal consumers don’t spend as much time on ChatGPT or chatbots any longer, they must be asking what they can do with AI? At a time when AI is inflationary and increasing healthcare costs, personal intelligence can’t come soon enough. I’m not in the U.S. or Canada (currently) so I haven’t spent much time on Muse outside of web usage, but here’s what I like it for and the general consensus:
Let’s just try and name a few use cases that average Americans might use Muse for (remember these are non-AI users in general):
Helpful reminders in personal life
Parenting and Pet care. Early adoption and usage indicates people find real value in Muse for navigating caring roles. Focusing on family relationships.
E-Commerce, repeat shopping and price comparisons for holiday spending
Getting rid of unused or unneeded discretionary subscriptions (if you link your bank account)
Potential help with navigating and responding to Emails via Voice like a scene from the movie HER (2013) (if you link your Email)
Personalizing an AI assistant to give me niche info on my pet hobbies, interests and areas of news to improve my engagement with these things
Helping me monitor and optimize my family and social life better
Investing, personal finance and setting up an “equity analyst” assistant filled with smart reminders, recommendations and notifications. So, literally making me money!
Giving me new ways to optimize my side hustle, small business or entrepreneurship (and automate Ads if I so want)
Helping me deal with my health, fitness and personal finances better. Meta’s models are very decent in healthcare benchmarks.
Special tasks: Flight Delay Compensation: Files claims with airlines for delayed flights and tracks down issued credits automatically.
Cliche example: Restaurant Reservations: Phones restaurants directly or checks reservation platforms at opening time to secure hard-to-get tables.
Real life tasks example: Grocery List Generation: Pulls ingredients from saved online recipes, merges duplicate items, and categorizes them by supermarket aisle.
E-commerce comparative shopping: Product Price Tracking: Monitors online products and alerts or purchases them when prices drop to a designated threshold.
Example: Parenting Comms: Kid's School Email Extraction: Parses endless school newsletters to extract due dates, sign-up links, and permission slip deadlines.
Taking care of parents: senior care: Medication Schedule Management: Logs doses, sends timely reminders to loved ones, and follows up if a medication confirmation is missed.
Nich curation for professional lifelong learning: Deep Research Aggregation: Conducts comprehensive web research across primary and secondary sources to build structured research digests.
Credit Card Auditing: Unclaimed Funds & Charge Disputes: Flags unrecognized credit card charges and prepares dispute letters or claims for missing funds.
Holiday Gift Management: Gift Tracking and Ideas: Remembers past gifts given to friends and family and recommends personalized ideas well in advance of upcoming dates.
Personal or Household finances: Expense & Budget Monitoring: Categorizes incoming receipts and keeps a running log of weekly discretionary spending.
Muse could help people and users improve tracking of what matters most to them. Whether that’s relationships, information, events or micro-transactions. For now, Meta is paying the tab.
So you get the idea. But what’s next and who might be the competition?
Meta Connect Announcements related to Muse
You can watch the Meta Connect Keynote (four days ago) here. Or check out some of the key announcements here.
Real time Voice Mode: “Talking to an agent as capable as Muse is different from voice chat with an assistant. You can have a long, in-depth conversation, and Muse gets work done in the background while you’re still talking.”
Muse on AI Glasses: “You’ll soon be able to say your agent’s name, and Muse can act on what you’re looking at, so you won’t have to describe what’s in front of you. Ask about a product on a shelf, a flier on a wall, or a long list of school supplies, and Muse can take action.” You can name your Muse whatever you like and use it on the go.
Muse Charm: Zuckerberg's closing surprise at Meta Connect was a weird pocket-sized, Tamagotchi-like gadget with a small screen showing Muse's animated avatar, built purely for real-time voice interaction. Front camera, fingerprint sensor, action button. No price yet, targeting the holiday season.
Will Meta Muse Hurt Media and Creators?
Since Meta Muse optimizes how consumers can clean up digital subscriptions they no longer care about, at scale it could implode subscriptions and accelerate Subscription fatigue we saw in 2026 even more so in 2027.
This is a big deal given the decline of journalism in the U.S. over the last decade in particular and this could significant hurt independent media providers like those on Substack.
From the perspective of ordinary people undergoing an affordability crisis, subscriptions are one of the key discretionary spending that gets cut when budgets get tight. Since Muse was released it is claimed, consumers who gave the AI access to banking and credit card statements found it to be a great budgeting coach, leading them to cancel unnecessary subscriptions.
Americans have been increasing spending on subscriptions in recent years, as many as 20 per month to the tune of $157 on average. When it comes to subscription management, some consumers forget what they joined. They stop using services but keep paying. A $9 or $15 monthly charge can disappear into a credit card statement for months or even years. As these are canceled, it hurts Creators and media providers that have sought to replace the disruption of media. The Newsletter economy is particularly vulnerable to this.
