Raising for the Endgame: An AI Safety Founder’s Primer

 | 28 views

So you’re considering making an AI safety startup?

Now you’ll just need to do a quick fundraise so you can pay a salary and make an awesome product. From there it should be smooth sailing!

Right?

Not exactly:

avgstartup_liveliness Original simulation of the average tech company’s journey

Over the next ten years, you’ll have to contend with a 70% chance of death and a 30% survival and exit probability.

What do each of these stages look like?

Tract of Determination

In the early stages of your startup, you’ll be busy with two things: Proving product-market fit (or what I like to call impact-market fit) and raising cash from investors.

This is where you should be ready to pivot in the face of hundreds of experiments, change your name, change your product, change your industry, and generally change yourself.

To understand what happens in an average startup in a business-as-usual world1, you’ll be going through a series of venture rounds with increasing valuations as you raise for blitzscaling so you can have the impact on the world that AI safety demands:

avgstartup_valuation%201 Raising cash from VCs over the business lifecycle

Traditionally, startups only raised Series A through Series E before they did an initial public offering (IPO), becoming available on the public markets.

Today, it has become a standard to raise seed and pre-seed investments before you’re even generating revenue and companies that would traditionally go public are staying private (such as Stripe).

An absolutely crucial read for technical founders is Venture Deals, a book that goes through how the industry of VC works.

The Slope of Counterfactuals

In the early days of your business, there’s a high chance you’ll go bust. This can happen for a variety of reasons, such as co-founder disagreements, lack of real product-market fit, premature scaling, and strong competition.

It’s crucial that you pivot if your product doesn’t work, that you maintain a culture to keep the highest product standards, and that you iron out any co-founder disagreements as early as possible.

In the context of AI safety, you’re building a startup to solve a crucial risk from machine intelligence. If your solution fails or the tasks necessary to safeguard humanity at the inflection point aren’t technological, there’s a chance you want to pivot either the company or your career.

The Evergreen

Your company isn’t a startup anymore. You have thousands of AI agents running your $20 million annual recurring revenue (ARR) company and you receive calls from private equity, Anthropic, and Microsoft while you consider whether the public markets deserve your attention.

This is what many VCs call the accelerated path to generational wealth and what people in EA would call an outsized opportunity for impact, as your money can now be used for good. This is where you cash in your Founders Pledge and save people’s lives.

In deep tech, companies were often acquired after just three years. This has changed dramatically, as tech companies stay independent and we often see high-profile exits after seven to ten years.

Sources

  1. In the face of machine intelligence, your plans should be more ambitious. You are in the AI industry and you want to realize the impact of running a high-growth startup before crunch time. ↩

Related writing