By joining Microsoft, they retain access to all the data, weights, and infrastructure they had at OpenAI. They don't have to start from scratch and ramp up. They can start up right where they left off.
This is like a spacecraft research nonprofit working on faster than light travel promising Boeing 100% of the rights to any of their technology that's sub-light speed. I give even odds that they'll never achieve "AGI," or when it happens it'll be an incremental gain made by simply wiring existing technologies together that'll be obvious to any engineer competent in the field and thus easily duplicated.
The OpenAI nonprofit board does, AIUI. That means OpenAI can, in theory, cut off everything from this day forward (by declaring GPT-5 or whatever as "AGI"), but they can't cut off access to GPT-4.
What makes you think it wouldn't change after Sam and Greg join the team? AFAIK the reason Microsoft scaled down their research division (including GPUs) was because they were no where close to OpenAI despite years of investment.
Does the Microsoft deal let Microsoft continue training from e.g. the GPT-4 weights?
I guess at least it gives them access to the OpenAI models to use internally, which they kinda need as their ways of working (Greg especially) will be highly dependent on having them now.
Your comment got me thinking, it's not just all the current access to all the data, weights, and infrastructure they had at OpenAI, it's also everything that will come out of OpenAI in the future.
Remember, Microsoft has an exclusive license to all models that come out of OpenAI until they reach the pre-agreed income threshold, which given the current trajectory of OpenAI, will not happen anytime soon.
I wouldn't put it past them. There's never in history been as large and as wilful destruction of value as what we saw this weekend. The lawsuits will be fascinating.
He wasn't the guy who built it, he was the guy who got things funded. Let's see how many of the core OpenAI people join. I.e. the ones that weren't (just) there for the money / post ChatGPT.
Every once in a while, and I actually do not care much about soccer, I read comment sections in a German newspaper about soccer (please don't ask why, I have actually no clue myseld). And there, you basically have the same discussion: that player / trainer is great / sucks / rightfully / wrongfully lost his job, that club will never ever win again without person A...
It is quite intrigueing to see tge same fan / cheerleading going on when it comes to comapnies and managers. But then everything is entertainment by now...
Sam joining MS was actually one of the theories I read in the initial, or one of the first, threads about his ouster. 10 billion dollar seems like a pretty steep recruiting cost, but MS knows what they are doing, right? Right?
And still they are hiring him. Different take: You are a CEO who just spend 10 billion to bevome a minority shareholder in the latest, hotest tech start up the world has ever seen. This start-up is controlled by a non-profit so. And then this non-profit kicks out the poster child of the whole industry, and you cannot do a thing about it. Well, you have to answer to a board as well. And what do you think that board will ask you about this whole affaire?
This move makes it exactly clear what was going on. Microsoft is doing to AI what they tried to do to Internet browsers back in the day. I wonder if they'd have been successful if they'd managed to buy the board of Netscape.
I suspect it's rather possible that there will be an ungodly-massive lawsuit in the offing.
All the claims about how OpenAI's board desperately wanted Altman back were based on leaks from "people close to Altman" which the press uncritically lapped up.
If it wasn't clear before, it should be clear in hindsight that the board's desire to welcome Altman back was, at best, overstated.
The leaks were probably an attempt to pressure the board or, failing that, undermine OpenAI.
This actually seems like a decent compromise. Sam and Greg can retain velocity on the product side without having to spin up a whole new operation in direct competition with their old levers of power, and Ilya + co can remain in possession of the keys to the kingdom.
Ilya and co are going to get orphaned, there’s no point to the talent they have if they intend to slow things down so it’s not like they’ll remain competitive. The capacity that MSFT was going to sell to OpenAI will go to the internal team.
Maybe they want it that way and want to move on from all the LLM hype that was distracting them from their main charter of pushing the boundaries of AI research? If yes, then they succeeded handsomely
Maybe I'm reading too much into it, but for me it is us framed as if they won't be working on GPT-based products, but on research.
The whole thing reads like this to me: "In hindsight, we should've done more due diligence before developing a hard dependency on an organization and its product. We are aware that this was a mistake. To combat this, we will do damage control and continue to work with OpenAI, while developing our in-house solution and ditching this hard dependency. Sam & Co. will reproduce this and it will be fully under our control. So rest assured dear investors."
How do you conduct research with sales people? even if they manage to bring in researchers from OpenAI, the only gain here is microsoft getting some of the researchers behind the products and/or product developers.
Well, the same way a man with drive, discipline and money but very little in the way of technical expertise can build a company.
Sometimes you need someone who can drive a project and recruit the right people for the project. That person does not always need to be a subject matter expert.
Except they only had AI model velocity and not product velocity. The user-side implementation of chatGPT is actually quite below what would be expected based on their AI superiority. So the parts that Sam & Greg should be responsible for are actually not great.
This is kind of true, I think programming even codellama or gpt3.5 is more than enough and gpt-4 is very nice but what is missing is good developer experience, and copy-pasting to the chat window is not that.
If I recall correctly, Mira Murati was actually the person responsible for productizing GPT into a Chatbot. Prior to that, OpenAI's plan was just to build models and sell API access until they reach AGI.
I know there's a lot of talk about Ilya, but if Sam poaches Mira (which seems likely at this point), I think OpenAI will struggle to build things people actually want, and will go back to being an R&D lab.
Sam and Greg were responsible for everything including building the company, deciding on strategy, raising funding, hiring most of the team, coordinating the research, building the partnership with Microsoft and acquiring the huge array of enterprise customers.
