58 comments

[ 3.9 ms ] story [ 118 ms ] thread
I second that
Why? neither MSFT or OAI are monopolies in the AI space.
You don't need to be a monopoly to be engaging in anticompetitive practices. The ideal time for an antitrust action is before the trust has successfully eliminated all competing businesses.

I'm not even saying that OpenAI/MSFT are being anticompetitive, but I certainly see no problem with the FTC keeping an eye on them, especially given Microsoft's track record.

On what planet are they ever going to eliminate other AI? Neither Amazon, nor Google are at any risk of going bankrupt, and I can say with 100% certainty both will continue to build their own AI while solvent. That’s to say nothing of competitors coming out of Europe and Asia.
You could have said the same thing about Windows Phone—Microsoft had every reason to keep trying to win this new market, and they certainly weren't going bankrupt, but eventually they had to concede it wasn't going to happen.
Because they are rapidly looking to build a monopoly. We are starting to see capital just take whatever it wants in all spaces. Early law didn’t anticipate this to this degree.
Early law didn't anticipate anything. Modern law can't keep up either.
I am so tired of reading this.

You do not need to be a monopoly to engage in anticompetitive practices that are illegal.

Monopolization begins to matter more with things like mergers. It is not necessary for other illegal anticompetitive practices, such as price fixing, for example.

Collusion is illegal - full stop.

(comment deleted)
Oh shit msft in its lifetime has never engaged in anti-competetitve practices. Same sentiment with the other reply. Tired of reading this.
> Why? neither MSFT or OAI are monopolies in the AI space.

Microsoft is a monopoly in other spaces, and could be leveraging those monopolies and their partnership with OpenAI to monopolize parts of the AI space, which would be illegal even without having succeeded.

You need a microscope because otherwise you can't even tell them apart?
> Microsoft's investment does not amount to control of the company.

We can't know this without seeing the agreement.

> ""While details of our agreement remain confidential..."

I wonder if this was triggered by Microsoft having a big part in forcing Sam back into OpenAI. Certainly, housing the infrastructure and compute of OpenAI gives them some effective power.
But that's not how it played out though. MS offered them a new home, so from the outside at least, they had leverage that any other player could have had.
MS very quickly had an agreement with the ex-CEO to form OpenAI 2.0 within Microsoft. Maybe another player could have done that, but it seems more likely that Sam Altman and Microsoft have a Special Relationship.
Not based on what Satya Nadella said in an interview:

> "We have all the rights and all the capability. I mean, look, right, if tomorrow, if OpenAI disappeared, I don't want any customer of ours to be worried about it, quite honestly, because we have all of the rights to continue the innovation, not just to serve the products."[1]

If you're an OpenAI employee looking to continue working on the product, would you rather go to SalesForce and start from scratch or Microsoft and continue (roughly) where you left off?

[1] https://stackdiary.com/satya-nadella-we-have-all-the-rights-...

What we do know is that Microsoft has exclusive permission to all OpenAI products, until AGI is reached.

Do we need to know the rest?

There are two parallel thoughts going in my mind.

1. smoke and mirrors

2. hatchet job as a favor to Google for old times sake (healthcare.gov resurrection etc. relations).

Either way, this will not go anywhere, because Microsoft is run by competent people and FTC is not.

edit: upon further reading, retribution for losing Microsoft-Activision merger could be a possible thing too. But my final observation still stands.

It is a shame that the FTC was not this proactive when Google and Facebook were buying up every company in sight that might have turned into competition for them. Does this represent progress for the FTC? Or does Microsoft get extra scrutiny because of its history with anticompetitive practices?
Since Carter/Reagan, FTC has historically not challenged these mergers because FTC and other legal agencies got swallowed up by Robert Bork who basically convinced them that monopolies are actually good. So FTC and other agencies basically listened to paid off economists who were like "This merger will be fine for consumers"

https://www.promarket.org/2019/09/05/how-robert-bork-fathere...

Lina Khan has changed FTC which is why you see wave of this action but obviously anything law, it's very slow going.

> Does this represent progress for the FTC?

Yes. The FTC has been largely dormant on antitrust issues for 40 years. That changed for the first time under the current administration, but it's going to take a huge lift and a lot of time to rebuild the FTC into an agency that can once again do antitrust stuff. I'm hoping they get that time to rebuild. As you noted, we desperately need to reign in these huge companies & mergers.

It is basically Lina Khan doing all this.

However, it is pretty useless as the courts have slapped down her attempts pretty regularly.

It makes great headlines, but I don't think anything will come of it.

It’s a bit hard to figure out what she thinks she’s going to accomplish here. This isn’t Call of Duty. MS/OpenAI has national security implications. There’s a reason SpaceX gets what it wants in the end with the FAA and EPA.
Matt Levine on Bloomberg pointed out that originally the OpenAI structure chart had MS as a minority owner:

https://web.archive.org/web/20231109140650/https://openai.co...

