OpenAI and Microsoft Tension Boils, Amazon’s Job Automation, Zuck’s Spending Spree

0:00Microsoft and OpenAI have some major differences, and things are getting heated as OpenAI moves to its next stage. Amazon CEO Andy Jassy says AI will reduce his workforce. And Mark Zuckerberg keeps spending for top talent. That's coming up on a Big Technology Podcast Friday edition right after this. Welcome to Big Technology Podcast Friday edition, where we break down the news in our traditional cool-headed and nuanced format. We have so much to speak with you about this week because the AI news cycle just rolls on with bigger stories each week. We have a fascinating battle between Microsoft and OpenAI that's just really heating up and might eventually prevent OpenAI's for-profit conversion.

0:42We also have this very interesting memo from Andy Jassy about generative AI and what it might mean for Amazon's workforce. And of course, Mark Zuckerberg is spending that cash and doesn't seem like he's stopping. So we'll pick up that thread from last week. Joining us as always on Fridays to do this is Ranjan Roy of Margins. Ranjan, good to see you. Good to see you. AI news doesn't sleep. Another week. Sure doesn't. Never a dull moment. And leading this week is this really interesting conversation that we've had on the show and we will continue to have because it could influence the future of the AI industry.

1:18And that is what's going to happen with OpenAI and Microsoft. This is from the Wall Street Journal. OpenAI and Microsoft tensions are reaching a boiling point. Tensions between OpenAI and Microsoft over the future of their famed AI partnership are flaring up. OpenAI wants Microsoft's grip on its AI products and computing resources to loosen. And it also wants to secure the tech giant's blessing for its conversion into a for-profit company. Microsoft's approval of the conversion is key to OpenAI's ability to raise more money and go public. But the negotiations have been so difficult in recent weeks that OpenAI's executives have discussed what they view as a nuclear option accusing Microsoft of anti-competitive behavior during their partnership.

2:04That effort could involve seeking federal regulatory review of the terms of the contract for potential violations of antitrust laws, as well as a public campaign. Well, what do we call this? The not-so-gentle singularity? I mean, this escalated in a hurry. So a few points on this. I think we'll definitely get into the for-profit company conversion, which is incredibly complex. We've talked about for months and months and months. Like, let's not even forget, remember, Elon Musk sued over this conversion. they're getting hit from many different sides. And the fact that Microsoft is actually kind of acting as a hurdle is definitely a big issue.

2:48And that can prevent them from $20 billion of that sweet Masa Sun SoftBank money. So that's going to be a huge issue for OpenAI this year. But the Federal Regulatory Review of Anti-Competitive Behavior, I love this. I mean, And they are saying that the deal that they struck is anti-competitive. From what I understand, like, what else could it mean? They're saying the money we took from Microsoft and then that whole arrangement is bad for the industry and us, OpenAI, as well. But can you read it another way? It is bananas. I mean, it's one of those things where OpenAI needed a partner like Microsoft to be able to get to the place where it is today.

3:35because, of course, the most important ingredient in growing its products up until this point has been compute. And what did Microsoft have? It had that compute. What was OpenAI? It was a nonprofit. I mean, it's interesting because the ambitions were artificial general intelligence. So it wasn't like OpenAI wasn't thinking big. But it's almost like the company surprised itself with how successful it's been. It's almost like it wrote in that clause in the contract that both entities wrote in that clause in the contract that OpenAI and Microsoft's deal dissolves once OpenAI hits AGI. Not really thinking that that would ever be possible because now it's really coming into a place where they are going to, they're going to have to work through these issues.

4:16And we right now have OpenAI owing, I think 20 % of its profits to Microsoft. This is coming from the information. OpenAI wants Microsoft to have roughly a 33 % stake in the reshaped unit in exchange for foregoing its rights to future profits. If the companies don't change the 20 % cut OpenAI owes to Microsoft, Microsoft could be in line to get$35 billion in payments in 2030 when OpenAI has projected it will generate$174 billion in revenue. Now, OpenAI is on track to generate$10 billion in revenue this year. So that's one hell of a projection. It's a good projection. It's a good extrapolation.

