Hey — It’s Nico.

Welcome to another Failory edition. This issue takes 5 minutes to read.

If you only have one, here are the 5 most important things:

• Freight Farms, a startup building farms in containers, has shut down — learn why below.

• State of the product job market in 2025.

• Improvements in ‘reasoning’ AI models may slow down soon, analysis finds.

• Elizabeth Holmes’s partner raises millions for blood-testing startup.

• YC published their Summer 2025 Request for Startups — see which startups they want to invest in below.

Let’s get into it.

This Week In Startups

🔗 Resources

State of the product job market in 2025.

How founders should think about runway.

You need a better testing strategy.

The hidden benefits of being an open-source startup.

📰 News

OpenAI brings its GPT-4.1 models to ChatGPT.

Y Combinator says Google is a ‘monopolist’ that has ‘stunted’ the startup ecosystem.

Improvements in ‘reasoning’ AI models may slow down soon, analysis finds.

OpenAI agrees to buy Windsurf for about $3 billion.

💸 Fundraising

Samaya AI, a startup building agents for Financial Services, raised $43.5M.

Nuclear battery startup Zeno Power raises $50 million to expand in space and at sea.

Flam, an AI infrastructure startup, has raised US$14 million.

Elizabeth Holmes’s partner raises millions for blood-testing startup.

Fail(St)ory

Container Farms

A few days ago, Freight Farms, a Boston agtech startup that once turned shipping containers into lettuce factories, shut down and filed for bankruptcy.

They had raised over $56 million, installed hundreds of farms around the world, and were once valued at nearly $150 million. Now? $26,000 left in the bank and two lawsuits hanging over their heads.

What Was Freight Farms:

Back in 2013, two founders had a clever idea: take old shipping containers, retrofit them with hydroponic tech, and let people grow food anywhere, even in a parking lot in Arizona.

They called their main product the Greenery. It was basically a self-contained farm: climate control, LED lights, sensors, water systems, all controlled remotely with their app, farmhand. One unit cost over $160,000.

The company envisioned their farms being used in diverse settings: urban areas with limited space, educational institutions for hands-on learning, hospitals aiming to provide fresh produce to patients, and even zoos looking to grow food for animals. Their goal was to make fresh, local produce accessible and sustainable.

Over the years, Freight Farms sold over 600 units worldwide. They claimed to have created the largest connected network of hydroponic farmers in the world. The idea caught on — especially with institutions looking for sustainable food sources.

But getting the product into the world was only half the battle. The economics were tough, the customer base was niche, and the company never quite figured out how to make it all add up.

The Numbers:

• 📅 Founded in 2013

• 💰 Raised $56M+

• 🚛 Sold 600+ container farms

• 📉 Tried to go public in 2023 via SPAC — deal collapsed

• 💸 $7M in liabilities, $26K in cash at time of bankruptcy

Reasons for Failure: 

• The numbers didn’t work: Selling a container farm for $160K sounds like a good deal until you realize the market for that kind of product is pretty small. Most customers were schools, hospitals, and startups with sustainability goals, not exactly high-volume buyers. Freight Farms had some sales, but not enough to cover their costs or support real growth. The unit economics just didn’t scale.

• The SPAC fell apart: In 2023, they announced plans to go public through a SPAC at a $150M valuation. It didn’t happen. That deal would’ve brought in much-needed capital, but instead, it disappeared — and so did the momentum. A failed SPAC doesn’t kill a company on its own, but it often signals that investors are starting to lose faith.

• They ran out of cash: When they filed for bankruptcy, they had $26,000 in the bank and $7 million in liabilities. That’s not a runway — that’s a crash site. Freight Farms wasn’t a scrappy startup anymore; it had grown into a company with real overhead and no clear path to profitability. Once the money stopped flowing, there was no plan B.

• Legal and operational messes: Two suppliers sued them for unpaid bills — one even accused them of fraud. That’s not just bad optics, it’s a sign of deeper problems in how the business was being run. It’s hard to keep a company alive when you’re fighting legal battles and dodging invoices.

