Is This the Start of the End for OpenAI?

Despite OpenAI being one of the most talked about and impactful companies of the last few years, it’s in a very precarious position. Without some incredible innovations, it looks more likely to give way to a death spiral than market dominance.

The shine of ChatGPT has worn off, leaving us only with the OpenAI reality TV show as we watch key staff quit and rejoin at paces nobody can keep up with. Alongside this, we’ve got a product strategy that, while producing some genuinely delightful products, hasn’t produced any outright winners despite their early lead in the AI assistant market. Unless there are some miracles on the horizon, I’m firmly of the belief now that OpenAI’s initially tight grip on the world of AI is only going to keep loosening with time.

I don’t think there’s a clear thesis for why OpenAI should win in any of the product categories they’re competing in.

As the performance of models used to power AI assistants continues to converge, the value of the data that can be leveraged during inference becomes a major differentiator. Google and, to a lesser degree, Facebook have a clear advantage over OpenAI here. They have massive historic data about users, and Google has access to data across all of its platforms that it can leverage. Outside of our chat history (which Google has too with Gemini), there’s no competitive advantage for OpenAI when it comes to being able to offer the best AI assistant experience.

When it comes to building an agentic platform for enterprise automation, Microsoft is better positioned to be able to easily deploy and sell a solution to enterprises (even if OpenAI powers parts of the solutions, owning the customer relationship is where the value will accrue). And with the buzz over Claude Co-Work in recent weeks, Anthropic looks like it’s rushing ahead as far as functionality goes as well. Additionally, I think a lot of this market will be taken up by smaller companies delivering automation solutions for specific industries; this is already the direction of travel and I think it’s just going to increase as startups work with customers to craft ever-improving solutions to meet the needs of niche industries.

The market for code generation tools is already one where OpenAI is a smaller player when compared to GitHub Copilot and everything on offer from Anthropic.

And finally, I think it’s worth mentioning OpenAI’s recently released Atlas browser; I can’t be the only one who’s already pretty much forgotten about it. I think trying to compete against Chrome with the might of Google’s AI and existing customer base is going to go about as well as it has for everyone else who’s launched a browser over the last few years. If you come for the king, you best not miss.

So that’s a pretty bleak picture… what would OpenAI need to do to be THE company of the AI age?

OpenAI’s consumer product developed with Jony Ive dominates the market for AI wearables (a product category currently characterised by flops and unmet promises). We’ve seen how quickly companies were able to respond to ChatGPT and start chipping away at the market share, so this would need to be a massive reveal, with a user experience that eclipses anything that’s come before to really push them back into a dominant position.

The other path, the one I think they’re betting on, would be OpenAI jumping significantly ahead in research and delivering a breakthrough on the level of the Transformer (the tech that started the AI boom) that gives them a huge technical edge in the AI assistant market. I think with Sam Altman calling a “Code Red” to focus on ChatGPT, this is what he’s hoping to realise. Then, if OpenAI develops AGI before anyone else, all bets are off. I’ll be busy watching Terminator, but from my bedroom window rather than the TV.

So let’s wait and see; the last 2 years have been impossible to predict, so I doubt the next year will be any different.