To act like they were just responsible for the "UI parts" is ridiculous.
I'm the first to defend CEOs and it's not a popular position to be in usually, believe me. But in this case, they did an experiment and it blew up based on their model's superiority alone.
Product-wise, however, it's looking like good enough AI is being commoditized at the pace of weeks and days. They will be forced to compete on user experience and distribution vs the likes of Meta. So far OpenAI only managed to deliver additions that sound good on the surface but prove not to be sticky when the dust settles.
They have also been very dishonest. I remember Sam Altman said he was surprised no one built something like chat GPT before them. Well... people tried but 3rd parties were always playing catch-up because the APIs were waitlisted, censored, and nerfed.
a) Meta is not competing with OpenAI nor has any plans to.
b) AI is only being commoditised at the low-end for models that can be trained by ordinary people. At the high-end there is only companies like Microsoft, Google etc that can compete. And Sam was brilliant enough to lock in Microsoft early.
c) What was stopping 3rd parties from building a ChatGPT was the out of reach training costs not access to APIs which didn't even exist at the time.
a) Meta is training and releasing cutting-edge LLM models. When they manage to get the costs down, everyone and their grandma is going to have Meta's AI on their phone either through Facebook, Instagram, or Whatsapp.
b) Commoditization is actually mostly happening because companies (not individuals) are training the models. But that's also enough for commoditization to occur over time, even on higher-end models. If we get into the superintelligence territory, it doesn't even matter though, the world will be much different.
c) APIs for GPT were first teased as early as 2020s with broader access in 2021. They got implemented into 3rd party products but the developer experience of getting access was quite hostile early on. Chat-like APIs only became available after they were featured in ChatGPT. So Sam feigning surprise about others not creating something like it sooner with their APIs is not honest.
Researchers can not be 'AI visionaries', almost by definition, as you focus on depth instead of breadth as a competent researcher.
Someone like Sam Altman is indeed more of a visionary than every hardcore AI researcher. The job here is to not push the boundaries of science, it is to figure out and predict the cascading effects of a new invention.
It really does feel like that. Like it’s mutually beneficial for Sam and MSFT to team up in the short term while Sam figures out his next move and MSFT tries to keep OpenAI afloat for the time being
If your goal is to produce a lot of value and you don’t care about others capturing it, then it may actually be a good way to go, especially with the non-profit setup.
But Satya is making a few 100 mil a year, tops. Sam could easily make himself a billionaire with one raise. And who wants to control all of Microsoft, that's a whole lot of headaches
I think the only edge cases are for executives of companies, and even then it's pretty limited, but I imagine this could be one of the examples. IANAL though - it's just from what I've seen discussed elsewhere.
The new models and data would stay at OpenAI. You can have thousands of researchers and compute, but if you don’t have “it”, you are behind (ask Google).
In Microsoft he still has access to the models, and that’s all he needs to execute his ideas.
> most of the current OpenAI staff would follow him
Source please? This just keeps getting repeated but there’s extremely limited public support and neither Sam’s nor the board’s decisions indicate he has a whole lot of leverage.
And if governments squeeze on AI your start up is worth pennies over night. Earning 100 MILLION per year already removes any possible financial restrictions you had. Why do you need to have 10x that? Heck even earning "just" 10 millions per year will make all of your financial concerns go away.
I suspect for people like Sam who are compulsively ambitious and competitive, it's not about the dollars. It's about winning.
Further, based on anecdotes from friends and Twitter who know Sam personally, I'm inclined to believe he's genuinely motivated by building something that "alters the timeline", so to speak.
AGI is decades if not centuries away. Cranking a plausible sentence generator to be even more plausible will not get there. I do not understand how people suddenly completely lost their minds.
The hype wave really is something else, eh? People are suddenly talking as if these advanced chatbots are on the precipice of genuine AGI that can run any system you throw at it, it's absolute lunacy
Sam is rich, I assume being CEO of one of the worlds largest companies is a far greater award than extra money when you're at the billionaire level, especially at 38. But I do think this is probably non-compete related too.
Lol, maybe. Ballmer was a friend of Gates, was 44 years old and had worked at Microsoft for 20 years (2000-1980) already when he became CEO. Nadella was also forty-something and had worked at Microsoft for 22 years (2014-1992) when he got the job.
So Sam & Greg can stay focus on their work rather than getting distracted by all the lawsuits. It isn’t a bad thing. Just not sure how they can get they want under the corporate culture?
It's complicated. In the case of the CEO it is possibly enforceable. But going to the primary funder, after being fired in a move without notification of that same funder? Likely with long complicated contracts that may contemplate the idea of notification of change of executive staff?
I don't know, even of strictly "enforceable" I doubt we will see it enforced. And if so. I'm sure the settlement will be fairly gentle.
Edit: Actually, a quick skim of the relevant code, the only relevant exception seems to be about owners selling their ownership interest. Seemingly, since Sam doesn't own OpenAI shares, this exception would seem to not apply.
non-competes are extremely hard to enforce in California. Sam would literally have to download Open AI trade secrets into a USB drive to get in trouble.
That is only the case for rank and file employees. From my understanding executives, particularly ones with large equity stakes, are not exempt from non-competes. Sam doesn't have equity though, and I am not sure if non-profit status changes anything, but regardless I suspect any non-compete questions would need to be settled in court. Probably not something to stop Sam from starting a competitor as he could afford the lawyers and potential settlement. I suspect the MSFT move has more to do with keeping the ball rolling and keeping Satya happy.