4:59You got to extrapolate this. Kudos to the person who put together the spreadsheet and kudos to whichever investor believed the PowerPoint. Yeah. It really does come down to this. It's like OpenAI, again, they made the deal. They needed what Microsoft had. They agreed to like, all right, if we're doing really well, you're going to do really well. I don't see where OpenAI has the wiggle room to back out of this. Well, they certainly have the wiggle room when Waymo comes to New York City and we declare AGI, which we'll get into later. But no, no, I agree that from a pure financial standpoint or a pure contractual standpoint, they don't have a lot of wiggle room.

5:38They don't even have leverage, which is why I think that Hail Mary of the, whether it's through the FTC or that anti-competitive review, it feels like a Hail Mary. But it is because I don't think OpenAI has much standing in any other direction. But again, I think going back to, as you said, they're on track to make$10 billion. They're losing tons of money. Microsoft's share is in the profit, not revenue. And there's no sign that they're going to be turning a profit, even though they're projecting one by 2029. So I think in a certain way, a lot of this is moot anyways, other than like from a financial standpoint, because the money is not going to actually change hands based on any kind of like current trajectory of revenue and profit.

6:22So it really is about control. And I think like it's telling that they are they're competing incredibly directly. Like we're seeing it more and more. They're going after the same use cases, customers, audiences. I want to talk a little bit about the control part because that is, I think, crucial. So this is also a new detail that emerged in the reporting this week. The deal with Microsoft is that Microsoft gets the right to use OpenAI's IP through 2030, which, by the way, if you do the math, is like well beyond a lot of these lab leaders' predictions for when we get artificial general intelligence.

7:01I don't know. I'm not bought in that AGI is going to be here by 2030, but we'll luckily, hopefully we'll be doing the show. And if it does, I'll, you know, eat crow on the air and put together my army of millions of agents to shame me on Twitter or whatever it is. But I'm going to create an agentic workflow just to shame Alex for when AGI is declared. I really wouldn't be very different from all the bots on Twitter that came after me for my perplexity acquisition take. But anyway, I digress. But there's another thing that this control is important because it's not just OpenAI's IP. The areas that OpenAI is expanding to also will end up competing with Microsoft and also are of interest to Microsoft to control.

7:46So here's another detail from the journal story. OpenAI and Microsoft are at a standoff over the terms of the startup's$3 billion acquisition of the coding startup, Windsurf. Microsoft currently has access to all of OpenAI's IP, but it offers its own AI coding product, GitHub Copilot, that competes with OpenAI. And OpenAI doesn't want Microsoft to have access to Windsurf's intellectual property. I mean, if we think that coding is going to be one of the big applications of GenAI in the near term, this is really bad for OpenAI because effectively the deal that it struck again serves to put it in service of Microsoft and not expand its own offer.

8:30You know what? Maybe I am starting to feel this anti-competitive posturing a little bit because I guess it's true. They are competing incredibly directly. Right now, even Microsoft co-pilot across the entire 365 ecosystem competes very directly with ChatGPT Enterprise. Basically, this kind of always-on assistant and agent it's a direct competitor. And I'm sure when salespeople are going in and there's been more reporting that salespeople at Microsoft have been complaining that they charge$30 per user per month, ChatGPT Enterprise, it's competing directly, but they could be discounting it, meaning they're trying to undercut Microsoft's pricing, which is kind of hilarious because they're heavily invested and part owned by Microsoft.

9:21But I think overall, it does present a good amount of problems in terms of they're trying to do the same things, going after the same customers, probably in direct competition when going through any kind of enterprise sales cycle. So at some point, something has to give. And one could argue that Microsoft, by giving its compute because of its size as this tech giant, is acting in an anti-competitive bullying fashion. So maybe if Lena Kahn was still here, she might agree. But she's not. And let me point again to the fact that OpenAI signed the deal. This is your deal. You signed it. You wouldn't be here without it.

10:04Wait, I'm trying to think of an example where a company ever took a lot of money and then calls that funding or acquisition anti-competitive. I mean, it actually, how could it happen? You know what you call that? It's the Sam Altman special. Yeah, the Sam Altman special. Give me money, and then I'm going to go and say that you're bullying me for giving me that money and investment. But also, if you think about it, Microsoft holds OpenAI's financial future in its hands as well, because it's a capital-intensive field, shall we say, to put it lightly. AI, you need money to build servers, to grow.