Close to half (44%) of U.S. consumers increased their subscription spending in 2025, with average annual spending rising to $1,887, or about $157 a month, according to a report published by Mastercard and FT Strategies in April. But Muse could reverse this at scale in 2027. I’m expecting a huge Subscription implosion that started the day Muse was released.
Muse like Products with Dedicated Memory + VM
Not all of these have a cyber secure VM and dedicated long-term memory but these are the best I think:
Codex Bot (“o”, it’s late)
Google’s Early Prototype: Gemini Spark
An early porotype of Cognition (acquired) called Poke
An early Enterprise AI prototype called Carly.
An early prototype of a digital employee, CellCog
ChatGPT Work (not even a viable competitor to Muse at all)
An early prototype of dedicated VM space for agents Zo Computer
An early prototype for unified messaging AI: Wingman (the orange logo)
A decent prototype for Coding tasks for Enterprise teams (beware the forward deployed engineers): Cognition AI
In short, nothing much approaches the scope and B2C appeal of Muse currently on the market at of Autumn of 2026. Nor is anything likely to arrive anytime soon outside of China e.g. ByteDance.
But is it Growing?
The first few weeks show Meta’s Muse is making decent headway and good growth. For fairly high token usage per user one has to imagine it’s quite a compute bill that Meta will have to deal with though.
Musing on Muse via X (curation)
Meta Could Transform Enterprise AI taking marketshare from OpenAI + Microsoft there as well
Is Meta’s Pivot from Ads Finally here?
Or is this just another ploy to get humans to make even more Ads?
Can Muse Block the Ascent of Amazon in Advertising?
Muse has a lot of Endorsements from the Financial Elite
Agents in the Back
A New Era of AI Products
Is this AI for Normies and Non-AI folk?
Saving People Money in an Affordability Crisis?
Full disclosure: a human operative likely made the call.
Making Automated Complaints
Can Meta even Afford Scaling Muse?
Conclusion and some Thoughts
Meta’s Muse is clearly a next phase of consumer B2C experiences of AI for those who aren’t really into AI. With over three Billion users, Meta has a lot of data on how to optimize such a product. The app had 730,000 downloads over the roughly five days following its release on Sep. 8, according to the analytics firm Sensor Tower but retention rates or knowing what to do with it are not necessarily good. Meta could learn a thing or two from gaming tutorials. Given the cost for compute, Meta’s FCF (free cash flow) is going to crater in 2027 to a very suspect degree. And like Oracle, that bet is only going to get worse.
Betting on a product at any cost, that is directly related an accelerating demand for compute - is borderline insane to me. Only Anthropic has shown how to scale ARR so far in this entire Generative AI cycle. Meta’s Muse has a lot of potential, but also comes with a lot of risks. This isn’t 2022, how do you retain users in a sea of AI products? How do you make it valuable enough to actually pay for? The ratio of app downloads to actual users seems fairly low. That’s suspicious in and of itself.
Meta needs the AI devices part of this to work with the AI glasses, Muse Charm and whatever else it can bundle into the emerging ecosystem. While they will go directly head to head with the likes of Apple, Google and OpenAI AI devices. Meta needs to make full use of the first mover advantage, because it won’t last long. Meta essentially needs to keep hitting home-runs at consecutive at bats to make the math work. While Meta historically as a company tends to struggle with product execution outside of its expertise of digital advertising. I don’t like those odds for Muse.
At a time when American consumers have very low levels of AI sentiment, the timing for Muse hitting the public marketplace is also challenging and the TAM (total addressable market) might be smaller than Meta estimates today. The capex drag for inference costs and token spend is going to be immense and Meta will thus need to generate revenue from this ASAP. It’s not immediately clear how they will do so from an operating margins standpoint. It’s such a big gamble, it could ruin the company. There are some meta risks involved. Execution risks have never been this high. Putting Alexandr Wang in such a position may have been a great tactical mistake. Read Why We’re Building Muse. Meta’s cotton candy ideals and values seem misplaced. We know what the business stands for.
To be kind, Meta has around $88 billion in total cumulative operating losses of its so-called Metaverse, where its vision of VR truly failed. In total, Meta's Metaverse vision easily exceeds $95 billion to $100 billion. By 2028, I estimate Muse as a project by Meta will be 3x the spend. It needs to be successful in order for Meta to have even a solid future in such a changing landscape. Zuck has shown he’s not averse to burning cash for a big bet, but this might also be the wrong kind of bet in a market that will be too competitive.