> From my understanding executives, particularly ones with large equity stakes, are not exempt from non-competes.
Your understanding is incorrect. There are some exceptions where noncompetes are allowed in California, but they mostly involve the sale or dissolution of business entities as such. There is no exception for executives, and none for people who happen to have equity stakes of any size.
> Do anti-compete clauses work when you’ve been ousted?
In jurisdictions where they are enforceable, yes, they generally are not limited based on the manner the working relationship terminated (since they are part of an employment contract, they might become void if there was a breach by the employer.)
It certainly sounds out of line with all the reporting that Altman was talking about starting a new company and could trivially fundraise for it. Was that just as much kayfabe as the idea of bringing him back?
Presumably they’ll both get their C-level positions out of the gate (for that AI entity MS is setting up specially for this) so not just “mere” employees.
But, yeah, kind of confusing, especially for Altman.
He was the kind of guy on the way to become worth $100 billion and more, with enough luck, meaning to be the next Musk or Zuckerberg of AI, but if he chooses to remain inside a behemoth like MS the “most” that he can aspire to is a few hundred millions, maybe a billion or two at the most, but nothing more than that.
They're likely going to be the ones who manage the OpenAI relationship...what better way to fuck the people who fucked them than by becoming the ones who literally control the resources that they need?
Hilarious. The look on Ilyas face when these two show up at the office for their "sync", or perhaps he's ordered to travel to a location of the owner/client's choosing.
Azure was already second nature for OpenAI and so there is very little friction in moving their work and infrastructure. The relationships are already there and the personnel will likely follow easily as well.
They are also likely enticed by the possibility of being heads of special projects and AI at the second largest tech company, meaning deep pockets, easy marketing and freedom to roam.
They will probably run a subsidiary under the MS umbrella and profit hugely in the next few years. Also, MS could easily dump OAI in the next few months to year.
Sam had no stake in OpenAI. So, any potential deca billion value is hypothetical. He would have to do a U-turn and fight with the board to get his cut. Now he'll get his cut from MS. This AI division will have some further restructuring.
OpenAI continues to develop core AI offered over API. Microsoft builds the developer ecosystem around it -- that's Sam's expertise anyway. Microsoft has made a bunch of investment in the developer ecosystem in GitHub and that fits the theme. Assuming Sam sticks around.
Also, the way the tweet is worded (looking forward to working with OpenAI), seems like its a truce negotiated by Satya?
I doubt it would be hard for Microsoft to rebuild, Microsoft Research has made many excellent contributions to transformers for many years now, DeepSpeed is a notable example.
I don’t think they’ve had the will/need to have done this but they most likely already have the talent.
Much as LLM is essentially industrial strength gaslighting, so is the meta around it.
It's not so important. There's not much there. No it's not going to take your jobs.
I am old enough to remember not only the How Blockchain Is Solving World Hunger articles but the paperless office claims as well -- I was born within a few weeks of the publication of the (in)famous "The Office of the Future" article from BusinessWeek.
Didn't happen.
No, a plausible sentence generator is just that: the next hype.
In fact some of the hustlers behind it are the same as those who have hustled crypto. Someone got to hold the bag on that one but it wasn't the rich white techbros. So it'll be here. Once enough companies get burned when the stochastic parrot botches something badly enough to get a massive fine from a regulator or a devastating lawsuit, everyone will run for the hills. And again... it won't be the VCs holding the bag. Guess who will be. Guess why AI is so badly hyped.
If you think the ChatGPT release happening within a few weeks of the collapse of FTX is a coincidence I have ... well, not a bridge but an AI hype to sell to you and in fact you already bought it.
OpenAI is doing a lot more work than just a LLM, despite that being there headline product for now. I'd rather have OpenAI leading the way than Microsoft or Google in this stuff. Despite it's own issues.
I get your pessimism, but the same has been said about a lot of tech that did go on to change the world, just because a lot of people made a lot of noise about previous tech that failed to come to anything doesn't mean to say this is the same thing, it's completely different tech.
A lot of OpenAI's products are out in the real world and I use them everyday, I never touched Crypto, now maybe LLM's won't live up to the hype, but OpenAi's stuff is already been used in a lot of products, used by millions of users, even Spotify.
'A plausible sentence generator is just that: the next hype' - Maybe, but AI goes far beyond LLM as does the products OpenAI produces.
While it can’t plug and play replace and employee yet in my experience at least every dev I see now has it open on their second screen and send it problems all day.
Comparing it to crypto and building that weird narrative you have is just not at all connected to the reality of what the product can actually do right now today.
It's probabilistic and not factual and so everything it outputs must be treated as something the actual answer might sound like and needs to be counterchecked anyways. If I am researching the actual answer already then why bother?
At this point in time a new AI company would be bottle-necked by lack of NVIDIA GPUs. They are sold out for the medium term future.
So if Sam and Greg were to start a new AI company, even with billions of initial capital (very likely given their street cred) they would spend at a minimum several months just acquiring the hardware needed to compete with OpenAI.
With Microsoft they have the hardware from day one and unlimited capital.
At the same time their competitor, OpenAI, gets most of the money from Microsoft (a deal negotiated by Sam, BTW).
So Microsoft decided to compete with OpenAI.
This is the worst possible outcome for OpenAI: they loose talent, pretty much loose their main source of cash (not today but medium to long term) and get cash rich and GPU-rich competitor who's now their main customer.