10:43We have this Stargate thing that OpenAI is trying to build. Your funders are not going to really be into giving you all that money if Microsoft really has control over your future or control over a good chunk of your sizable profits. That's why I think SoftBank told OpenAI, you better convert to a for-profit or we have the right to withdraw our money. So I just want to ask you this, Ranjan. I mean, what does OpenAI have to stand on here? Again, funded by Microsoft, really built by Microsoft, competing with Microsoft. Am I crazy that OpenAI shouldn't be able to dictate the terms to Microsoft? Like, where in Microsoft's interest is it to change this deal?

11:26No, no. Okay. I agree. In any normal, like, flow of logic, they would have no right to dictate the terms, given, they not only took the deal, they probably pushed the deal themselves to work like this, because it basically, rather than kind of like traditional venture funding, where they would have had to be growing even faster, seeing more returns more immediately, it really was this sweetheart deal where it was just kind of a lot of compute, a lot of like future looking projections and possible financial returns. But I mean, it was a pretty sweet deal for them. And to try to say that it was problematic or back out of it now, again, is ridiculous, the Sam Altman special, but it is becoming more, I don't want to say existential, but it's becoming more problematic right now the way this is unfolding.

12:23Do you think Microsoft may just say to OpenAI, sorry, I know this is important to your future, but this is what we agreed upon. So even if you're worth a little less in the future because of your inability to spend the money you need to get bigger, we're just going to keep it as is. Thanks for playing. Yeah, I think they're going to have to. And again, there's been a lot of chatter around in terms of the success of co-pilot products and whether they're actually working or how happy people are using them versus ChatGPT Enterprise and OpenAI on the product side keeps swimming along. So at a certain point, maybe Microsoft does start to be a little bit anti-competitive and a little bit bullying.

13:09And they have the deal and the contract to allow them to, and OpenAI signed that deal. So I could definitely see it start to, yeah, explode. I think this, the more I'm thinking about this, this is going to explode in some direction this year. And again, like, I don't see how this is anti-competitive. I mean, it's simply a company trying to hold another company to the deal that it signed. And if you think about it, Microsoft has already been more than generous in allowing OpenAI to go and work with other companies for compute. It's allowed it to work with Oracle for Stargate. And there's recently news that it's going to work with Google to build the infrastructure it needs to run.

13:50I'm trying. I'm trying here to take the other, in the interest of nuanced conversation, trying to figure out what could be anti-competitive. But I mean, I don't think it is in this context. In another context where like Microsoft, again, Microsoft using its ecosystem to push its own products in a preferential way. In any other context, you can start to see how that's kind of like classic anti-competitive behavior, but not when it's the company that you funded and gave them all the good terms and sweetheart. I don't know. We will take projected profits down the road, even though you're losing billions of dollars.

14:36And that's all we're asking in exchange for giving you billions of dollars of compute and cash. I mean, yeah, I tried. I tried. I'm just imagining the meeting between Sam Altman and Satya Nadella where Sam goes, Satya, you helped OpenAI become what it is today. We owe you billions of dollars. But here is a different idea. How about we owe you less billions of dollars? I mean, that is exactly what this is. If anyone somehow could pull that off. I don't know. Sam Altman's pulled off a lot over the last few years. So who knows? I can just imagine Satya Nadella just like hating this guy right now.

15:17Yeah. But also, let's not forget, there is a political element of this that could kind of start to filter in. If they're serious, and this really starts to kind of move into the realm of like federal review, let's not forget that Sam Altman and the Stargate announcement got some good press for the administration. is, I'm not sure at this exact moment where he stands with the administration, but like, overall, he's had some good moments. Satya, I don't think Satya was at the inauguration, right? Correct. Yeah, yeah, exactly. So like, maybe starting to use that as a lever in this negotiation could be one way.

16:02Maybe. But I think we started the Trump administration with like, Zuck and Sundar and Bezos and Musk all behind Trump in the inauguration. But as it tends to happen in politics, he has bigger things to worry about right now. The tariff stuff, the trade war, Iran. I don't think he even remembers who Sam Altman is. I think that's fair. And now imagine, you mentioned OpenAI and Microsoft selling the same thing, whether it's ChatGPT Enterprise or Copilot, right? Very similar. Imagine you're OpenAI and you are going to go into federal court to argue or make some play to argue Microsoft is anti-competitive and you're doing this.

