The New York Times just ran a piece titled "People Loved the Dot-Com Boom. The A.I. Boom, Not So Much." Hacker News lit up. Twitter piled on. The consensus? AI is overhyped, the bubble will burst, and everyone should be worried.
Here's the thing: the "AI bubble" everyone keeps talking about has nothing to do with you.
Not if you're an indie developer. Not if you're building a focused AI tool. Not if you're charging $29/mo and serving a real niche.
The bubble conversation is about Big Tech burning cash at a pace that makes Wall Street physically uncomfortable. And that discomfort is your opportunity.
The $650 Billion Bonfire
Let's look at the actual numbers. In 2026, the five biggest hyperscalers (Amazon, Google, Microsoft, Meta, Oracle) are on track to spend over $650 billion in capital expenditures. Roughly $450 billion of that is directly tied to AI infrastructure: GPUs, data centers, cooling systems, and enough power to run a small country.
Meta alone told investors it would spend $115 to $135 billion this year. Microsoft is running at roughly $80 billion annually. These are staggering, almost absurd figures.
Wall Street is nervous. And honestly? They should be. When Sam Altman himself admits there's "more resistance to the diffusion, the absorption of A.I." than he expected, and 80% of firms in a National Bureau of Economic Research survey report AI is having no impact on their productivity, the vibes are not great for companies betting hundreds of billions on adoption curves that haven't materialized.
But here's what gets lost in the panic: this conversation is about mega-cap infrastructure spending, not about whether AI tools are useful.
You Are Not OpenAI
There's a category error happening in the public discourse. People conflate "the AI bubble" with "AI doesn't work." Those are completely different statements.
AI models are genuinely useful. People use ChatGPT, Cursor, Midjourney, and dozens of niche AI tools every single day. The question isn't whether AI creates value. The question is whether spending $650 billion on data centers will generate returns fast enough to satisfy Wall Street.
That question doesn't apply to you. You didn't raise $13 billion. You don't need 50 million daily active users to justify your existence. You need 500 paying customers at $29/mo, and you have a business doing $174K ARR.
The economics of indie AI tools are fundamentally different from the economics of OpenAI.
The Dot-Com Playbook (You Already Know How This Ends)
We've seen this movie before. The dot-com bubble inflated through the late '90s and burst spectacularly in 2000. Pets.com, Webvan, and a thousand other companies evaporated overnight. The NASDAQ lost 78% of its value.
And then something interesting happened.
Amazon survived. Google launched. And thousands of small web businesses, the ones nobody wrote breathless NYT articles about, thrived in the aftermath. The infrastructure the bubble built (fiber optic cables, server farms, a generation of web developers) became cheap, abundant, and available to everyone.
The people who built boring, profitable web businesses in 2002 and 2003, while VCs were licking their wounds, ended up doing extremely well. Basecamp (launched 2004). Mailchimp (grew through the bust). Plenty of Fish (built by one guy, sold for $575 million).
The same pattern is forming right now with AI. The billions being poured into GPUs, model training, and infrastructure will make AI APIs cheaper, faster, and more accessible. When the bubble pops (or deflates, or whatever it does), the cost of building AI products will drop even further. The developers who are building right now will have a massive head start.
Indie AI Builders Are Already Profitable
While the NYT publishes doom pieces, solo founders and tiny teams are printing money with AI tools. These aren't hypotheticals. These are real numbers:
BoredHumans — Built by solo developer Nick Dobos. Over 100 AI tools on a single domain, driven by SEO and organic traffic. Revenue: roughly $733K/month (~$8.8M ARR). No VC funding. No massive team. Just one developer and a lot of AI-powered utility pages.
SiteGPT — Built by Bhanu Teja Paccha, a 24-year-old solo founder. Custom AI chatbots trained on business data. Revenue: ~$95K MRR (~$1.14M ARR). Started as a side project. Grew through clear B2B positioning and product-led growth.
Dozens more on Indie Hackers and Twitter — AI writing assistants, code review bots, image generation tools for specific verticals, AI-powered SEO tools. Many doing $10K to $100K per month with teams of one or two people.
The pattern is consistent: find a specific problem, wrap an AI model around the solution, charge money for it. No billion-dollar infrastructure play required.
Fear Creates Opportunity
Here's the contrarian take that most people won't internalize: the best time to build is when everyone else is scared.
When the narrative shifts to "AI is overhyped," several things happen in your favor:
- Competition thins out. The tourists and hype-chasers move on to the next thing. The people still building are serious.
- Talent becomes cheaper. When big AI labs slow their hiring (or start layoffs), skilled developers become available.
- Infrastructure gets cheaper. The overcapacity from billions in spending means API costs and compute prices drop.
- Customers still have problems. The business problems that AI tools solve don't disappear because the NASDAQ had a bad month.
Warren Buffett's famous line about being greedy when others are fearful applies to products, not just stocks. If you can build something useful while your competitors are frozen by fear, uncertainty, and doubt, you win.
You Don't Need VC Money. You Need a Checkout Page.
The VC bubble and the AI tool market are two different things. VCs are worried about returns on $100M+ investments in foundation model companies. That's their problem.
Your problem is simpler: can you build something people will pay for, and can you collect the payment?
The first part is on you. The second part is easier than it's ever been.
If you're building an AI tool and you're ready to sell it, you don't need to spend weeks integrating payment systems, handling tax compliance across 40 countries, or figuring out subscription management. You need a Merchant of Record that handles all of that so you can focus on your product.
That's what CREEM does. Global payments, tax compliance, subscription management. You build the AI tool, plug in CREEM, and start collecting revenue. No fundraising required.
The Bottom Line
The "AI bubble" is a story about $650 billion in infrastructure spending and whether Big Tech will see returns. It's a valid concern for investors in NVIDIA and Meta.
It is completely irrelevant to you, the developer building an AI-powered tool for a specific audience at a reasonable price point.
The infrastructure is being built (with someone else's money). The models are getting cheaper. The customers are there. And the hype-chasers are about to get scared off.
Build now. Sell now. Let Wall Street worry about the bubble.
