No, OpenAI isn't killing startups

Image created by OpenAI's DALL·E

"OpenAI’s ‘Extinction Event’ For Other AI Startups" read a dramatic headline from The Information. "Beware AI Startups, OpenAI Can Take You Out Anytime" warned Gizmodo.


OpenAI DevDay was exciting for most. Apparently scary for some.

In an event that reminded many of the iPhone and Facebook Platform launches in 2007, OpenAI put their best platform foot forward.

"We believe the most incredible GPTs will come from builders in the community. Whether you’re an educator, coach, or just someone who loves to build helpful tools, you don’t need to know coding to make one and share your expertise."

This is a huge statement of intent. With one paragraph OpenAI said what many aspiring platforms struggle to say in their early days: we can't create everything for everyone, so we're arming the ecosystem to do that.

Add the imminent launch of a transactional marketplace, and OpenAI are in rarefied air - a company creating not just a platform ecosystem, but a platform economy.

So what's with all the dramatic headlines? For me, it's a déjà vu moment.

Filling gaps

Around 2012 - the year of Facebook's IPO - Facebook's Ads product and engineering groups were arguably the most under-invested. Ads were a side project, a means to an end.

"We don’t build services to make money; we make money to build better services" was Zuck's mantra. It was true.

Back then, billions of dollars of Facebook ad revenue was bought through Power Editor - an open sourced browser extension that was built atop Facebook's open Ads API. It was maintained by one or two engineers. I say maintained because it was a side project - not a core focus.

They did a great job, but Power Editor had many sharp edges. It was plagued with bugs and broke frequently. Internal teams struggled to use it.

Armed with that knowledge, clever developers saw an opportunity.

An ecosystem of startups emerged, building "AdTech for Facebook." Most started life as media agencies, who struggled - like all Facebook advertisers - with Power Editor. They built features beyond Facebook's core ads capabilities, and integrated Facebook Ad buying into existing marketing and advertising stacks. Many - like Brighter Option and Alchemy Social - sold to publicly-listed companies like Salesforce and Experian.

They created solutions far, far better than anything Facebook had built.

A quick buck

Yet, for every Brigher Option and Alchemy Social there was another developer eager to make a quick buck. They weren't motivated to create customer value. They were motivated by fast money. I encountered one developer that - eager to join our PMD program - demoed me a forked version of Power Editor, with different colours and a different logo.

It obviously didn't make the cut.

I recall another developer demoing a simple ad buying interface - basically a "lite" version of Facebook's ad create flow - that the developer made money from by arbitraging 90% of ad spend. You think you spend $100 on Facebook Ads, but in reality you'd spent just $10. 10% of the usual results, for 100% of your money. Hardly bleeding-edge innovation. Or good for customers.

That didn't make the cut, either.

It was the wild west. And we were the sheriffs in the middle of it all.

Gold rush

What I've just described happens every time a new economic opportunity emerges. In 1848 over 300,000 people rushed to California because one man struck easy gold near Sacramento. Only a few of those people emerged richer.

Isaac Newton - an apparently rational genius - lost $3 million in the 1720 South Sea Bubble. We had the 2021 NFT craze. That didn't end well.

Humans are attracted to easy money like bees to honey. We can't resist.

Obvious opportunities generate easy money. But - by their nature - they can only generate easy money for a short period. If it's obvious, everyone will want in on the action. Things that obviously need to get built, will get built.

Many of the most vocal reactions to OpenAI's launch event eminated from this cohort - developers that built obvious solutions to fill obvious gaps.


Some say they've been sherlocked. They believe their invested time was a waste of time. They forgot that - for a time - they made nice, easy money.

Much of what OpenAI announced last week was obvious and inevitable - they'd just not got round to building it yet.

But much of what they launched opens up even more innovation opportunities for value-creating developers. GPT-4 Turbo is better and cheaper. And it now includes vision capabilities.

A new Assistants API makes it super easy to integrate AI assistants in almost any environment. It includes persistent and infinitely long threads.

This, on its own, is incredible.

Heck, they even built an assistants playground to help newbies get started quickly. Text-to-speech is now a thing. A cheap, high quality thing.

A majority of developers are excited. They should be.

OpenAI DevDay 2023 will go down in history as the official "birth" of a new decade-long wave of developer-led value creation.

The polar opposite of an "extinction event."

Join the Newsletter

Each week I send fresh content to readers from companies like Meta, LinkedIn and HubSpot. Subscribe to join them.