16:45This is from the information. A software company that had purchased OpenAI models through Microsoft over the past years was in talks to sign a new agreement to spend$8 million on the models over the next three years, according to a Microsoft salesperson involved in the talks. But that firm notified Microsoft that OpenAI had offered it a 20 % discount on the same models, reducing the cost over three years by$1 million. Microsoft salespeople asked the company's finance department if they could match the discount, but were rebuffed. As a result, the firm told Microsoft that it was choosing to buy the models through OpenAI instead.

17:24I mean, this is exactly what must be happening now and is going to only grow in scale, these exact kind of negotiations. And again, Microsoft is a publicly traded company that has to show certain projected financial metrics. OpenAI can light cash on fire right now, so they can be pretty aggressive and literally undercut at every step of the way using Microsoft's capital and cash and compute. Can you imagine being that salesperson? I mean, that's a pretty nice commission on an$8 million deal, I would imagine. Have that taken from underneath you, same model from OpenAI. This partnership is in trouble.

18:06So let's just look ahead. Ranjum, what do you think is going to happen with OpenAI and Microsoft? What are the scenarios you're thinking about? I somehow still think OpenAI comes out ahead, mainly because the product at the consumer level at a like, it's not going to just be crushed by Microsoft in that conventional way. anytime soon. I think from like a courtroom contractual legal standpoint, I think Microsoft certainly has both like standing and like the legal firepower to, you know, hold this over OpenAI. But I don't know. I don't see how what Microsoft can do to stop this at this moment. Like they can't be like, just shut it down.

18:56They can't be like, don't you have a separate sales force, but you are not allowed to undercut our pricing. So what could they do? I think that they just say, we're not making, we're not going to budge an inch and we're not going to let you get this money. We're not going to let you IPO unless we get some better terms. Stuff that like holds to the earlier agreements that we had. And maybe they end up owning 40 % of the company or something like that. and everybody just has to go about their business. But I don't think Microsoft made what some people called the tech bet of the century in OpenAI, only to lose it because OpenAI worked.

19:33Oh, that's a... Okay. I can agree with that. I think it's... Yeah. I think that's a good way to put it. I have an idea why Sam Altman may have thought this would work with Microsoft. Go on. Go on. That is that he's using ChatGPT too much because there was a study this week and written up in time that ChatGPT may be eroding critical thinking skills. Here's the story. Does ChatGPT harm critical thinking abilities? A new study from researchers at MIT Media Lab has returned some concerning results. the study divided 54 subjects, 18 to 39-year-olds, from the Boston area into three groups and asked them to write several SAT essays using OpenAI's ChatGPT, Google search engine, and nothing at all.

20:25Researchers used an EEG. I guess that's, yeah, it's a machine that records writers' brain activities across 32 regions and found that of the three groups, ChatGPT users had the lowest brain engagement and consistently underperformed at neural linguistic and behavioral levels over the course of several months chat chp t users got lazier with each subsequent essay and often resorted to copy and paste by the end of the study there were some people on x that were being like look at the way that these people designed this study they hate ai they were trying to set traps for people to fall into it studies garbage i i my reaction here is totally not surprising i mean are we even going to debate that this is happening?

21:09Well, one, if Microsoft investing in open AI was the tech bet of the century, I think you just nailed the segue of the century with that transition right there. I'm trying, man. Yeah. But okay. So this study, I was thinking about a lot because how or if generative AI will totally alter our brains is something I have wondered about. I always think of maps. And I literally grew up in the Boston area. And when I was first driving and before any kind of Google Maps, I literally knew how to get all over the city, all around my town, just by memory. I've been in New York for many years now and have a car now.

21:52And I cannot get to JFK or LaGuardia driving without Google Maps. I don't know directions from driving, but I think that's okay. I mean, if, if, if the entire like internet collapsed, I might be in a little trouble there, but the way I learned kind of location is completely changed, but I can now do it at a scale that it was unimaginable before I can go into a new locate locality and like, just move around and navigate. So it's a different way of using the brain. I think this is again, And writing an SAT essay was already, it's such a specific kind of mindless thing, I think. Like, in reality, the act of doing it is not actual, like, intelligence in my mind.

22:38And I say this as someone who went through the process. It's like, so to use that really specific thing, I do think is kind of misleading. because taking a very mindless task, and then of course the person using ChatGPT, it's not going to trigger the level of brain activity because it was already a mindless task anyways, and it's perfect for generative AI. So for me, I think our brain, the way we learn is going to change. I think that's good. We have to adapt the way like learning is done for generative AI, but I don't think people are going to get dumber. Maybe that's optimistic. But I think this is one of the points is a lot of the stuff, let's say you're thinking about using this for work, a lot of the stuff that you're doing at work will get your brain sort of deep into the material and help you think critically about a broader business activity.