While I’m optimistic on Muse in some ways, Meta’s track record for building models and products is not fronier or that of a leading lab. They are an advertising monopoly incumbent not well suited for building the next era of AI products. Kudos to them for trying, but this is likely too ambitious of a project even for them. AI is bigger than all of us. You don’t need to be Boris Cherny to understand the stakes or repeat the tired quotes of Dario’s talking points. Meta poached and paid its way to Muse, but that’s not how the magic happens. That’s not the formula for winning against China or the likes of Anthropic.
What do you think?
Addendum - Notes on AI
These are my musing on some of the macro trends that stood out to me.
Could Muse Cause a Bank Run?
Very weird coverage of Muse being a risk of a bank run by Torsten Slok. If users give an autonomous agent like Muse continuous permission to optimize their idle cash, Muse will automatically route liquidity to whichever FDIC-insured account or money market fund offers the highest return. Thus it could usher in a cascade. Muse era agents could be dangerous for Big Banks.
“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.” - Torsten Slok
Also see Morningstar on this.
ChatGPT has Harmed Culture
Generative AI is inflationary, increases healthcare costs, is making our utility bills rise and oh yeah, it’s harming culture, reading and literacy and funding of digital and live arts and entertainment.
Brookings Paper on Cost of AI Buildout
AI buildout is costing America. Read Financing the AI buildout. Big tech's spending on AI infrastructure went from $97 billion in 2020 to about $400 billion in 2025, and it's on track for $800 billion next year. Over $1.3 trillion in debt is already committed to it. $10.3 trillion between 2025 and 2032. That's 3.6% of the entire US economy, every year, for eight years. Not a rational scheme considering the U.S. National debt. That estimate is higher than the 2.2% of annual GDP absorbed by railroads in the late 1800s.
Stijn Van Nieuwerburgh, a finance and real estate professor at Columbia Business School, wrote in a paper prepared for a Brookings Institution conference last week. It begs a lot of questions. The paper is only 28 PDF pages.
“The projected buildout would be larger relative to the economy than the major U.S. canal, railroad, electrification, highway, and telecommunications investment booms,” writes the author, Stijn Van Nieuwerburgh of Columbia University.”
Like many of us have been writing on Substack for the past few months: “The paper warns that the risks embodied in the AI investment race are migrating from transparent on-balance sheet financing by major corporations to opaque off-balance sheet financing through joint ventures (i.e. circular funding my emphasis), private credit, securitization, special-purpose vehicles, lease commitments, loan guarantees, and other structures.”
You don’t need an economist to tell you the obvious in 2026: "Silicon Valley wants all of us to believe that this is a miracle technology... it has to generate trillions of dollars of revenues to be financeable. I'm just not sure how likely it is."
Employment in Tech Labor Market is Worse than Pandemic Slump
If AI was boosting productivity wouldn’t you expect to see a better tech jobs snapshot by now and a healthier labor market? Not much data that AI is leading to either productivity boosts or anything resembling ROIC.
$3 trillion in off-balance-sheet AI commitments
Moody’s and Michael Burry warnings are not enough. Muse like products will explode token inference spending. Nine tech giants carry $3 trillion in off-balance-sheet AI obligations, a figure triple their reported debt and nearly double a July estimate of $1.65 trillion. This is expanding at an alarming rate that is not sustainable. It’s 10x the 2023 amounts. But with HBM spiking and other bottlenecks, the cost of datacenters is increasing. The negative FCF we will see from BigTech incumbents might break the equity AI bubble in the coming months to years.
As an example, Meta hides $27 billion in Hyperion data center debt by making Blue Owl Capital the majority owner while Meta serves only as tenant. Debt and compute catches up with us all in the end. The flashy launch of Muse is a bleak reminder of how capital intensive and token wasting this entire AI race has become.
Meta’s off-balance-sheet debt alone is about $420 billion, nearly triple its reported debt. The price to being a winner for incumbents may indeed be a zero sum game for their stock and shareholders. The $1.65 trillion sitting in the footnotes will not make the headlines but it will continue to grow.
Really hard to watch: American prioritizes doom over creatives in the Generative AI “slop” era.
Increasingly diminishing marketshare with Google Gemini and others. In fact even in Search Google has taken back considerable marketshare in recent quarters.
Traffic to ChatGPT's website has been slowing down and much slower than most top websites for quite some time.
Also very much related Instinct just raised a $1 billion series c: https://techcrunch.com/2026/09/28/viral-ai-agent-instinct-raises-1b-series-c-at-a-10b-valuation/
Just after I published this meta announced Muse for small businesses:
Meta launches Muse for Small Business as Zuckerberg pushes beyond consumer AI market
https://www.cnbc.com/2026/09/29/meta-launches-muse-for-small-business-zuckerberg-pushes-enterprise-ai.html