As I think about my next startup, I’ve been speaking with dozens of AI founders and VCs. What I’m seeing is unlike any other market cycle before. The pace is brutal.

A startup announces a round and ships a product. Within weeks, five clones appear. One is cheaper. Another is open source. A third is already running ads. Suddenly, your launch becomes their lunch.

In this environment, speed doesn’t look like an advantage anymore. Without real lock-in, it’s starting to feel like a liability.

What VCs Are Seeing

From the investor side, there are clear patterns emerging:

  • Most GenAI startups are wrappers on public APIs with a slick UI
  • Many founders claim to be building “infrastructure,” but in reality, the product looks more like prompt templates
  • Pricing trends toward zero unless there is a clear ROI story
  • Frontier labs are moving into the application layer, directly threatening portfolio companies

Several VCs have admitted they are considering a pause, given how quickly things are shifting.

What Founders Are Facing

For founders, the challenges are equally stark:

  • Proofs of concept are easy to land, but renewals remain a struggle
  • Enterprises are curious, but security reviews and on-prem demands often kill momentum
  • Competitors can offer extreme levels of customization because the underlying build is simple
  • Many teams mistake early interest for genuine product-market fit

Even experienced founders are feeling the strain. Burnout is visible, and strong operators admit that every new announcement from Sam Altman or another major player reshapes the ground under their feet.

Where Real Opportunities Exist

Despite the noise, certain areas show promise:

  • Products directly tied to revenue or cost savings, not vanity outputs
  • Workflows that deliver end-to-end value rather than surface-level automation
  • Systems that learn from customer behavior, not just respond to prompts
  • Tools that integrate deeply into messy, real-world environments such as CRMs, ERPs, emails, and internal databases

The dream remains pricing on performance, though it carries risks. Covering API costs or running your own infrastructure can be expensive, even with AWS and GCP credits to soften the blow.

Most GenAI products are still flashy demonstrations — the equivalent of saying “Look what it can do.” The more durable opportunities are grounded in a different question: “Did it work?”

Rethinking Visibility

I used to believe every founder should build in public and announce frequently.

Lately, I’m questioning that approach. There are real merits to staying in stealth.

After all, do I really want to invite more competition before I even begin?

Conclusion

This market is moving at dizzying speed. For founders and investors alike, the landscape is full of excitement but also exhaustion. Amid the noise, the lesson is simple: the future belongs to products with depth, defensibility, and direct impact, not just those that look impressive in a demo.

Here are the key takeaways:

  • Speed without defensibility is no longer an advantage in GenAI
  • Many startups are wrappers with little moat, and pricing races to zero
  • POCs come easily, but renewals and enterprise adoption remain tough
  • Real opportunities lie in ROI-driven products, end-to-end workflows, and deep integrations
  • Building quietly may be smarter than building loudly in today’s climate

I am still exploring ideas and 100% committed to starting another company. For now, knowing what not to build is its own form of clarity.