23:32I'm just saying this for the sake of argument, because I do believe LLMs could be really useful in the workplace. I think less of that in education. If people are going home, not just writing their SATs, but writing essays for class, they're not going to retain anything. Because actually the pain that you endure in that writing is where you retain. But I actually like, just to sort of move from education to work, I wonder if this will make better workplaces if people are using more AI. And whether it will make more competitive companies. if people are using more AI because of the impact that it could have on the brain.

24:06Hold on, hold on. Just quickly on the education thing, I think it really is going to demand rethinking how the classroom works. And I think that's good. Writing essays, as you said, it was the enduring the pain of writing is where you learned in the past. That potentially goes away, but like presenting the essay, having to be able to defend it, like more Socratic learning, like, do you think this is going to be net bad for education? Or do you think we just have to update the way we teach? I think we've had this discussion before. I am, I'm hopeful that we can find a way to teach better. But I think if you marry this style of learning with our current education system, you end up getting a disaster.

24:53I won't argue with that, I guess. The flip side, then you also get into, to me, the most optimistic part of this is the democratization of learning and access to knowledge, learning tools. Now the fact that anyone in the world can, wherever you are just with an internet connection, can actually create entire curriculum tailored to you like i think it's going to really be good for learning at a large scale i think in isolated cases a couple of uh 18 to 39 year olds in boston uh writing an sat essay yes that format is not going to be good and it's going to be change like people will just cheat their way out of it and not actually develop their brain but i i kind of look at this like what youtube did for learning this is going to be at like a thousand X scale.

25:48Yeah. So Dwarkesh, after we turned the cameras off for our conversation on Wednesday, it was like, yeah, I just basically drop stuff like papers I'm trying to learn and have Chad GPT use the Socratic method and help teach me that way. So I think, yeah, hopefully we will develop new modes of learning. But I also think we're living at a time where we've known we need new modes of learning for a long ass time. And we're still doing the rote memorization and spit back stuff that we've been doing forever. Now, I'm sure they're going to be education professionals who are listening to this, who are going to know much more than me and say that there are areas where this is improving.

26:23I don't doubt that. But by and large, I think my generalization about the education system really across the globe holds true. So let's talk about work because we have limited time today and we should definitely get it to this Andy Jassy message to Amazon employees about AI in the workplace. He says, we have strong convictions that AI agents will change how we all work and live. Agents let you tell them what you want, often a natural language, and do things like scour the web and various data sources, summarize results, engage in deep research, write code, find anomalies, highlight interesting insights, translate language and code into other variants, and automate a lot of tasks that consume our time.

27:05There will be billions of these agents across every company and in every imaginable field. There will also be agents that routinely do things for you outside of work, from shopping to travel to daily chores and tasks. Many of these agents have yet to be built, but make no mistake, they're coming and coming fast. Can I just pause here before we get to the fact that he's going to sort of now, you know, soon pledge to decrease Amazon's workforce to be like, come on. I mean, we've heard this promise for how many years now, and it's not there. And what makes him so confident? And of course, he's closer to the technology than I am.

27:43But what does he make it? What makes him so confident to think that this is going to happen? I would I would push back. I think the vision that's being presented is going to happen in terms of like timescale when that happens. I think like again, the same way learning has to change. Work is definitely going to change. A lot of the repetitive things that we've all done over and over again start to be automated in some way. I mean, even thinking about like this podcast, being able to upscale our audio, being able to edit very quickly, getting video clips from Riverside, like this is something that would not have been possible a few years ago, even like two years ago at the scale that we're able, and quality we're able to do for our listeners.

28:32Little plug there. But, you know, like overall, I think it's definitely going to happen. I think I'll agree that it's in every CEO's interest to say this, especially if you're a publicly traded company. It just is kind of easy to say we're going to be getting efficiency gains and stuff. But I do think from a messaging standpoint, it's important because this will force the workforce to actually start learning how to use these tools. And I think that's whoever, whichever companies like figure that out. And I've thought about this a lot. Like, are the winners of the next decade going to be AI first, AI native companies?