Ask HN: What are the terms of the $10B Microsoft / OpenAI deal?

43 points by MrThoughtful ↗ HN
I have not found any good description of the deal yet. Are the terms public?

54 comments

[ 2.9 ms ] story [ 83.1 ms ] thread
My mother asked me a few days ago how to invest in ChatGPT. I explained that it's probably the easiest expressed with some long dated ITM calls on MSFT

Anyone know if MSFT actually has equity in OpenAI or if its just a "mutual partnership" ?

Before thinking about investing in AI, it might be worth watching the all in podcast from last week (not the most recent one) https://www.youtube.com/watch?v=qQ544sWC8ZQ

I am not saying the hosts are right about everything but they made a point that the first to the market may not be the eventual winner and provided some examples of how revolutionary ideas such as a cell phone/smart phone were iterated several times before major leaders broke away.

The same can be seen with search and other technologies.

right, but was openai "first to market"? google and amazon have had AI offerings for a while now.
Was apple ever first to market anything?
It’s this simple, something that a surprising number of intelligent people fail to grasp. Apple was rarely the first to it, they just did it well. Google and other giants had the technology but they never made a super powerful chatbot available to everyone with the ingredients they had on hand. Now they’re scrambling.
The iPhone was the first to market for many things including an all-screen phone, multi-touch, full-web browser experience on a mobile device.
Palm would like to talk to you
No.

It was the first good device with those features.

But I skipped the first iPhone because at the time, my Palm appeared to be superior.

Both Amazon and Google did not have usable language models prior to ChatGPT.

Certainly ML was used, but not in that particular application.

this was reeeaaallly f good. Thanks
It's a delicate one.

I would bet on Alphabet for the long term, the original inventors of the paper, together with DeepMind.

But chatgpt like tech could have an effect on their advertising income.

The question would be if it affects their advertising or if it actually adds a service to sell to their pool.

If it adds a service to their pool, both Alphabet and Microsoft could be winners. But it takes a huge amount of profits/revenue for a service to have any affect on the stock price for both of them.

Additionally, AWS and Meta won't stand by. We've all seen how much Meta is prepared to fund Zucks vision and that's transformers now.

They even released one before ChatGPT.

https://www.technologyreview.com/2022/11/18/1063487/meta-lar...

Additionally, besides the impact, it's possible that it's not going to be as profitable as one would expect ( eg. because of price pressure due to competition)

There's been a brain drain from alphabet, why do you bet despite this?
Most of the people from the original Transformers left Google as far as I know, (including for OpenAI), so...

NicoJuicy: Have you ever tried Bard, the "ChatGPT-Killer" (English-only btw) ?

No, I didn't. I know the results aren't that good yet.

My guess was that they didn't apply RLHF much. There's going to be plenty of research coming in that area, I suppose.

I can’t bet on Alphabet with the current leadership. You either need a visionary CEO or the original founders to have the courage to gut your entire revenue stream (ads) in pursuit of a new tech.

Pichai is not that guy.

You're betting against a 1,4 T behemoth on "code red" that invented Tensor flow, has enough compute for anything with custom hardware versus a company that "got 10 billion in azure compute credits".

OpenAI may have a headstart. But it's still early for transformers.

I'd guess that the brain drain that went into the media is a bit exaggerated for a company with 190 k. Employees.

Every release of chatbots in the past of a multinational ( meta, Microsoft, alphabet) got ridiculed because it hallucinated.

For some reason, the media ignored it more with OpenAI.

Side-Notes: Apple is doing well and Cook isn't Jobs either. Ballmer still helped prepare everything for the next in line ( cloud, office for Mac, ... ). He was not popular and he made expensive mistakes ( Nokia). But Azure was released in 2008 during his leadership too.

As an end user, all I’ve seen from Google over the last few years is a search engine that can’t beat blogspam, a maps product that increasingly throws up bizarre directions, and a Google Docs product that has had no meaningful updates.

Google has been fumbling the ball even in things it used to lead in.

You're timing is a bit unfortunate, 4 days ago - https://www.theverge.com/2023/3/6/23627809/google-docs-drive...

Blogspam has always been a cat and mouse game

Maps works fine for me though, but that could be just my perception

---

Microsoft has been rolling out updates fast though, their delivery pipeline is well done

It doesn't change the point for Google docs & co, it's just a refresh of the UI not a breakthrough update with new features if I'm reading the news that you have posted
What features were missing from Google docs?
At the very least, at this point, I'd expect Google Docs to bake in some basic AI-supplemented creation features. Just autosuggest doc titles as a start.

Google Sheets could also use some AI magic to make data analysis a little easier.

I think Google has had the tech for a long time, they just see it as a stupid and wreckless product. Now their hand has been forced to respond, so we’re in a kind of dangerous arms race.
I'm currently talking to chatGPT and asking it how to fetch and filter records from a mongoDB database with a NextJS frontend. I described my record and use case. It told me that I should use a combination of aggregation pipelines and standard queries. Then it gave me the exact code matching my model.

This is far from a stupid and wreckless product.

I found your typo funny - wreck and reckless it is both, but wreckless, it will not be!
Exactly. We will see the next time that Google announces a new competitor [0] to ChatGPT or any other product but this time with DeepMind's direct involvement.

Many here have been sucked into the O̶p̶e̶n̶AI.com hype and have already forgotten about DeepMind. The media has completely exaggerated the 'death of Alphabet'.

[0] https://www.wsj.com/articles/google-ceo-sundar-pichai-says-s...

Bard has improved drastically over the last three weeks. Pichai is a good leader when it comes down to delivering on definitive products.
> I would bet on Alphabet for the long term, the original inventors of the paper, together with DeepMind.