Satya probably offered the one resource they couldn’t buy at the scale/speed they need: GPUs. Both time on Azure’s cloud, as well as promise of some of the first Azure Maia 100 and Cobalt 100 chips.
Satya probably offered the one resource they couldn’t buy at the scale/speed they need: OpenAI models & future work. Altman wouldn't have had (legal) access to these anywhere else, and Microsoft wouldn't have had Sam Altman controlling OpenAI tech in any other arrangement. This arrangement may be the best for all involved: Microsoft gets it's LLM geegaws based on OpenAI tech, Altman gets to build GPT marketplaces and engage whatever growth-hacking schemes he can dream of that may have been found distasteful by colleagues at OpenAI, and OpenAI can focus on the core mission and fulfilling contractual obligations to Microsoft
I foresee this new group building on top of (rather than completing with) OpenAI tech in the near-to-mid term, maybe competing in the long term of they manage to gather adequate talent, but it's going to be going against the cultural corporate headwinds.
I wonder if Microsoft will tolerate the hardware side-gig and if this internal-startup will succeed or if it will end up being a managed exit to paper over OpenAIs abrupt transition (by public company standards). I guess we'll know in a year if he'll transition to an advisory position
I bet there was no hardware side-gig. More likely it was a ruse to trigger the push from openai, so they can exfiltrate gpt5 to MS. Openai won't exist soon, since they rely on vouchers from MS to run. I can't see MS being a very forgiving partner, after being publicly blindsided, can you?
No. They need a lot of money and computation resources to work on. In order to continue their work, they either A). raise a massive fund B). be employed by a big corp. There's no surprise they chose the latter. After all, MS has a research department on this domain.
Special unit mate... Gonna have special rules. You think these cats are gonna be in the basement pushing papers? This is grade AAA talent that can go anywhere including a fresh outfit with 1 billion in the bank VC money day 1.
Don't believe me? Check out the VC tweets... Sand hill pulled the checkbook the moment these guys might have been on the market.
I think Sam's goal is to create AGI, same as most of the other founders of OpenAI. If he just wanted money and power, he probably would have continued with YC or some other startup instead of joining the nonprofit and unproven OpenAI at the time.
His opinion on the ideal path differs from Ilya's, but I'm guessing his goal remains the same. AGI is the most important thing to work on, and startups and corporations are just a means of getting there.
I imagine Sam's vision, both before and after this company change, is that he'll keep improving GPTs, while also setting up a thriving ecosystem through APIs, and AI will become a trillion dollar industry with him at the center.
From there, maybe someone will come up with the revolutionary advance necessary to reach AGI. It may not necessarily be under his company, but he'll be the super successful AI guy and in a pretty strong position to influence things anyway.
Just because that's the goal they have written on the tin doesn't mean that that is/was their actual goal.
Especially in the early days where the largest donor to OpenAI was Musk who was leading Tesla, a company way behind in AI capabilities, OpenAI looked like an obvious "Commoditize Your Complement" play.
For quite some time where they were mainly publishing research and they could hide behind "we are just getting started" that guise held up nicely, but when they struck gold with Chat(GPT), their was more and more misalignment between their actions and their publicly stated goal.
It's gonna be a special unit. He's not gonna be an employee.
Once you lead at that level... It's max autonomy going forward. Source: Elon. Guy hates a board with power as much as Zuckerberg. Employee? Ha
.. Out of the question.
There are more structures available than simply gobbling something up and everyone is your employee.
See openai investment with technology transfers and sunset clauses. They just did a new dance.
They'll prod do something special for these guys.
They would never be employees. That's for non Sam Altman's and non Brockmans. Brockman is prob already a billionaire from openai shares. No employees here. Big boys.
I guess they were fired exactly for this reason: more money, less research and being actually "open". A "non-profit" called "Open"AI hiding GPT-4 behind a paywall with no source code with just a few hints in the papers, surreal.
Or just accept that their image is overinflated just because they happened to be in the right place at the right time. Ofcourse they had a hand on building that successful team but do not underestimate the fact that, that successful team was build with the promise of nonprofit, AI for the benefit of all And few of them would have joined Microsoft out of principle.
They won't have to worry about raising capital or getting access to GPUs, and they've likely been promised a high degree of autonomy, almost certainly reporting directly to Nadella.
We don't know the structure of their new unit, do we? Sometimes "startup in a big corp" may really bring the best of both worlds (although in reality, 90% of such initiatives bring the worst of the two worlds).
For many years, Microsoft Research had a reputation for giving researchers the most freedom. Probably even that's the reason why it hasn't been as successful as other bigcorp research labs.
"OpenAI, the next huge tech company to rival Google" is a huge loser of this whole process, probably dead.
"OpenAI, the non-profit who only has a for-profit subsidiary to get enough resources to fund its mission to develop AGI" is probably a winner, and gets to live instead of slowly die.
This was most unexpected. Hopefully it's a compromise reached mutually. Now OpenAI can fulfill what their mission is and these guys can work on AI products at MS.
I thought for sure the only two outcomes were that Altman raises money for a new startup or he comes back to OpenAI with a new governance structure (which is still a wild and crazy outcome, but crazier things have happened). Now that this happened though, I feel stupid for not considering this as a possible outcome at all.
The whole timeline of events over the last two events still leaves me scratching my head though.
It's confusing because no one beyond the direct negotiating parties knows exactly why any of this is happening in the first place. The media scoops about commercialization disputes don't seem that important to warrant such a dramatic showdown
Did Microsoft just effectively acquire 100% of OpenAI without actually needing to cross the 49% regulatory hurdle? Depending on who follows them to MSFT, that might actually be the outcome here. And OpenAI is suddenly Xerox Parc.
