In June 2023, Inflection announced that it had raised $1.3 billion to build what it called “more personal artificial intelligence.” The main investor was Microsoft.
Today, less than a year later, Microsoft announced that they were essentially eating Inflection alive (although I think they worded it differently).
Co-founders Mustafa Suleyman and Karén Simonyan will go to Microsoft, where the former will lead Microsoft’s fledgling AI division, along with “several members” of their team, as Microsoft put it — or “most of the staff,” as reports Bloomberg the. Reid Hoffman will stay behind with new CEO Sean White to try to save what’s left of the company, which, I feel I should repeat, raised $1.3 billion 9 months ago and $225 million in mid 2022.
Inflection’s thesis was a conversational AI that you could talk to normally across multiple platforms and that remembered you and past conversations, making it more personal and useful. A great idea, but Pi, their AI, never really came close. It was fine, but even with that much money they couldn’t keep up with the rapid advances in capabilities and services offered by OpenAI (which Microsoft, hedging its bets, also supports), Google’s Gemini (which has the advantage in the home field. search) and Anthropic (which made its game for safe, boring applications where AI can excel).
Maybe the pitch was meant to fail: it’s not really clear that people want a personalized AI when things already sit so comfortably in the uncanny valley of humanity. They may not want their business writing expert AI, their architectural sketch AI, and their therapeutic conversation AI to know each other at all. Combining them is certainly an approach, but whether the market rejected it or Inflection’s technology just didn’t meet its goals, the Pi always seemed respectable to us.
Of course, this was extremely expensive to make, probably hundreds of millions of dollars, with, I imagine, negligible revenue, relatively speaking.
How long do you keep throwing money down the hole? When does it stop being a great game and become a waste of capital? (I’m not saying wasted effort, as the company produced a lot of valuable work in the field, and its researchers are no doubt proud.) And who makes this request?
Whoever it was, here’s “The new Inflection” now: minus its two most technically savvy co-founders (sorry Reid), minus the product he invested a fortune in (there will be no “immediate changes” to the Pi, which is basically a death warrant), minus some or most of the team that built it, and one imagines minus a lot of money contingent on performance. Hoffman and White have their work cut out for them.
The new focus, on an “AI studio business where custom AI models are built, tested, and customized for commercial clients,” would probably have been a good focus a year or two ago, but I suspect at this point they’ll be fighting over crumbs.
And then there’s Microsoft.
What am I hearing? An echo from the 90s? Embrace… extend… erase…
Is he the good guy here, rescuing a precious team from the wreckage of an enterprise that threatened to take them down with it? Or are they opportunists who backed several horses in the same race, with the express intention of gleefully gobbling up any that stumbled before the finish line?
I mean, it was only in November that we saw them tinkering with OpenAI — literally the same thing!
For a few hours, OpenAI was the fallen horse and Sam Altman and Greg Brockman and their followers were the meat! In the end, Microsoft got a little extra leverage in the company instead of eating it alive.
Now they have their second choice, and probably much cheaper. Inflection lacks the momentum of OpenAI, which was perhaps at the zenith of its ambitions so far when the coup was attempted. We’ll see if Suleyman and Simonyan can effectively lead the new AI division — I don’t know about you, but I’d be pretty skeptical of their instincts after watching a billion-dollar business evaporate so quickly. I will you should I hire them after all this?
Whatever the circumstances behind this sudden drop, it underscores the dominance of legacy tech companies in this space. Whether it was OpenAI or Inflection, Microsoft was feeding their cash and computing addictions, whispering in their ear about partnerships, and then, once they stumbled, out came the hidden fork and knife. Think Google isn’t ready to do the same thing to Anthropic? Don’t you think Apple would do the same if they could?
As always in technology: Those who can, build. Those who can’t, buy… by all means.