OpenAI meltdown: How could Microsoft have let this happen after betting so many billions?
A quick summary of the past three days of chaos. And Redmond has questions to answer
Comment Microsoft is stuck in a hard place. It needs OpenAI co-founder and fired CEO Sam Altman back at the helm of the upstart, or working internally at the Azure giant.
Either way is awkward: if Altman returns to OpenAI, changes need to be made to prevent a fiasco like this happening again, the lab's board and some colleagues may well have to go, and questions asked about how Microsoft – as a significant investor in OpenAI - was somehow completely blindsided by this meltdown. And if Altman takes up Microsoft's offer and joins the cloud goliath, will he and any loyalists who follow him have the freedom and agility to continue shaping machine learning as they did at OpenAI?
What is clear is that right now Microsoft is trying hard to save face. It is in damage-control mode. It went all in on OpenAI, that blew up in a hugely embarrassing manner, and it must now show it has a grip on the situation.
There's no guarantee Altman will join the Windows giant or even go back to OpenAI. The vast majority of OpenAI's staff – by now reportedly 700 of 770 – have pledged they'll quit and possibly join Altman at Microsoft unless he is reinstated at OpenAI and the board resigns. Nothing so far has been officially decided or explained. This saga twists and turns every other hour.
Timeline
On Friday, the tech industry was stunned to learn the OpenAI board of directors had turned on Altman. After sacking him, the panel issued a damning statement accusing the now-former chief exec of not being "consistently candid in his communications" with the board.
It's not clear what the upstart's leader did exactly to be kicked out so abruptly, though it's obvious his fellow board directors had lost confidence in him and wanted him out. There are many theories being thrown around, and a leading one right now is that there were concerns among at least some board members that OpenAI was commercializing its content-generating neural networks at too fast a rate without sufficiently addressing the risks. Friction there led to Altman's ousting, we can well imagine.
Another rival?
Hours after Altman was chopped, OpenAI co-founder and president Greg Brockman - who had also been removed from his board chairman role on Friday - announced his resignation, and a handful of senior engineers followed suit. Rumors that the group would go off and form their own shiny new AI venture immediately began swirling around.
That would have been an obvious next move for Altman – a well-known figure in Silicon Valley who has more than a decade of experience in the startup world, and someone who would have no problem raising money and attracting talent. (OpenAI has already had a split of sorts, with ex-staff leaving a while ago to form safety-conscious AI startup Anthropic.)
OpenAI's investors knew Altman's removal was going to be sub-optimal for them and for the org. Thrive Capital, Khosla Ventures, and Tiger Global Management are among the shareholders trying to bring the top boss back to OpenAI, and are still reportedly keen for him to return.
He could just go back
There was and still is talk of Altman possibly coming back to OpenAI as long as certain conditions are met – such as changes to the organization's governance, presumably to stop another fiasco like this happening. OpenAI deliberately has a small board, and it was supposed to have the ability to quickly and easily take significant steps, like what we saw at the end of last week, to curb the lab if (say) its AI systems got out of hand.
As OpenAI's largest investor, with a 49 percent stake, Microsoft had hoped to repair the situation – after all, it has plowed more than $10 billion into the compute-hungry lab, and injected the upstart's family of content-generating, conversation-driven neural networks into all corners of the MSFT empire, from Azure and Bing to Windows and GitHub, under the Copilot brand. Microsoft wasn't about to let this big bet go south so easily; the world was surprised Microsoft was as much in the dark as the rest of us.
Microsoft signaled on Friday it was committed to OpenAI and that it had a long-term agreement to access the upstart's tech as needed to roll out products and services as planned. We can't imagine Redmond being left totally high and dry if OpenAI imploded, though it's obvious the tech giant needs the lab's whiz-kids to continue their research and development work one way or another, and that Microsoft would like some stability brought to this cluster-fsck.
Crucially, Microsoft's share price was dropping. Wall Street was losing confidence. The mega-corp had to do something – and it managed to revive its stock, and even pushed it higher, by having CEO Satya Nadella boast on Monday that Altman and Brockman will join the enterprise IT giant to lead an AI research arm (and presumably keep that GPT tech flowing into Microsoft 365 etc.)
To avoid looking like it's been left holding a busted flush, Microsoft must keep Altman, Brockman, and the OpenAI team within arm's reach, either within Redmond or at OpenAI. But there's no guarantee the team will thrive or last long within Microsoft if they get onboard, and there's no guarantee OpenAI can continue as-is with or without Altman.
The worst-case scenario for the Windows goliath, in our view, is for it to have invested heavily in a machine-learning lab that has not only lost its Steve Jobs but also the majority of its engineers and boffins, and is facing destruction.
