Paying for AI is still something only a small minority of people do, whatever the technology is called: artificial intelligence or super intelligence. That was the thread running through the 2 October 2026 episode of TechCrunch’s Equity podcast, titled “Call it AI, call it Super Intelligence, only 2% of consumers are buying it”. In one week, the White House renamed the technology by executive order, tech chiefs signed a safety pledge, and Meta and OpenAI gave their assistants cuddlier faces. The share of consumers paying for AI barely moved.
Hosts Kirsten Korosec, Anthony Ha and Sean O’Kane linked those stories to “the ugly economics of consumer AI”, the title of a TechCrunch column that week, and to a run of start-up deals in which nearly every buyer was a business. The point is not that consumers dislike AI. Many use it every day. The point is that use and payment are very different things.
This article looks at where the 2% figure comes from, whether a new name or a friendly mascot can change consumer behaviour, where the money in AI is actually coming from, and what that means for businesses deciding how to sell, or buy, AI products.
Table of contents
- What the Equity Episode Said About Paying for AI
- How Many People Are Paying for AI?
- Will a New Name Get More People Paying for AI?
- Will a Friendly Face Get People Paying for AI?
- Who Is Paying for AI Instead: Businesses
- What Oura’s Shelved IPO Says About Paying for AI
- Paying for AI: The Week in Numbers
- What the Paying for AI Gap Means for Businesses
- Signs That Paying for AI Is Changing
- Paying for AI FAQ
- References and Further Reading
What the Equity Episode Said About Paying for AI
Equity is TechCrunch’s weekly start-up podcast. The episode notes set out the week’s argument in a few lines, and each claim maps to a story TechCrunch reported between 28 September and 2 October.
The rename
President Trump signed an executive order on 29 September titled “Inaugurating the Era of Super Intelligence”. It tells federal agencies to use “Super Intelligence” and “SI” instead of “Artificial Intelligence” and “AI”. We covered the first announcement of the idea in our report on Trump’s super intelligence rename; the order made it official a week later.
The pledge
The same week, the White House gathered leading tech chief executives to sign a voluntary “Joint Commitment on Frontier Responsibilities”, which Trump called “morally binding”. TechCrunch noted that the signed copy misspelled the country as “Unites States”. Our explainer on how tech leaders will self-police AI safety walks through what the accord does and does not require.
The friendlier faces
Meta’s Muse assistant, with its plush mascot Jolly, has been a surprise hit, and OpenAI launched Dots, which TechCrunch described as a “bubbly agentic avatar”. Both aim to make AI feel approachable to ordinary users rather than to developers.
The IPO market
The episode also covered public markets: Oura pulling its IPO, Anthropic’s leaked S-1 and OpenAI returning to private funding. The hosts read these together as a sign of “how picky public markets have gotten”. Investors want proof of revenue, and for AI companies that proof has mostly come from businesses rather than consumers paying for AI.
The money
Against all that, the episode argued that “the biggest money in AI still appears to be coming from the enterprise”. The start-up deals discussed on the show, from maritime sensors to satellite insurance, all sell to businesses.
How Many People Are Paying for AI?
The “2%” in the headline is not a single survey result. It is a shorthand for a set of payment studies that all land in the same low single digits. We examined these in detail in The Ugly Economics of Consumer AI, so here is the short version.
| Source | Method | Finding |
|---|---|---|
| a16z State of Markets II (30 Sep 2026) | Chartbook drawing on PNC research | “As of April, barely ~2% of US households were paying for some AI service” |
| PNC, cited by TechCrunch | Payment data | 2.2% of consumers paying as of May, average $31 a month |
| Bank of America Institute (March 2026) | Household payments data | About 3% of households pay; paying households up 38% on the 2024 average |
| Menlo Ventures (2026) | Survey of US adults | 64% of adults use AI; 55% of users say they pay for at least one product |
Payment data says 2 to 3% are paying for AI
Bank data counts charges it can recognise as AI subscriptions. On that basis, PNC and Bank of America both find only a few per cent of customers paying for AI, and that is the figure the Equity episode used. Bank of America’s March report, “Not quite mAInstream”, found that the share of households paying $21 to $40 a month had grown 50% since 2024, so the trend is up, but from a very small base.