> But chatgpt like tech could have an effect on their advertising income.

An AI response costs a lot more to serve than a search query.

Microsoft is integrating GPT into it's SaaS offerings (see copilot), where it adds new features (and is gaining traction for it's adds and search products). Meanwhile, for every GPT query served by Google, it costs them significantly more and can't be monetized right now with adds.

Microsoft may make a lot of money from chatgpt, but that doesn't mean they will topple google in search. Being able to democratize model training so that an enterprise could run gpt like tooling on their own internal data with no privacy concerns would be exceptionally valuable to a lot of enterprise and govt customers.

If they add 5% to bing, that is a fantastic side effect, but i dont know that Microsoft care about that as much as the pundits think they do. Enterprise gpt as a service and copilot powered software development would be world changing enough.

> If they add 5% to bing, that is a fantastic side effect, but i dont know that Microsoft care about that as much as the pundits think they do. Enterprise gpt as a service and copilot powered software development would be world changing enough.

GPT will shrink the search business. Why go browse adds from a search engine when you can just ask the assistant already built-in the app you're using?

> it's probably the easiest expressed with some long dated ITM calls on MSFT

Just buy equity. No need to sidecar long vol in the middle of a resolving banking crisis.

> if MSFT actually has equity in OpenAI or if its just a "mutual partnership"

Yes, Microsoft are taking a significant stake in OpenAI. But it’s not equity, because OpenAI isn’t a corporation. (Partnership interests. Economically similar.)

> No need to sidecar long vol in the middle of a resolving banking crisis.

as a former options trader, yup excellently put. love that this kind of "sophistication" (its simple in the grand scheme of things, but it passes) is here on HN. +1. stop using options for leverage if dont have a view on vol. way cheaper ways to borrow money.

what ways are cheaper to borrow money / leverage to be long MSFT?
not at all being condescending but if you dont know you may be better off not knowing, leverage is a mean master.

answer is margin accounts at any brokerage.

Wouldn't buying the MSFT stock be easier?
This is what I do. Also helps that I like the rest of the business. For me, this is mostly a win-win arrangement - I believe strongly in GitHub/Visual Studio/.NET/Azure/etc too.

If you dont believe in MSFT but want to invest in OpenAI, buying their stock is probably a terrible idea. You would be purchasing a gigantic liability (relative to your risk preferences) in hopes of gaining access to a miniscule component of the overall business.

From https://news.ycombinator.com/item?id=35367194

"I mean the conspiracy argument would be that the $10B isn't a normal investment. It's a special government project investment facilitated by OpenAI board member and former clandestine CIA operative and cybersecurity executive Will Hurd through his role on the board of trustees of In-Q-Tel the investment arm of the CIA. It's funneled through Microsoft instead of through Google in part because of Google's No-Military-AI pledge in 2018 demanded by its employees, after which Microsoft took over its military contracts including project Maven. The new special government project, the Sydney project, is the most urgent and ambitious since the project to develop nuclear weapons in the mid twentieth century.

Of course I don't necessarily believe any of that but it can be fun to think about."

> $10B isn't a normal investment

It’s a few hundred million of cash and billions of Azure credits.

> I don't necessarily believe any of that but it can be fun to think about

You’re quoting yourself disabusing yourself of an argument you’re quoting yourself making.

[flagged]
(comment deleted)
(comment deleted)
One interesting thing is that my homie https://news.ycombinator.com/user?id=ta08131942 made a throwaway account ten days ago, like 'throwaway 08/13/1942' which is the start date of the manhattan project and then he put only a single detailed comment https://news.ycombinator.com/item?id=35366484 relating OpenAI to the government/intelligence/military
First you quote yourself mentioning this CIA conspiracy elsewhere in this thread, then mention this green account, your "homie", with the same conspiracy.

I'm not saying you're right or wrong, just it is obvious you two are the same people, but I do not understand creating a throwaway account and then doxing yourself 10 days later.

I guess you hoped the throwaway account felt more credible.

> just it is obvious you two are the same people

Well I'm not the same as them but I see how it looks that way.

Nobody would've "just guessed" the meaning of that string of numbers in the username like you claim to, lol.
Well maybe no one will believe me but I'm a nerd and hacker (we are on hacker news) and I guessed it. I love guessing stuff like that. If you check my account, I'm named ftxbro because my first comment ever was to guess the meaning of a number related to ftx https://news.ycombinator.com/item?id=34473811

If you are so impressed by number guessing you can look at that one and be impressed if you think I am right about it too! Almost no one looked at that comment I think, because I posted it after the thread had scrolled off the front page. When I put that guess, I was so worried that it was such a smart number guess that people would accuse me of being some kind of ftx insider, so I tried to put a disclaimer at the top of that comment with my guess lol.

It's the trying hard to defend for me.
idk if you have ever been accused of doing something, and you didn't do it, and there is literally no way for you to prove your innocence of it. And the more you try to defend the more others pile on. It feels pretty bad.

It feels especially bad if you were even proud and happy that you did a clever thing (like guess the number meaning) and then this very thing that you were proud and happy of (the cleverness) is itself used as the rationalization that you are guilty of something bad (it was too clever to be true).

EDIT: maybe it could be possible @dang or some mod could back up that I am saying the truth, to stop the dogpiling and witchhunting me. They could backtrace the ip addresses. Some might say that I just used a VPN to make the throwaway, but if the one who actually made it didn't use one, the mods could see their real account and verify that it wasn't me without revealing who it really was. If that happened, maybe the hunters would say that my account is another of the throwaways of that user on a different IP address, but maybe that would be too much of a stretch of belief.