What Would You Pay for Personalized AI?
What Would You Pay for Personalized AI?
The math for the $1T+ Personal AI market
Last week, we talked about the trillion-dollar opportunity in Personal AI. As a reminder, a Personal AI is “a chief of staff for your life…a scheduler, an organizer, an advocate, a buyer, a booker,” per Mustafa Suleyman, Inflection AI’s CEO. More than that, a Personal AI is trained on our data and knows us intimately. Whether we control how that data is used to serve us will depend on whether we pay for the service or not.
I ran a recent poll where ~70% of respondents say they’d pay some amount for a Personal AI. In this post, we quantify the potential of the paid Personal AI market given the survey data.
The Deload explores my curiosities and experiments across AI, finance, and philosophy. If you haven’t subscribed, join nearly 1,500 readers:
Disclaimer. The Deload is a collection of my personal thoughts and ideas. My views here do not constitute investment advice. Content on the site is for educational purposes. The site does not represent the views of Deepwater Asset Management. I may reference companies in which Deepwater has an investment. See Deepwater’s full disclosures here.
The People Want to Pay for Personal AI
When you use free software, you’re the product. When you pay for software, you’re the customer.
Credit where due, that’s not my original idea. I forget where I heard it first, but the idea sticks. On social networks, you are most obviously the product. The game of social is to keep your attention on the content, which means in a not so subtle way, we work for the network. Most of us would admit that we serve our social networks more than they serve us if we could pry ourselves away from Instagram or TikTok.
Is this the best path for AI?
Do we want our AIs to be attention harvesting tools, trapping us in an endless flow of stimulating dialogue?
Or, do we want our AIs to work for us?
If we could similarly pry ourselves away from ChatGPT, most of us would choose the latter. Our desire to be the customer of AI rather than the product demands a follow up question:
What are you willing to pay for an AI that works for you?
I ran a poll on Twitter and LinkedIn asking just that. With 245 responses, the answer was that almost 70% of people would pay something for a Personal AI.
The results favored a paid model more than expected. I assumed most people wouldn’t be willing to pay anything for a Personal AI since we’re spoiled by so many great free software products.
Maybe I shouldn’t have been that surprised. I pay $240/year for ChatGPT Plus just to get access to GPT 4 and plugins. Maybe it’s just that AI has captured our imagination in the current moment. Or maybe we instinctively understand the power of AI and the desire for a customer vs servant relationship.
If ~70% of people really would pay something for a Personal AI, then a paid approach might be the best way to monetize AI in a massive way.
The Math to $1 Trillion in Paid Personal AI
Assume my Personal AI polling data is correct and apply it to the US and EU markets for customers 18+. You get a $1.25 trillion Personal AI market.
I can already hear the objections. ElevenLabs prompt, in a whiny voice say:
“The survey sample size is small. No one will pay for an AI when they can get it for free. People are never going to spend $12,000 per year for a Personal AI. This is never going to happen.”
Rational skepticism is the best path to discover truth, but these objections are boring.
First, the survey is meant to be directional, not definitive. No survey is. Only events in the market are definitive. Since there’s no paid Personal AI available now, the survey will have to do. Some people will pay. Maybe more than we think.
Second, people already pay for AI en masse at costs similar to Internet or wireless bills. I already mentioned the $240/year ChatGPT Plus. Midjourney offers plans that cost between $90-1,200/year. Tesla FSD is $2,400/year.
Third, the average monthly car payment in the US is $725. That’s more a subscription than a loan because we often get a new car with a new payment when the old one runs its course. If AI is to be as transformative, powerful, and potentially dangerous as we all think, it will be more important to us than a vehicle, and we should be willing to pay more than a vehicle. There will be a class of people who pay more for AI than their cars to do everything they can to harness AI for their interests.
But let’s assume my survey is overly optimistic and ratchet down how many people will pay and how much. If only 50% of the market was willing to pay, and most of that was the lower tier, there’s still an almost $700 billion annual market opportunity for paid Personal AI.
Market sizing is more art than science, but even a skeptic should be able to see a path to a trillion dollar market for Personal AI.
Updating My Personal AI Mental Model
In the trillion-dollar Personal AI piece, I outlined three Personal AI business models:
The Neo - a fully self-managed Personal AI solution for highly technical users.
The Aristocrat - a Personal AI managed by a third party that users would pay.
The Commoner - a centralized Personal AI offered for free where the user is the product.
Given the survey data break down between low end ($300/year), mid end ($3,000/year), and high end ($12,000/year) customers, there seems to be a sort of aspirational Aristocrat and a true Aristocrat. A Mercedes vs a Rolls Royce.
The Mercedes Aristocrats ($300-3,000/year customers) will own their data, stay free of centralized control, and not be the product, but not get the same level of customization and capability as the Rolls Royce Aristocrats ($12,000/year customers).
Based on the survey and estimates above, the market for Mercedes Aristocrats and Rolls Royce Aristocrats might be roughly equal.
Are the Mercedes and Rolls Royce Personal Ads built by the same company?
You could make an argument either way. The company that figures out the best scaled solution to helping manage paid Personal AI might win all customers. That winner could offer the common tiers we see across SAAS models of good, better, best.
The other argument is that which AI service we use may become a status symbol. Rolls Royce Aristocrats might want the world to know they’re using the top service. That’s sort of the point of owning an ultra luxury product.
Either way, the winner or winners of the paid Personal AI space could earn hundreds of billions in revenue, equivalent to Google’s current scale (~$300 billion). That would create another trillion-dollar company, maybe something even bigger.
Disclaimer: My views here do not constitute investment advice. They are for educational purposes only. My firm, Deepwater Asset Management, may hold positions in securities I write about. See our full disclaimer.
AI Index Update
AI stocks performed well last week. Our AI Average Index (AIAI) was up 4.1% WTD vs the QQQ up 3.6% and S&P 500 up 2.2%. Our AI Hardware Index (AIHI) was up 3.1%.
Megacap tech continues to drive a significant portion of AI stock strength with GOOG, NVDA, and CRM all up over 7% last week. Both Alphabet and NVIDIA could be players in the Personal AI market. Alphabet could be a Personal AI provider, although it’s hard to imagine them moving away from ads. NVIDIA is the front runner to provide compute regardless of the winner.
You can track the indices on Thematic:
Later this week, I’ll introduce two new indices to answer another question:
Can ChatGPT beat the market?
We’ll find out…
Subscribe to follow along.
I love the product concept. We already see products like Calendy nibbling at pieces. Not sure any of these point products have scaled up. One products I've fantasized about that could take advantage of LLM would be contact manager, that would aggregate all your personal contact info such Apple's Contacts, Linkedin, FB, etc. But this would be private only to the customer, in sharp contrast to LI, FB, etc.
Precisely because the contact manager is private only to the owner, subscription makes sense. Not sure how the bot would use your LI info, i.e. what is yours and what belongs to MSFT.
This is super interesting and I think about it all the time. I guess I wonder though, what's the working definition of personal AI?