I don't know, way before all of this drama started, the rumors were that he was barely contributing any original or significant ideas to research and the grounbreaking ideas had come from lower level researchers.
Haven't worked at Microsoft, but usually, when folks up high have their balls at stake, resources and budgets magically start getting approved faster than the Concorde.
This is no research group, this is OpenAI 2.0, Sam/Greg will have enormous autonomy. It will be foolish to think Satya just recruited them to tangle them in MSFT bureaucracy
BigCos generally have a hard time keeping their autonomous groups actually isolated from bureaucracy. Lab126 has been thoroughly corporatized, and Area 120 got outright reabsorbed.
I think people underestimate how much of a company’s value is in their key leadership, select talent, and technology. When a company is acquired those are typically the reasons to do so other than pure revenue acquisition. Microsoft already has their technology, now has the key leadership, and will soon have the select talent.
Pirate more like. He's not just poaching "talent" he has likely stolen IP and will hope to destroy OpenAI in court costs. Microsoft is a terrible company and I hope this backfires on them.
Nadella was heavily involved in the talks to get Altman et al. back in OpenAI. This must have been brought up, so I’m guessing the OpenAI board made their decision knowing this would be the outcome?
This might turn out to be a lot more stable structure long term: the commercialization of AI under Microsoft's brand, with Microsoft's resources, and the deep research into advanced AI under OpenAI. This could shield the research division of OpenAI from undue pressure from the product side, in a way that it probably couldn't when everything was under one roof.
Ugh. I’m not keen on AGI being an eventual Microsoft product, or after this circus, even the hangers on at Open AI. Hope it’s still decades off and this all is a silly side show footnote.
I find your theory more plausible. Microsoft, Google and Amazon were lagging in AI. You can simply look at their voice assistants for an example. That's why they started investing billions in OpenAI and other think tanks in this space. Now capital turns things around to be as they should (from their perspective) and reacquires control.
MS/G/A didn’t put this into voice assistants not because they don’t have it, but because it doesn’t scale to fit the commercials at the moment. Google invented transformers and Deepmind had GPT scale LLM’s at least a year before CGPT came out.
Altman just rushed everyone’s hand by publishing it into the world at cost
My friends and family had an awful opinion of AI in general because it was the voice assistants were sold to them as the best example of AI. That changed with ChatGPT.
Google invented really useful AI but failed to deliver. OpenAI did so in record time. Now it's Google that's playing catching up with the technology they invented themselves, ironically.
But my comment applies more to Microsoft and Amazon, tbh.
This wasn’t a result of product genius in this case - OAI just didn’t have the regulatory and PR oversight that big tech has - I know for a fact Meta and Google had CGPT equivalent models ready but couldn’t launch them as they’d get rightfully berated for the model being racist or hallucinating. Things OpenAI avoided because it’s a startup non-profit.
And OAI delivered with enormous per-user cost that doesn’t scale - in an app that is a showcase and doesn’t really have latency requirements as people understand it’s a prototype.
And the vas majority of people play with CGPT, they don’t use it for anything useful. Incidental examples of friends and family of tech workers to the side.
Incredible how much has changed in one weekend… or not?
Confused what this really means. So Microsoft still has access to OpenAI’s pre-AGI tech that Sam and Greg can leverage for their more product-focused visions.
More than that, it looks like Microsoft has become a major AI player (internal research) overnight up with the likes of Meta, Google, and OpenAI. Incredible.
Microsoft was already set to spent 27 billion usd on research for 2023. They dedicate huge standout double digit percentages of budget to research every year. Their in house AI research division was already huge.
They didn't become a major AI player overnight... They already were long ago.
OpenAI is small, in raw numbers of AI researchers, compared to the big players in the space. That's a major reason why it's so compelling that they have been able to consistently set the bar for state of the art.
They were a dream team... But small. Msft is adding AAA+ talent to their existing A+ deck. Also they won't have to rewrite the code base. Can hit the ground running.
Lastly, there is no evidence that openai has the greatly quoted and so hard to define 'agi'. That's Twitter hearsay and highly unlikely... If folkes can even agree what that is. By the overwhelming percentage of definitions... Even gpt-5 is unlikely to meet that bar. Highly speculative. Twitter is a cesspool of conspiracy theory... Don't believe everything you read.
Hopefully this motivates a lot of people who don't want Microsoft to be the AI company. Slowing down research would mean Microsoft wins everything.
MS now has both the accelerationists and the deccelerationists. They can keep accelerating themselves when pushing for regulatory capture through their deccelerationist branch to slow down any competition.
1,368 comments
[ 4.4 ms ] story [ 413 ms ] threadI was confused when the whole thing was going down.
I was more confused when the whole "board wants to backtrack and maybe resign" thing was going down.
I got even more confused when Emmett Shear was announced as the CEO.
...but never in a hundred years would I have imagined "haha just join Microsoft" as an actual alternative.
I remain, confused.
I think everyone is confused.
Wouldn’t surprise me if Sam and Greg are back on the startup path by week end.
This just seems like PR to give MS a way to paper things over after such an abrupt firing.
I guess at least it gives them access to the OpenAI models to use internally, which they kinda need as their ways of working (Greg especially) will be highly dependent on having them now.
Remember, Microsoft has an exclusive license to all models that come out of OpenAI until they reach the pre-agreed income threshold, which given the current trajectory of OpenAI, will not happen anytime soon.