Making Office 365 spreadsheets a bit more clever isn’t something that rallies a team like theirs
We're not the only ones thinking along these lines. Cloudflare CEO Matthew Prince was blunt in his assessment here: "Sunday morning [Microsoft] were at their lowest point. There was real risk they lose all their $11B investment [in OpenAI] and look like absolute fools for making that size of investment without any real governance controls."
"I think the chances of senior OpenAI folks still being at Microsoft in three years is asymptotically approaching zero," the Cloudflare supremo added.
"Where the independence and clear mission of OpenAI was exactly what could have kept that group of incredible talent motivated and aligned over the long term, making Office 365 spreadsheets a bit more clever isn’t something that rallies a team like theirs.
"Sure they’ll try and have some level of independence, but the machinery of a trillion dollar-plus business software behemoth is hard to not get caught up in and ground out by ... This was a very bad weekend for Microsoft."
- Copilot coming to Windows 10 to help navigate the OS's twilight years
- You get a Copilot, and you get a Copilot – Microsoft now the Copilot company
- OpenAI hits the GPT-4 Turbo button plus promises copyright shield for fans
- Microsoft 365 Copilot 'generally available' – if you can afford 300 seats
Despite Nadella's confidence earlier, a deal hasn't actually been finalized, seemingly because Altman would rather go back to the $80 billion startup he co-founded than bury himself in Redmond's corporate machine. Rumors persist that Altman may indeed return to the organization he started, and Microsoft has indicated it would be OK with that, provided those aforementioned conditions are met. The OpenAI board would most likely have to go in that case. Anything to fix the situation, stat.
Speaking of the board, OpenAI co-founder, board member, and chief scientist Ilya Sutskever - who was reportedly instrumental in the board rebellion against Altman over safety concerns regarding the lab's state-of-the-art models - has since backtracked on his defenestration of the CEO.
"I deeply regret my participation in the board's actions. I never intended to harm OpenAI. I love everything we've built together and I will do everything I can to reunite the company," he said on Monday.
Sutskever has previously, along with Altman, banged on about the potential dangers AI poses to society and the need for guardrails, regulations, and other protections to rein it in. Critics have said this fear-mongering suits OpenAI because it encourages lawmakers and watchdogs to put in place rules and requirements that smaller rivals can't cope with, leaving the market to well-resourced incumbent players like OpenAI.
The OpenAI board made CTO Mira Murati interim CEO, and after she reportedly indicated her support for Altman to return, the directors on Sunday instead installed former Twitch boss Emmett Shear as chief exec.
Ramifications
There are three things that AI companies need in order to take off: money, talent, and compute power. Under Altman, OpenAI had all three. It may not have that in the future, if the majority of the remaining staff from the 700-plus headcount follow through on their threat to resign unless Altman is reinstated as CEO.
Podcast: Where does this leave OpenAI, Microsoft, and you?
LISTEN NOWBear in mind, OpenAI offers huge compensation packages to staff, and presumably if Microsoft absorbs those workers, as it's offered to do, it'll match those figures. There's no requirement to go to Microsoft, though, and on Monday Salesforce CEO Marc Benioff threw his hiring hat into the ring to snap up disgruntled OpenAI staff.
"Salesforce will match any OpenAI researcher who has tendered their resignation full cash & equity OTE to immediately join our Salesforce Einstein Trusted AI research team under Silvio Savarese," Benioff said.
"Send me your CV direct ... Einstein is the most successful enterprise AI Platform completing 1 Trillion predictive & generative transactions this week! Join our Trusted AI Enterprise Revolution."
Doom and boom
The reasons for Altman's ousting remain opaque, though as we said there is speculation it's down to a deep ideological clash between AI doomers and boomers within the biz. Those in the doomer camp believe AI is too risky and that development must be slowed down, and they didn't approve of Altman's push to commercialize the technology so quickly. Those in the boomer camp believe the promise of AI is just too great to ignore, and it should be built as fast as possible.
If Altman does indeed switch to Redmond, and if he and his followers can execute there, it potentially spells the end for OpenAI as it exists today. Because who will Microsoft favor: an external business that may hold back on releasing state-of-the-art technology over safety concerns, or an internal team that promises to be bolder and faster in its execution to unleash more powerful models? What would then happen to the upstart lab?
Doomers should realize that Altman's abrupt exit isn't well-aligned with their vision. It has only driven him into the arms of Big Tech, whose ultimate goal is money and growth, or into setting up a competitor, or returning to OpenAI emboldened with an overhauled organization.
Altman at Microsoft would have access to a ton of funding and resources, and maybe hundreds of his former employees. Meanwhile, Microsoft would get what it really wants, at least in the short term: a chance to directly control what could eventually be the most powerful computing technology in the world. And best of all, it didn't really cost Redmond anything it can't easily afford.
Assuming, of course, that scenario plays out perfectly. As we've seen in the past few days, there's always room for one more baffling blunder. ®
Newsflash: Sam Altman is heading back to OpenAI.