Why It Matters: 

• The SPAC hype cycle is full of ghosts. Freight Farms is just one more.

• They sold to schools, hospitals, even zoos — but none of those are fast-scaling customers. Selling to institutions feels stable, but it can kill your growth and cash flow.

• The market was too narrow. There just aren’t that many people who need and can run a $160K lettuce machine.

Trend

YC’s Request For Startups

Last week, Y Combinator published their Summer 2025 Request for Startups (RFS) — a document where they spell out what types of companies they want to fund. Think of it as YC’s way of whispering, “Hey, founders… build this.” And if history is any guide, paying attention is worth it: RFSs tend to predict where the startup world is heading months in advance.

This is already the third RFS of 2025, with others released in December (Winter 2025) and January (Spring 2025). So yes, they’re coming out fast, every couple of months now, and each one acts as a mini roadmap for builders.

Why It Matters:

• A crystal ball for trends: YC’s RFS is one of the best ways to spot early signals of what investors will care about in six months.

• Validation for ideas: If you’re already building something that matches one of these requests, it’s instant validation that the market might be ready.

• A blueprint for differentiation: Even if you’re not applying to YC, knowing what types of companies everyone else might build helps you stay one step ahead.

The Most Important Requests

🧠 Full-Stack AI Companies

YC isn’t just looking for tools for law firms or hospitals — they want you to replace the law firms and hospitals.

Instead of selling AI software to incumbents, build your own vertically integrated business, fully powered by AI. It’s a bold approach: start a law firm staffed with AI, not lawyers. Take on legacy players directly, not by enabling them, but by outcompeting them.

🎨 More Designer Founders

Design isn’t a finishing touch anymore — it’s a founding skill. With no-code and AI speeding up product development, design is what separates good from great. YC wants more designers at the helm. They argue that design taste, user empathy, and problem-first thinking are exactly what modern startups need from day one. As AI commodifies execution, taste becomes the moat.

🗣️ Voice AI

Despite all our tech, we’re still stuck yelling “press 9 to talk to a human.” YC sees a huge opportunity in replacing outdated phone systems with conversational AI. Voice bots are finally getting good enough to feel human and with over a trillion business calls per year, the upside here is massive.

🎓 AI Personal Tutors

Online education hasn’t lived up to its promise — yet. Most courses treat everyone the same, offering one-size-fits-all content. YC thinks AI is finally ready to change that. With multimodal models that can explain concepts visually, verbally, and interactively, we’re getting closer to building the ultimate teacher: personalized, scalable, and always available.

🧰 Internal Agent Builders

Soon, everyone in a company will have their own AI assistant automating their annoying tasks. But we need infrastructure to build those agents — tools that are secure, connected to your internal systems, and easy to use. YC already built one for themselves and use it to automate things like contract reviews and accounting. The bet here: tools that help every employee build their own copilots.

So, Whats the Trend?

This RFS makes one thing very clear: YC is done with AI tools. What they want now are AI-native companies that go full-stack — ones that own the entire product, service, and customer relationship, not just improve workflows from the sidelines.

In other words: stop building the plugin — build the platform.

The request for “full-stack AI companies” sums it up best. YC wants startups that go direct, bypassing slow-moving incumbents, and delivering the final experience themselves. Don’t help others modernize. Outrun them.

This shift echoes across the rest of the RFS. The call for voice AI isn’t about powering call centers — it’s about replacing them. The push for AI personal tutors isn’t about helping edtech platforms — it’s about becoming the edtech platform.

Even the spotlight on design founders ties in: if you’re building end-to-end, execution and taste aren’t optional — they’re your moat. YC isn’t just funding products anymore. They’re funding challengers — teams willing to rethink entire categories from first principles, with AI baked into the foundation.

Help Me Improve Failory

That's all of this edition.

Cheers,

Nico