I think Sam just took the easier route to rebuild OpenAI within MSFT.
Now the trouble comes to the SV VCs they now will be furious.
OpenAI already has a very clear business model, that is selling completion/chat/agent API based on their model. What they need is to productize it.
Their roadmap is GPT4/5/6/7
Like if they don’t like OpenAI they can go to 10 other places that pay more and treat researchers better than MSFT
MSRA invented ResNet. MSFT also contributed DeepSpeed to the open source, which is critical in OSS LLM scene.
It is now more of just a branding thing. It will become the new cool again.
And OpenAI? After this week, how would the people view them? Definitely not envious or prestige.
Or you think Ilya wrote every line of code of GPT4?
It is quite intrigueing to see tge same fan / cheerleading going on when it comes to comapnies and managers. But then everything is entertainment by now...
Sam will leave soon enough to start his own thing, but in the meantime there is no narrative problem for MSFT to deal with
This move makes it exactly clear what was going on. Microsoft is doing to AI what they tried to do to Internet browsers back in the day. I wonder if they'd have been successful if they'd managed to buy the board of Netscape.
I suspect it's rather possible that there will be an ungodly-massive lawsuit in the offing.
If it wasn't clear before, it should be clear in hindsight that the board's desire to welcome Altman back was, at best, overstated.
The leaks were probably an attempt to pressure the board or, failing that, undermine OpenAI.
The whole thing reads like this to me: "In hindsight, we should've done more due diligence before developing a hard dependency on an organization and its product. We are aware that this was a mistake. To combat this, we will do damage control and continue to work with OpenAI, while developing our in-house solution and ditching this hard dependency. Sam & Co. will reproduce this and it will be fully under our control. So rest assured dear investors."
Can you help me understand how you came to the conclusion?
Sometimes you need someone who can drive a project and recruit the right people for the project. That person does not always need to be a subject matter expert.
I know there's a lot of talk about Ilya, but if Sam poaches Mira (which seems likely at this point), I think OpenAI will struggle to build things people actually want, and will go back to being an R&D lab.
To act like they were just responsible for the "UI parts" is ridiculous.
Product-wise, however, it's looking like good enough AI is being commoditized at the pace of weeks and days. They will be forced to compete on user experience and distribution vs the likes of Meta. So far OpenAI only managed to deliver additions that sound good on the surface but prove not to be sticky when the dust settles.
They have also been very dishonest. I remember Sam Altman said he was surprised no one built something like chat GPT before them. Well... people tried but 3rd parties were always playing catch-up because the APIs were waitlisted, censored, and nerfed.
b) AI is only being commoditised at the low-end for models that can be trained by ordinary people. At the high-end there is only companies like Microsoft, Google etc that can compete. And Sam was brilliant enough to lock in Microsoft early.
c) What was stopping 3rd parties from building a ChatGPT was the out of reach training costs not access to APIs which didn't even exist at the time.
a) Meta is training and releasing cutting-edge LLM models. When they manage to get the costs down, everyone and their grandma is going to have Meta's AI on their phone either through Facebook, Instagram, or Whatsapp.
b) Commoditization is actually mostly happening because companies (not individuals) are training the models. But that's also enough for commoditization to occur over time, even on higher-end models. If we get into the superintelligence territory, it doesn't even matter though, the world will be much different.
c) APIs for GPT were first teased as early as 2020s with broader access in 2021. They got implemented into 3rd party products but the developer experience of getting access was quite hostile early on. Chat-like APIs only became available after they were featured in ChatGPT. So Sam feigning surprise about others not creating something like it sooner with their APIs is not honest.
Good get by MS though!
Weird situation for him.
Someone like Sam Altman is indeed more of a visionary than every hardcore AI researcher. The job here is to not push the boundaries of science, it is to figure out and predict the cascading effects of a new invention.
ding ding ding
It sounds very out of line of what you'd expect.
https://www.ottingerlaw.com/blog/executives-should-not-ignor...
https://leginfo.legislature.ca.gov/faces/codes_displaySectio...
In Microsoft he still has access to the models, and that’s all he needs to execute his ideas.
Source please? This just keeps getting repeated but there’s extremely limited public support and neither Sam’s nor the board’s decisions indicate he has a whole lot of leverage.
Greed is hell of a thing
Further, based on anecdotes from friends and Twitter who know Sam personally, I'm inclined to believe he's genuinely motivated by building something that "alters the timeline", so to speak.
I am old enough to remember the "How Blockchain Is Solving the World Hunger Crisis" articles but this new wave is even crazier.
So, like 15 year old?
I do think it's funny how the Blockchain Consultants have become AI Consultants though.
[1] https://microsoft.gcs-web.com/node/31056/html
Do you have a source for your assertion?
All the other somewhat reliable sources do not have him as one.
So what is your source for your assertion?
Except if Sam and Greg have some anti-compete clauses. If they join MS, they have a nice 10 billion USD leverage against any lawsuites.
I don't know, even of strictly "enforceable" I doubt we will see it enforced. And if so. I'm sure the settlement will be fairly gentle.
Edit: Actually, a quick skim of the relevant code, the only relevant exception seems to be about owners selling their ownership interest. Seemingly, since Sam doesn't own OpenAI shares, this exception would seem to not apply.
https://leginfo.legislature.ca.gov/faces/codes_displaySectio....
