AI startup costs just crashed 90%

I recently saw internal numbers from a Sequoia-backed AI startup. Eighteen months ago, they were paying $30/user/month in API costs.

Today? $3/user/month.

What used to cost $30 per user now costs just $3.

That’s a 90% drop—and it's still trending down.

Every 2 generations of AI models = 100x cost reduction.
Let that sink in: something that cost $100 in 2022 now costs $1.

So what does this mean for entrepreneurs?

• AI-first startups can bootstrap longer.

• Barriers to entry are disappearing.

• Unit economics are getting ridiculously good.

• Competition is about to explode.

We’re seeing a perfect storm:
Smarter models ✅
Cheaper compute ✅
Faster inference ✅
Energy costs falling ✅

The impact?
We’re entering a new era where entire categories of work that were economically impossible 18 months ago are now totally viable.

Example:
We built an agentic workflow that coaches our team daily on their highest-leverage task.
This would've been cost-prohibitive before.
Now? It runs at scale and took just hours to set up.

And this isn’t just theory—OpenAI jumped from $3.5B to $10B in projected revenue while cutting costs at the same time. They're eyeing $125B by 2029.

The question isn't if AI will reshape your industry.
It’s whether you’ll be the disruptor or the disrupted.

So here’s what you should be thinking about right now:

• How will you integrate AI before your competition does?

• How does cheaper AI unlock new business models for you?

• What becomes possible now that wasn’t before?

The AI advantage window is closing fast.

To faster, smarter execution,
Eric Siu