Surveys say much more
Menlo Ventures asked people directly about paying for AI and found that more than half of AI users pay for at least one AI product. The gap is large, and it is partly about method. A card study may miss subscriptions bought through app stores, paid by an employer, or covered by a family member. Menlo found that only 48% of payers cover the cost entirely themselves.
Either way, use runs far ahead of payment
Whichever number you trust, the conclusion holds. Hundreds of millions of people use AI assistants, and only a small fraction are paying for AI directly. Menlo found that 25% of Americans now use AI every day, while the payment studies still show low single digits.
Will a New Name Get More People Paying for AI?
If paying for AI were held back by branding, the rename would be the obvious fix. The evidence so far suggests the name is not the problem.
What the order actually does
The executive order says the executive branch “shall use the terms ‘Super Intelligence’ and ‘SI’ in place of ‘Artificial Intelligence’ and ‘AI'” in official correspondence, websites, reports and other non-statutory documents. It does not require changes to existing regulations, contracts or grants. Its definition of super intelligence is simply the existing statutory definition of artificial intelligence, so for now the two terms describe the same technology.
A deadline in November
The order gives the President’s science adviser 60 days to propose legislative language for a federal definition of “Super Intelligence” that reflects the capabilities described in the order. Sixty days from 29 September falls on 28 November 2026. Until Congress acts, the change is one of vocabulary, not law.
What the public prefers
A YouGov poll reported by Axios found that 53% of Americans prefer the term “artificial intelligence”, against just 9% who favour “super intelligence”. According to Axios, some executives expect to use the new term when dealing with the government but do not expect it to spread into everyday speech. California’s governor went further: Forbes reported on 30 September that Gavin Newsom had signed an order to keep calling it AI.
The .si gold rush
The rename did create one new market, and it had nothing to do with consumers paying for AI: domain names. Registry SI, which runs Slovenia’s .si domain, told the BBC that .si purchases rose 2,199% in September, although its spokesperson was cautious about crediting Trump alone. TechCrunch reported 11,000 new .si addresses on 30 September and almost 13,000 in a single 24-hour period.
Public reaction to the rename, % (YouGov via Axios; Hostinger via TechCrunch)
Speculators, not customers
The hosting company Hostinger told TechCrunch that only around 3% of the .si names registered through it were explicitly related to the technology. Most were “unclassified”, and some buyers were openly speculating. One public relations founder said he bought the .si version of his company name “to protect the brand in case this takes off like .ai”. A .si name costs about $12 a year at Hostinger, against $90 a year for .ai with a two-year commitment.
Will a Friendly Face Get People Paying for AI?
The second part of the week’s story was product design. If the rename was about how governments describe AI, the mascots are about how consumers feel about it.
Muse and Jolly
Meta’s Muse assistant, fronted by a plush-like mascot called Jolly, has grown quickly. It is also free to use, which is the key fact for anyone thinking about paying for AI. Meta can fund Muse from advertising and is already exploring a business version, which TechCrunch reported on 29 September.
Dots
OpenAI’s Dots, launched on 29 September, gives ChatGPT’s agent features a cartoon-like avatar. But Dots sits behind ChatGPT’s paid plans, which sets up a direct contest with a free rival. We looked at that contest in OpenAI’s new agent and whether it can compete with free.
Friendlier is not the same as paid
A friendly face lowers the barrier to trying an assistant, and that matters for paying for AI later. So do years of progress in natural language processing, which has made assistants far easier to talk to. Menlo’s survey found that payers are nearly twice as likely as non-payers to use AI daily, so habit does come before payment. But a mascot cannot make paying for AI feel necessary. Usefulness does that, and the strongest consumer products this year have either stayed free or found other ways to earn money.
Other ways to earn
TechCrunch’s column on consumer economics noted that the agent start-up Instinct plans to take a cut of purchases made through its agent, rather than charge only a subscription. That model sidesteps the question of whether people will pay for AI directly, because the AI earns its keep inside a transaction the customer already wanted to make.
