metered ai 1

AI Is Now Sold by the Meter. Are You Using It?

AI Is Now Sold by the Meter. Are You Using It?

By Raymond Miyawa, a developer and AI automation engineer at Digital4Africa,

The biggest IPO in history, and a question for you

On 19 November 2026, Anthropic, the company behind the AI assistant Claude, started trading on Nasdaq. Its shares were priced at $83, opened at $111.50, and the offering raised about $81 billion. By that measure, it was the biggest stock market debut ever, ahead of SpaceX earlier in the year.

Here’s what struck me. The business underneath that number is surprisingly simple. Anthropic doesn’t mainly sell software licenses or advertising. It sells intelligence by the meter, and businesses pay for every bit of it they use.

Meanwhile, most people I meet, including business owners, managers and smart professionals, still use AI the way they used Google in 2005. They type a question, read the answer, and close the tab.

So the world’s markets just valued AI-by-the-meter at around a trillion and a half dollars. The question is simple: are you plugged in, or are you still standing outside the meter?

What is metered AI?

Metered AI means you pay for AI the way you pay for electricity, water, or mobile data: by how much you use.

AI models read and write text in small pieces called tokens. A token is roughly three-quarters of a word. Every time a business sends a request to an AI model, the meter counts two things:

  • Tokens in: the question, plus any document, email, or data attached to it.
  • Tokens out: the answer the AI writes back.

The bill is the number of tokens multiplied by the price of the model used. Smarter models cost more per token, the same way premium fuel costs more per litre.

In practice, that means a short email summary might cost a fraction of a cent, while analyzing a 300-page contract costs more. A business that builds a customer-service bot on top of an AI model gets a bill each month based on the conversations it actually handled.

If you’ve ever bought prepaid electricity tokens or a data bundle, you already understand the model. This is the same idea, applied to intelligence. Industry analysts have started calling it “metered intelligence”.

The consumer apps you may use, such as the paid chat plans, hide the meter behind a flat monthly fee. But underneath, at the level where businesses build products, the meter is always running.

Why the meter matters

Anthropic’s IPO filing gives us a rare look inside a metered AI business. According to analysts’ reading of the filing, the company describes its primary revenue model as “pay-for-intelligence”: customers pay for the tokens they use.

The growth is hard to ignore:

  • Revenue: $33.2 billion in the first nine months of 2026, up roughly twelve times on the same period a year earlier.
  • Customers: more than 14,000 business customers, about 250 of which spend $10 million or more a year.
  • Productivity: revenue per employee rose from about $1.4 million to over $6.6 million in a year.

That tells you something important. Companies aren’t just experimenting with AI. They’re paying for it every day, at scale, because it’s doing real work.

But the meter cuts both ways, and the filing is honest about that too:

  • Revenue can fall as fast as it rises. Customers can stop at any time. Total customer spending reportedly dropped about 20% in the two weeks around New Year.
  • Costs are committed early. Computing power has to be secured 12 to 18 months ahead, and the company reports roughly $900 billion in long-term spending commitments.
  • It’s not yet profitable on paper. The company reported a nine-month net loss of about $104 billion, though much of that came from non-cash accounting charges, and it generated positive operating cash flow.

For you and me, the lesson isn’t about stock prices. It’s this: intelligence has become a utility. It’s sold like electricity, priced per unit, and available to anyone with a card and an internet connection. A five-person business in Nairobi, Lagos, or Manchester can now rent the same intelligence as a Fortune 500 company.

The other side of the meter: businesses built with AI

If some companies sell intelligence by the meter, someone has to be buying it. And the most interesting buyers aren’t always the biggest.