But now call it a minority economic interest: https://openai.com/our-structure

There are many financial arrangements that are not traditional ‘ownership shares of a company’ that the SEC treats as ownership in some ways for reporting, eg Total Return Swaps. As the agreement isn’t public it’s right that it should be investigated.

Are there any circumstances under which a revenue-share agreement with employees (e.g. a commitment to split x% of revenue or x% of profit with employees as a bonus) gets considered any form of ownership, as opposed to simply compensation?
FTC is different than the SEC though.
(comment deleted)
Making an offer to hire all the employees because you don't like what the board is doing does tend to attract regulatory attention.
I think Microsoft offered them the jobs because a large number signed a petition suggesting they were unhappy. Salesforce made a similar offer so I guess by that logic, Salesforce should receive similar scrutiny

https://m.economictimes.com/tech/technology/salesforce-ceo-o...

Salesforce didn't also announce that they'd hired the ex-CEO, nor did Salesforce have a $10 billion investment in the company prior to that. Salesforce also didn't get a board observer after the ex-CEO was hired back.

It's the combination of factors that are raising eyebrows, not any one of them.

There's also this from Satya Nadella:

> "We have all the rights and all the capability. I mean, look, right, if tomorrow, if OpenAI disappeared, I don't want any customer of ours to be worried about it, quite honestly, because we have all of the rights to continue the innovation, not just to serve the products."[1]

Microsoft has access to OpenAI IP that SalesForce or others do not.

[1] https://stackdiary.com/satya-nadella-we-have-all-the-rights-...

So once the profit cap is hit, won't all employees just move to Microsoft for further stock compensation with a chance of appreciation, vs OpenAI's dead end stock? And likewise what stops the non-profit from just creating a second for-profit with all the IP too: "No Profit Cap OpenAI 2, inc."
I guess it depends on why the employees joined OpenAI? If I want to maximize my comp I don't work for a nonprofit in the first place.
I believe in workers right to organize and work wherever they want. I don't believe in non compete although I don't think they were subject to one.

What is the issue?

Refusing to hire the workers of partners would be a worse problem. But their offer sounded a lot to me like they were presuming access to salary information, etc. In a proper system neither microsoft nor salesforce would know if a candidate was bluffing about their pay, internal evaluations, etc, beyond customary things like references. Being guaranteed a job could mean being unable to play biggest a-hole with HR without an information disadvantage. In the worst case that data could have transferred before anyone even attempted to utilize the offer and as it turned out never needed the offer.

Given past crimes by big tech they should be examined extremely carefully.

Was Salesforce actually set up to absorb these engineers, do they have the ability to take advantage of their potential availability?

Regardless of whether or not there was any sort of plan beforehand (I bet there almost certainly wasn’t), the fact that OpenAI’s engineers have shown the ability to make the “we’ll switch to MS threat,” and have it be seen seriously enough to change the company, changes facts on the ground in a way that regulators need to at least think about.

As set up as any other giant tech company
They could have absorbed them financially, but I doubt their ability to effectively utilize and keep them. Management issues are causing a ton of churn now
I honestly don't get what the outrage is. Should engineers not be allowed to organize and threaten to leave as a group? Should they be disallowed from joining some company based on some relationship their parent has with them?

This is all very weird and contrived outrage

I don’t see the outrage. It seems like a signal that the companies are closer than previously described. It is worth checking out, that’s all.
help, there's a new company that just made ~500 employees into millionaires. That raises inequality, not everyone can be a millionaire, fear not, the government is here. no more inequality.
And how much of this would be going on if OpenAI was some thrift store company? Oh right... none.
Some questions:

Could OpenAI exist without Microsoft?

Which other tech company has meaningful ownership / economic interest of OpenAI?

Is OpenAI collaborating with any of Microsoft's competitors (e.g. Google, Apple, Facebook etc.) in a meaningful way?

What practical difference (e.g. price, privacy) would people experience if OpenAI would be a 100% Microsoft owned subsidiary vs. the current situation?

Openai did exist without Microsoft and there's no need for cloud when all you really need is compute.
Could a 2nd OpenAI (with the same scale as OpenAI) exist without Microsoft (or another cloud provider) giving preferential access to resources?
Is Microsoft the only company with compute?
I imagine not, hence 'or another cloud provider' in the comment you're replying to.
What about training openais models requires 9 9s distributed availability? Cloud is not the only form of supercomputing in fact it's probably the worst since you don't need the cloud aspect you only need the compute. Once trained you have no need for a cluster and your dependence on cloud is not an essential part of the business. Turning it around.. what am I missing? What cloud features are essential to openais?
> Cloud is not the only form of supercomputing

Yes. Microsoft rents compute to others so I simply mentioned other providers rent compute to others too.

Microsoft is a scary company. I try to avoid all business with them if possible.
Non-profits are required to take meeting minutes by law. The FTC will have information about OpenAI’s board.