Who Is Paying for AI Instead: Businesses
If consumers are not paying for AI in large numbers, someone else is paying for AI on their behalf, or instead of them. The Equity episode’s answer was businesses, and the week’s deals bear that out.
| Company | Deal reported | Who pays for the product |
|---|---|---|
| Quartermaster | $140M Series B (28 Sep) | Governments, shipping firms and insurers buying maritime data |
| Atomic | $12.5M Series A (29 Sep) | Companies such as DoorDash and HelloFresh automating inventory |
| Charter Space | $5M seed (30 Sep) | More than 50 space and defence companies buying insurance |
| OpenAI | Talks on a $30B round at about $1.4T (29 Sep) | Increasingly enterprises; bookings reportedly doubled since July |
Quartermaster
Quartermaster mounts camera and radio packages, which it calls SmartMast, on ships to gather real-time maritime data. Its $140 million Series B included about $100 million of equity, led by Insight Partners, and a $40 million debt facility from Stifel. Its customers are governments, shipping companies and insurers.
Atomic
Atomic, founded by former Tesla supply-chain staff, raised a $12.5 million Series A. Former Tesla president Jon McNeill, who sits on its board, told TechCrunch that DoorDash is “running, I think, 90% of its purchasing across hundreds of sites” on the system. That is AI making operational decisions inside a business, with the business paying.
Charter Space
Charter Space, a TechCrunch Startup Battlefield finalist, raised a $5 million seed round to grow its space insurance brokerage. It says it already serves more than 50 companies across the US space and defence industrial base.
The labs follow the same path
TechCrunch’s consumer economics column argued that frontier labs have shifted towards “the Anthropic model”, focusing on enterprise contracts. It said OpenAI’s enterprise bookings had reportedly doubled since July, and that even the Dots launch leaned on business uses. The same column pointed out that AI is unusually expensive to run compared with earlier waves such as social networking or cloud computing, which is why a mass consumer audience does not guarantee profits.
What Oura's Shelved IPO Says About Paying for AI
The episode also discussed the IPO market, and one consumer company stood out. Oura, the smart ring maker, postponed its planned listing of up to $2.2 billion on 29 September, citing “uncertainty in the IPO market”.
A consumer subscription that works
Oura is proof that consumers will pay a subscription when a product is useful enough. TechCrunch reported that it has 5.7 million paying members, up from 5 million at the end of June. Memberships carry an 89% gross margin and make up about 20% of sales.
Strong numbers, cautious markets
Oura expects revenue to rise 90% in its 2026 financial year from $907.9 million a year earlier. By our arithmetic, that would be about $1.73 billion. Even so, it chose not to list. Equity’s hosts tied this to a public market that has become picky, alongside Anthropic’s leaked S-1 and OpenAI’s move back to private funding. Our coverage of Anthropic’s prospectus sets out the losses investors are weighing.
The lesson for paying for AI
Oura’s members pay for health insights wrapped in hardware, not for “AI” as such. Whether a study counts that as paying for AI depends on how it classifies the charge. That may be the most useful clue about consumer behaviour. People pay for outcomes they care about. The label on the technology underneath matters much less.
Paying for AI: The Week in Numbers
The week’s announcements make more sense side by side. The table below lists each item, the number attached to it and what it says about paying for AI.
| Date | Event | Number |
|---|---|---|
| 29 Sep | Executive order renames AI as “Super Intelligence” | 60-day deadline for a legal definition |
| 29 Sep | Oura postpones IPO | 5.7M paying members |
| 29 Sep | OpenAI launches Dots | Paid ChatGPT plans only |
| 30 Sep | a16z State of Markets II | About 2% of US households paying |
| 30 Sep | .si registrations at Registry SI | 11,000 in one day |
| By 1 Oct | YouGov naming poll reported by Axios | 53% prefer “AI”, 9% “super intelligence” |
| 2 Oct | Equity episode airs | “Only 2% of consumers are buying it” |
Reading the table
The pattern is consistent. The policy and branding events produced headlines and domain sales. The payment data produced a small number that changes slowly. The products that grew fastest were free or business-facing.
What would actually move the number
Three things have historically turned free users into payers: a product that becomes part of a daily routine, a clear outcome worth money, and a price that fits a household budget. Menlo found the typical payer spends $20 to $49 a month. That is the band any consumer product asking people to start paying for AI has to fit into.
What the Paying for AI Gap Means for Businesses
For a business deciding how to sell or buy AI, the low level of consumers paying for AI is a practical issue, not an abstract one.