Your understanding is incorrect. There are some exceptions where noncompetes are allowed in California, but they mostly involve the sale or dissolution of business entities as such. There is no exception for executives, and none for people who happen to have equity stakes of any size.
In jurisdictions where they are enforceable, yes, they generally are not limited based on the manner the working relationship terminated (since they are part of an employment contract, they might become void if there was a breach by the employer.)
Microsoft is happy. They get to wrap this movie before the markets open.
Edit: I also agree with bayindirh below. These things can both be true.
Also, that doesn't mean Microsoft won't collect the outcome of this deal with its interest over time. Microsoft is the master of that craft.
Microsoft did not offer this because they're some altruistic company which wanted to provide free shelter to a unfairly battered, homeless ex-CEO.
But, yeah, kind of confusing, especially for Altman.
He was the kind of guy on the way to become worth $100 billion and more, with enough luck, meaning to be the next Musk or Zuckerberg of AI, but if he chooses to remain inside a behemoth like MS the “most” that he can aspire to is a few hundred millions, maybe a billion or two at the most, but nothing more than that.
Was he though? If I understand correctly he didn’t have any equity in the for profit org. Of OpenAI.
IIRC he also publicly said that he doesn’t “need” more than a few hundred million (and who knows, not inconceivable that he might actually feel that).
Azure was already second nature for OpenAI and so there is very little friction in moving their work and infrastructure. The relationships are already there and the personnel will likely follow easily as well.
They are also likely enticed by the possibility of being heads of special projects and AI at the second largest tech company, meaning deep pockets, easy marketing and freedom to roam.
Oh, and those GPUs.
Edit: Sam is CEO of the new AI division.
OpenAI continues to develop core AI offered over API. Microsoft builds the developer ecosystem around it -- that's Sam's expertise anyway. Microsoft has made a bunch of investment in the developer ecosystem in GitHub and that fits the theme. Assuming Sam sticks around.
Also, the way the tweet is worded (looking forward to working with OpenAI), seems like its a truce negotiated by Satya?
Apparently Microsoft already had plans to spend $50 billion on cloud hardware.
Now they are getting software talent and insider knowledge to replace OpenAI software with in-house tech built by Sam, Greg and others that will join.
Satya just pulled a kill move on OpenAI.
Not sure if its obvious that people would leave OpenAI in troves to join Microsoft just to be with Sam.
I don’t think they’ve had the will/need to have done this but they most likely already have the talent.
Either there in house team wins out and Microsoft wins.
Or OpenAI wins out and Microsoft wins with there exclusive deal and 75% of OpenAI profits.
Better to have two horses in the race in something so important, makes it much harder than one of the other companies will be the one to come out top.
Much as LLM is essentially industrial strength gaslighting, so is the meta around it.
It's not so important. There's not much there. No it's not going to take your jobs.
I am old enough to remember not only the How Blockchain Is Solving World Hunger articles but the paperless office claims as well -- I was born within a few weeks of the publication of the (in)famous "The Office of the Future" article from BusinessWeek.
Didn't happen.
No, a plausible sentence generator is just that: the next hype.
In fact some of the hustlers behind it are the same as those who have hustled crypto. Someone got to hold the bag on that one but it wasn't the rich white techbros. So it'll be here. Once enough companies get burned when the stochastic parrot botches something badly enough to get a massive fine from a regulator or a devastating lawsuit, everyone will run for the hills. And again... it won't be the VCs holding the bag. Guess who will be. Guess why AI is so badly hyped.
If you think the ChatGPT release happening within a few weeks of the collapse of FTX is a coincidence I have ... well, not a bridge but an AI hype to sell to you and in fact you already bought it.
I get your pessimism, but the same has been said about a lot of tech that did go on to change the world, just because a lot of people made a lot of noise about previous tech that failed to come to anything doesn't mean to say this is the same thing, it's completely different tech.
A lot of OpenAI's products are out in the real world and I use them everyday, I never touched Crypto, now maybe LLM's won't live up to the hype, but OpenAi's stuff is already been used in a lot of products, used by millions of users, even Spotify.
'A plausible sentence generator is just that: the next hype' - Maybe, but AI goes far beyond LLM as does the products OpenAI produces.
While it can’t plug and play replace and employee yet in my experience at least every dev I see now has it open on their second screen and send it problems all day.
Comparing it to crypto and building that weird narrative you have is just not at all connected to the reality of what the product can actually do right now today.
At this point in time a new AI company would be bottle-necked by lack of NVIDIA GPUs. They are sold out for the medium term future.
So if Sam and Greg were to start a new AI company, even with billions of initial capital (very likely given their street cred) they would spend at a minimum several months just acquiring the hardware needed to compete with OpenAI.
With Microsoft they have the hardware from day one and unlimited capital.
At the same time their competitor, OpenAI, gets most of the money from Microsoft (a deal negotiated by Sam, BTW).
So Microsoft decided to compete with OpenAI.
This is the worst possible outcome for OpenAI: they loose talent, pretty much loose their main source of cash (not today but medium to long term) and get cash rich and GPU-rich competitor who's now their main customer.
They already do, though, has everyone forgot they got a Microsoft Research division?
Or they write the AI that runs on your M3
That said the Microsoft offer came quickly than Amazon can deliver a 3090 to your house so…
I foresee this new group building on top of (rather than completing with) OpenAI tech in the near-to-mid term, maybe competing in the long term of they manage to gather adequate talent, but it's going to be going against the cultural corporate headwinds.
