The AI You Couldn't Afford Last Year Just Got Cheap

There's a version of this year's AI story that nobody put on a billboard, and it's the one that actually affects your bottom line.

The price of using AI has been falling off a cliff. Not slowly. Not "a little cheaper each quarter." Over the past couple of years, the cost to run a top-tier AI model has dropped by an amount that would make any CFO do a double take. The tools that felt like a splurge in 2024, the ones you looked at and thought "neat, but not for a business my size," now cost a fraction of what they did.

So why did nobody tell you? Because "prices went down" isn't a headline that gets clicks. But if you run a business, it might be the most useful AI news of the year. Let's translate it into plain English.

What actually happened

Three things collided this year, and together they crushed the price of AI.

First, the model makers went to war with each other. When one company releases a model that's nearly as good as the most expensive option but costs a lot less, everyone else has to respond. This summer we saw exactly that. New releases from the major players landed at roughly half the price of the versions they replaced, and in some cases matched the quality of last year's premium tier for a small share of the cost. Google, OpenAI, Anthropic, and the newer challengers are all undercutting each other, and you're the one who benefits.

Second, the technology under the hood got more efficient. The chips got better. The software that runs the models got smarter about doing more with less. That means the same answer that used to take a lot of expensive computing power now takes much less. Cheaper to produce means cheaper to sell.

Third, smaller models caught up. You don't need a Ferrari to drive to the grocery store, and you don't need the biggest, most expensive AI to write an email or summarize a report. The industry figured out how to make small, cheap models that are genuinely good at the everyday stuff. For most business tasks, "good enough and cheap" beats "brilliant and pricey" every single time.

Put those together and you get what people are calling the AI price war. The practical result: the meter is running a lot slower than it used to.

Why a normal business owner should care

Here's the part that matters. For the last two years, the honest reason a lot of small businesses stayed on the sidelines was money. The good tools felt expensive, the pricing was confusing, and it was hard to justify a monthly spend on something you weren't sure would pay off.

That excuse is quietly expiring. When the cost of using AI drops this much, a few things change for you specifically.

• The math flips. A task that wasn't worth automating at last year's prices might be very worth it now. Drafting proposals, cleaning up your customer list, turning meeting notes into follow-ups, answering the same five customer questions for the hundredth time. When each of those costs pennies instead of dollars, the "is this worth it?" calculation lands on yes a lot more often.
• You can stop rationing. People used to treat AI like an expensive consultant they only called for the big stuff. Now it's more like electricity. You can use it for the small annoying tasks too, the ones that don't feel important individually but eat your whole afternoon in aggregate.
• The tools you already pay for got better without a bigger bill. A lot of software you use, your email, your CRM, your accounting app, has AI baked in now. As the underlying costs drop, those features get more capable and more generous. You're getting upgrades you didn't ask for and didn't pay extra for.

The catch nobody mentions

Now for the honest part, because a story with no downside is usually a sales pitch.

In fact, plenty of companies are reporting the opposite. Their AI spending went up this year, not down.

How does that happen when prices are falling? Same reason your phone bill didn't shrink when texting became free. When something gets cheap, people use way more of it. A business that ran a handful of AI tasks last year is running dozens this year, because now they can. The price per task fell, but the number of tasks exploded, so the total went up.

This is worth watching. Cheap AI is only a win if you're pointing it at things that actually matter. If you spin up automations for busywork that didn't need doing in the first place, you've just found a very efficient way to waste money. We've written before about automating the wrong thing and about buying AI that nobody ends up using. Cheaper prices make both traps easier to fall into, not harder. Low cost lowers your guard.

There's also a quality question. The rock-bottom-cheap models are great for routine work and genuinely bad for anything that needs deep reasoning, careful judgment, or a high tolerance for being wrong. Part of using AI well now is knowing when to reach for the cheap, fast option and when to pay for the smart one. It's the difference between a text and a phone call. Both have their place. Sending a legal contract by text would be a mistake.

What this changes for you tomorrow

You don't need to understand model architecture or pricing tiers to act on this. You need to shift how you think about one question.

The old question was "Can I afford to try AI on this?" For most small tasks, the answer is now clearly yes, and it's going to keep getting cheaper. So the better question is:

Make a short list. The repetitive stuff. The copy-paste stuff. The "I do this every Monday and it takes an hour" stuff. Start with whatever interrupts you most, the same way you would if you were writing the process down. That's your opportunity, and the price tag that used to protect those tasks from automation just got a lot smaller.

A few concrete places to look:

• Customer replies. The same questions come in over and over. Cheap AI is very good at drafting first responses you can review and send.
• Turning messy notes into something usable. Meeting notes into action items. A voice memo into a proposal. A pile of feedback into a summary you can actually read.
• First drafts of anything. Emails, listings, job posts, social captions. Not the final word, the starting point that saves you the blank page.
• Cleaning and organizing data. Deduping a contact list, standardizing a messy spreadsheet, pulling the key numbers out of a report.

Notice what's not on that list: the big judgment calls, the relationships, the strategy, the moments where being wrong is expensive. Cheap AI doesn't touch those. It clears the underbrush so you have more time for the work that only you can do.

One warning as you start. Cheap makes it easy to try things and just as easy to stop there. The gap between an impressive test and a changed workflow is where most of these efforts quietly stall, and a lower price tag does nothing to close it.

Is this hype or does it matter?

It matters, and here's the tell. Hype is usually about a flashy new feature that demos well and disappears. This is different. This is a structural shift in what things cost, and cost is the thing that decides whether a small business can actually use a technology or just read about it.

Electricity didn't change the world when it was a novelty for rich households. It changed the world when it got cheap enough that a corner shop could afford a light bulb and a fan. AI is somewhere on that curve right now, sliding from "expensive novelty" toward "cheap utility." You don't have to be early to benefit from that. You just have to notice it's happening and stop pricing AI in your head at last year's rates.

The barrier to using this stuff was never really the technology. It was the cost and the confusion. One of those just fell away. The other is where a little guidance goes a long way.

If you're trying to figure out where AI actually fits into your business or daily workflow, and where it doesn't, that's exactly what we help people do at Humanity AI. No jargon, no overpromising. Just the practical version.