If you sell to consumers
Do not build a plan that depends on large numbers of people paying for AI as a standalone subscription. Bundle AI into a product people already pay for, take a share of a transaction, or use AI to cut your own costs. The consumer appetite for a separate “AI” line on the bank statement is still small.
If you sell to businesses
The money is flowing your way, but so is the competition. Every week brings another well-funded start-up selling to the same operations, finance and supply-chain teams. Clear outcomes, such as Atomic’s purchasing decisions or Quartermaster’s maritime data, are what win budgets.
If you are buying AI for your team
Enterprise is where AI companies make their money, which gives buyers leverage. Ask for outcome-based pilots, check data handling carefully, and compare prices across vendors. The same vendors chasing consumers with mascots are chasing your contract with discounts.
If you are a UK business
Nothing in the US order applies in the UK, and British public bodies still say “AI”. But UK firms that sell into American government supply chains, or that use US vendors’ marketing material, will start to see the new term. The commercial point is the same on both sides of the Atlantic: consumers paying for AI directly remain rare, so most UK AI revenue will come from business customers.
Watch the language in contracts
If you supply the US federal government, expect “Super Intelligence” and “SI” in new documents. The order says existing contracts do not need to change, but suppliers should make sure their own terms map cleanly to whichever word a customer uses, especially once a statutory definition is proposed.
Signs That Paying for AI Is Changing
The rename and the pledge were the headlines. These are the signals that will show whether paying for AI is changing for ordinary households.
The next payment data
Bank of America and PNC publish consumer spending research regularly. A clear rise from the 2% to 3% range would mean something is changing. A flat line would confirm the Equity episode’s point.
The legal definition
The proposed legislative language is due by 28 November 2026. If it gives “Super Intelligence” a narrower or broader meaning than “artificial intelligence”, the rename stops being cosmetic.
How Muse and Dots make money
Watch whether Meta introduces paid tiers for Muse, and whether OpenAI keeps Dots inside its paid plans or opens a free version. Both choices will say a lot about how the largest companies read consumer willingness to pay.
Paying for AI FAQ
Where does the “only 2% of consumers” figure come from?
It reflects payment studies. a16z’s State of Markets II says about 2% of US households were paying for an AI service as of April, and PNC data cited by TechCrunch put the share at 2.2% in May.
Did the executive order change the law?
No. It changes the words federal agencies use. Its definition of “Super Intelligence” is the existing legal definition of artificial intelligence, and a proposal for a new definition is due within 60 days.
Do Americans like the name “super intelligence”?
Not much. A YouGov poll reported by Axios found 53% prefer “artificial intelligence” and 9% prefer “super intelligence”.
Why are companies focusing on enterprise customers?
Because businesses pay more reliably, and AI is expensive to run. TechCrunch reported that OpenAI’s enterprise bookings had doubled since July.
Are consumers paying for AI at all?
Yes, but not many through direct subscriptions. Surveys such as Menlo’s find higher figures than payment data, partly because employers, families and app stores pay on many people’s behalf.
What was the Equity episode about?
The 2 October 2026 episode covered the rename, the safety pledge, friendlier AI products from Meta and OpenAI, consumer AI economics, the IPO market and start-up deals.
References and Further Reading
Call it AI, call it Super Intelligence, only 2% of consumers are buying it (TechCrunch Equity)
Inaugurating the Era of Super Intelligence (The White House)
The ugly economics of consumer AI (TechCrunch)
Not quite mAInstream: A consumer AI profile (Bank of America Institute)
2026: The State of Consumer AI (Menlo Ventures)
Poll: Americans Prefer AI to Super Intelligence (Newsmax via Yahoo News)
Slovenia’s .si domain sees a surge in registrations (TechCrunch)
Pledge signed by President Trump and top AI leaders misspells the United States (TechCrunch)
Oura shelves its $2.2B IPO (TechCrunch)
Quartermaster raises another $140M (TechCrunch)
Ex-Tesla team raises $12.5M to put supply chains on autopilot (TechCrunch)
Charter Space raises $5M to bring insurance to the stars (TechCrunch)
More AI coverage: explore Progressive Robot's AI Models, Tools & Releases hub — hands-on reviews, setup guides and benchmarks in one place.
