> Sam Altman, the recently ousted CEO of OpenAI, has been working on a new artificial intelligence venture he is planning to launch, sources briefed on the plan said on Saturday.
Doesn't seem misleading to me.
Maybe Reuters's can't disclose or doesn't know what the venture is, but its reasonable to cite a previous, seemingly relevant report from another outlet.
This goes back to what the original commenter was saying. The rumor is that Altman was planning to launch an AI chip startup to take on Nvidia before his ejection from OpenAI.
Unattributed sources quoted by some media outlets say that Altman has recently been fund-raising for a possible new venture with potential investors.
Given the context, I assume this was probably presented to the board, at least initially, as a personal side project outside of his duties as OpenAI CEO. If so, then the "not consistently candid" in the board's statement might be related to aspects of that, perhaps in terms of degree, focus, timing or even potential fund-raising or recruiting.
Speculating, I can imagine a scenario where Altman was working with the board to orchestrate his eventual exit to start the new endeavor or perhaps a dual role. If so, the board may have concluded Altman's outside activities were more extensive than he had disclosed and crossed the fuzzy line between discussion and exploration to execution - which could obviously create untenable conflicts of interest for Altman which the board might be required act on.
I like this hypothesis because it's consistent with the statements and actions so far as well as seeming reasonable given the massive opportunity and Altman's entrepreneurial nature. Recall that GPUs have been OpenAI's largest costs and limiting factor. In fact, as OpenAI's CEO, Altman could have argued to the board that OAI needed to vertically integrate 'down the stack' to designing it's own hardware, especially since their largest competitors already have. If the board balked at this direction as inconsistent with the non-profit's charter (and potentially even adverse to the mission of "safe AI"), Altman may have asked the board for permission to personally explore the opportunity. Assuming the board agreed, there would likely have been specific disclosure requirements of Altman as well as limits on certain activities. Such a fast-moving, dynamic environment, seems like it could lead to miscommunication and at least the appearance of material omission or even misrepresentation.
Such a breakdown in trust might result in a situation where Altman could reasonably think he's still working with the board toward a negotiated and stage-managed exit but the board could reasonably conclude the situation had gotten out of control and exceeded their agreement with Altman - requiring board action to remain in compliance with the organization's charter and their legal duties as board directors.
Since some undisclosed time period before he was even sacked according to an article in The Guardian, but it offers little else in addition to the Reuters article.
That was something to do with a VC fund he was raising. This is news that he plans to start a new company following his firing. Not sure what is misleading about this at all.
OTOH everyone right now is fighting for Nvidia's GPUs, and I bet (ex)OpenAI people know exactly what hardware they'd like that would be best suited for their task.
> know exactly what hardware they'd like that would be best suited for their task.
Thats the fun thing about AI hardware. By the time you develop it (which takes years and billions), the workload you're targeting is already changed or obsolete.
Nvidia is the exception because if you are an AI scholar, you are building your research around whatever capabilies Nvidia GPUs have. Thats why AMD and Intel have quite literally chosen the "emulate Nvidia" path and axed more specialized accelerators.
I suppose if the new 'venture' is for-profit it might give credence to the view that Altman and the board diverged because of his desire to turn OpenAI toward exactly that.
From a phenomenal demo day to this revelation Sam has managed to get out from under the limits of heading a nonprofit to something that could be immensely profitable, all while engendering sympathy. I’m not saying it was orchestrated with this intent, but if it was it was a masterful bit of reputation engineering.
Until last night, Microsoft seemed to have a lock on the world's most advanced AI. Now if the top talent follows Altman to his new company, Microsoft's investment could be stranded.
We should start a pool on which Microsoft competitor will overpay the most to freeze them out of the new company.
28 comments
[ 4.0 ms ] story [ 75.5 ms ] threadDoesn't seem misleading to me.
Maybe Reuters's can't disclose or doesn't know what the venture is, but its reasonable to cite a previous, seemingly relevant report from another outlet.
What does "has been working on" mean? He's been fired for a day. When was he working on it?
Given the context, I assume this was probably presented to the board, at least initially, as a personal side project outside of his duties as OpenAI CEO. If so, then the "not consistently candid" in the board's statement might be related to aspects of that, perhaps in terms of degree, focus, timing or even potential fund-raising or recruiting.
Speculating, I can imagine a scenario where Altman was working with the board to orchestrate his eventual exit to start the new endeavor or perhaps a dual role. If so, the board may have concluded Altman's outside activities were more extensive than he had disclosed and crossed the fuzzy line between discussion and exploration to execution - which could obviously create untenable conflicts of interest for Altman which the board might be required act on.
I like this hypothesis because it's consistent with the statements and actions so far as well as seeming reasonable given the massive opportunity and Altman's entrepreneurial nature. Recall that GPUs have been OpenAI's largest costs and limiting factor. In fact, as OpenAI's CEO, Altman could have argued to the board that OAI needed to vertically integrate 'down the stack' to designing it's own hardware, especially since their largest competitors already have. If the board balked at this direction as inconsistent with the non-profit's charter (and potentially even adverse to the mission of "safe AI"), Altman may have asked the board for permission to personally explore the opportunity. Assuming the board agreed, there would likely have been specific disclosure requirements of Altman as well as limits on certain activities. Such a fast-moving, dynamic environment, seems like it could lead to miscommunication and at least the appearance of material omission or even misrepresentation.
Such a breakdown in trust might result in a situation where Altman could reasonably think he's still working with the board toward a negotiated and stage-managed exit but the board could reasonably conclude the situation had gotten out of control and exceeded their agreement with Altman - requiring board action to remain in compliance with the organization's charter and their legal duties as board directors.
Big-die silicon is hard, like unimaginably hard, even with a bottomless pit of capital.
Thats the fun thing about AI hardware. By the time you develop it (which takes years and billions), the workload you're targeting is already changed or obsolete.
Nvidia is the exception because if you are an AI scholar, you are building your research around whatever capabilies Nvidia GPUs have. Thats why AMD and Intel have quite literally chosen the "emulate Nvidia" path and axed more specialized accelerators.
We should start a pool on which Microsoft competitor will overpay the most to freeze them out of the new company.
Perhaps there is an AI hardware start-up out there Altman wants to buy.
May be Sam could start NextAI ?