A new kind of business is appearing: small teams, sometimes a single person, running operations that once needed a whole office. Inc. magazine ran a feature titled “Claude Is My Co-Founder”. Solo founders publicly describe running several products alone, with AI handling research, writing, coding, design, and customer support. Anthropic’s own CEO, Dario Amodei, has predicted that the first billion-dollar company run by a single person is coming.
Is it me? We’ll figure it out eventually.😌

You don’t have to believe in billion-dollar solo companies to see the shift. Look at what’s already practical today:

  • A consultant who used to spend a day on a proposal drafts it in an hour and spends the rest refining it.
  • A small shop answers customer questions on WhatsApp around the clock with an AI assistant trained on its own price list.
  • A developer builds and ships a working app in a weekend that once took a team a month.
  • An organization turns its policy manuals into an assistant staff can actually ask questions of.

None of these people own an AI model. They rent intelligence by the meter and build on top of it. That’s the real opportunity: you don’t need to build AI to profit from it. You need to know how to use it.

There’s a warning in here too. Some startups have found that the AI they built on top of became good enough to replace their product overnight. Building on the meter rewards people who solve a specific problem well, not people who simply wrap a chatbot in a nice interface.

And still, most people don’t know how to use AI

Here’s the uncomfortable part. The technology is available to almost everyone, yet the gap between people who use AI well and people who don’t is growing fast.

In my work training teams and building automation, I keep seeing the same patterns:

  1. Using AI as a search engine. Asking one question, taking the first answer, and leaving. That’s like hiring a brilliant assistant and only asking them the time.
  2. Giving it no context. “Write a proposal” gets a generic proposal. Your client, your prices, your tone, and your goal get a useful one.
  3. Never giving it real work. Most people chat with AI but never hand it a spreadsheet, a contract, a set of meeting notes, or a messy process to fix.
  4. Not checking its work. AI can be confidently wrong. Using it well means reviewing what it produces, especially facts and numbers.
  5. Waiting for someone else to figure it out. Many organizations are waiting for a perfect AI strategy while their competitors simply start.

The difference between someone who “has tried ChatGPT” and someone who actually works with AI is rarely technical skill. It’s habit. The people getting the most value treat AI like a capable new colleague: they explain the task, share the materials, review the result, and give feedback.

The meter doesn’t care who you are. It only measures who is using it.

How to start plugging in

You don’t need a big budget or a technical team. You need a starting point.

If you’re an individual or professional:

  1. Pick one repetitive task you do every week: reports, emails, summaries, research. Do it with AI for a month.
  2. Give it real context. Paste in the actual document, explain who it’s for and what good looks like.
  3. Work in rounds. Ask for a draft, then push back: shorter, sharper, more formal, add an example.
  4. Always review. Treat the output as a strong first draft, not the final word.

If you run a business or organization:

  1. Start with a problem, not a tool. Where do your people waste the most time? Customer questions, quotations, reports, data entry?
  2. Run a small pilot. One team, one process, a few weeks. Measure time saved and quality.
  3. Write down what works. Turn good prompts and workflows into shared team procedures.
  4. Set simple rules. Decide what data can and can’t go into AI tools, and who checks the output.
  5. Then automate. Once a process works by hand, connect it properly: an assistant on WhatsApp, AI inside your ERP or CRM, or automated reports.

The businesses that will win in the metered era aren’t the ones with the biggest AI budgets. They’re the ones whose people learned to use it early and kept learning.

The meter is already running

A trillion-dollar listing is a signal, not the story. The story is that intelligence is now something you can rent by the unit, as easily as electricity or data. Some people are already using it to run leaner businesses, serve more customers, and build products alone. Others are still deciding whether AI is worth trying.

The gap between those two groups won’t be decided by access. Almost everyone has access. It will be decided by who learns to use it.

If you or your team are ready to move from asking AI questions to having it do real work, from training your staff to automating your processes, talk to us at Digital4Africa. The meter is already running. Make sure it’s running for you.

 

About the author: Ray Miyawa is a developer and AI automation engineer at Digital4Africa, where he builds ERP systems, custom software, AI agents, and websites, and trains teams to put AI to practical use.