I wonder if Microsoft will tolerate the hardware side-gig and if this internal-startup will succeed or if it will end up being a managed exit to paper over OpenAIs abrupt transition (by public company standards). I guess we'll know in a year if he'll transition to an advisory position
Which means, starting a competing startup means they can’t use it.
Which makes their (potential) competing startup indistinguishable from the (many) other startups in this space competing with OpenAI.
Does Sam really want to be a no-name research head of some obscure Microsoft research division?
I don’t think so.
Can’t really see any other reason for this that makes sense.
Desperate... Right...
The guy met with the Arabs a few weeks back about billions in financing for a new venture. The guys desperate like I'm Donald duck.
Don't believe me? Check out the VC tweets... Sand hill pulled the checkbook the moment these guys might have been on the market.
What moon are y'all on.
He can secure billions with a text message.
Love ya anyway, cya this evening for the fuzzy meetup.
His opinion on the ideal path differs from Ilya's, but I'm guessing his goal remains the same. AGI is the most important thing to work on, and startups and corporations are just a means of getting there.
Supposedly his goal was the same as OpenAi --> AGI that benefits society instead of shareholders.
Seems like a hard mission to accomplish within Microsoft.
From there, maybe someone will come up with the revolutionary advance necessary to reach AGI. It may not necessarily be under his company, but he'll be the super successful AI guy and in a pretty strong position to influence things anyway.
Especially in the early days where the largest donor to OpenAI was Musk who was leading Tesla, a company way behind in AI capabilities, OpenAI looked like an obvious "Commoditize Your Complement" play.
For quite some time where they were mainly publishing research and they could hide behind "we are just getting started" that guise held up nicely, but when they struck gold with Chat(GPT), their was more and more misalignment between their actions and their publicly stated goal.
Once you lead at that level... It's max autonomy going forward. Source: Elon. Guy hates a board with power as much as Zuckerberg. Employee? Ha .. Out of the question.
See openai investment with technology transfers and sunset clauses. They just did a new dance.
They'll prod do something special for these guys.
They would never be employees. That's for non Sam Altman's and non Brockmans. Brockman is prob already a billionaire from openai shares. No employees here. Big boys.
Satya runs the biggest race track.
Altman trains pure breds trying to win the Kentucky derby repeatedly.
Totally diff games. Both big bosses. Not equivalent and never will be. Totally diff career tracks.
For many years, Microsoft Research had a reputation for giving researchers the most freedom. Probably even that's the reason why it hasn't been as successful as other bigcorp research labs.
"OpenAI, the non-profit who only has a for-profit subsidiary to get enough resources to fund its mission to develop AGI" is probably a winner, and gets to live instead of slowly die.
But what a play, MSFT the winner here.
They now owns the actual OpenAI
Edit: PM->am
Again. Very unexpected.
The whole timeline of events over the last two events still leaves me scratching my head though.
No, Ilya alone is like 75% of the brains and I'm fairly certain he's not going anywhere.
Meanwhile Google remains oblivious
IMO, to lead a research group you need some decent research skills, Sam is good at business
This is no research group, this is OpenAI 2.0, Sam/Greg will have enormous autonomy. It will be foolish to think Satya just recruited them to tangle them in MSFT bureaucracy
Satya wins, OpenAI is walking dead.
Large companies are primarily purchased for their moats
Sam wins
Ilya and the board continue to look like fools
He takes advantage of this situation and make OpenAI's assets in his control more than ever.
He is the genius, scary even.
Great pickup by MSFT. The exodus is only beginning and MSFT will not have to buy OpenAI for the billions in valuation it was getting. East win.
Anthropic is probably next in line.
Altman just rushed everyone’s hand by publishing it into the world at cost
My friends and family had an awful opinion of AI in general because it was the voice assistants were sold to them as the best example of AI. That changed with ChatGPT.
Google invented really useful AI but failed to deliver. OpenAI did so in record time. Now it's Google that's playing catching up with the technology they invented themselves, ironically.
But my comment applies more to Microsoft and Amazon, tbh.
And OAI delivered with enormous per-user cost that doesn’t scale - in an app that is a showcase and doesn’t really have latency requirements as people understand it’s a prototype.
And the vas majority of people play with CGPT, they don’t use it for anything useful. Incidental examples of friends and family of tech workers to the side.
Confused what this really means. So Microsoft still has access to OpenAI’s pre-AGI tech that Sam and Greg can leverage for their more product-focused visions.
More than that, it looks like Microsoft has become a major AI player (internal research) overnight up with the likes of Meta, Google, and OpenAI. Incredible.
Microsoft was already set to spent 27 billion usd on research for 2023. They dedicate huge standout double digit percentages of budget to research every year. Their in house AI research division was already huge.
They didn't become a major AI player overnight... They already were long ago.
OpenAI is small, in raw numbers of AI researchers, compared to the big players in the space. That's a major reason why it's so compelling that they have been able to consistently set the bar for state of the art.
They were a dream team... But small. Msft is adding AAA+ talent to their existing A+ deck. Also they won't have to rewrite the code base. Can hit the ground running.
Lastly, there is no evidence that openai has the greatly quoted and so hard to define 'agi'. That's Twitter hearsay and highly unlikely... If folkes can even agree what that is. By the overwhelming percentage of definitions... Even gpt-5 is unlikely to meet that bar. Highly speculative. Twitter is a cesspool of conspiracy theory... Don't believe everything you read.
MS now has both the accelerationists and the deccelerationists. They can keep accelerating themselves when pushing for regulatory capture through their deccelerationist branch to slow down any competition.