Side Quests
This week saw a bunch of Open AI's announced products get shut down as the vibes shifted from experimentation to prioritising of enterprise and productivity, and a cull of "side quests".
Products that have been cut in the last couple of weeks include Sora (video), Adult Mode (porn) and Instant Checkout (eCommerce).
The news has triggered much grave dancing across the board, exacerbated by recent data from Ramp that suggests Claude is winning the race for new paid sign-ups in the enterprise.
The popular view this quarter is that Open AI is an overvalued, overfunded commodity on a path to an imminent meltdown. The popular view last year (and, for some, still) is that it's a generational $multi-trillion+ start-up primed to become a public company. As ever, we seems unwilling and unable to appreciate that the truth could well be somewhere in the middle.
A few thoughts in defence of Sam and friends:
- There's nothing wrong with closing things down that don't work. Any start-up that is <4 years post product launch would still be trying (and failing) at new things. The problem for Open AI is that its celebrity status and enormous valuation are increasing scrutiny on those experiments. Every false step is an "I told you so" moment observed on a global scale.
- The grave-dancing logic is inconsistent: In one breath, the narrative seems to be that LLMs in general are a commodity with no moat and no path to sustainable value. At the same time, the narrative seems to be that Anthropic are the inevitable winner because their products and models are so much better. Surely those two things can't be true at the same time? Any doomerism around LLMs in general must also apply to Anthropic, but right now it seems popular to ignore that logic in order to kick OpenAI.
- ChatGPT remains by far the most popular product: The Ramp data shows that Anthropic is leading in terms of new AI sign-ups, but in total installed base, Open AI remains the market leader. We've also seen the same theme reflected in recent data from a16z's Top 100 Consumer Gen AI Apps report:
Source: a16z
- Some experiments are working out: There was also news this week that Open AI's ad business is at $100m run-rate in less than two months, which shows pretty immense value could still be coming down the path on the consumer side. 900m weekly active users is an installed base not to be sniffed at.
The final, most obvious, point is this: everything is changing so quickly that nobody really knows. We seem to be able to swing wildly from extreme view to extreme view, without any acknowledgement that the world is shifting beneath our feet. Yes, Open AI has had a bad few months. But their team and the investors who are backing them aren't idiots, so it's a bit early to be calling "Game Over".