1. Podcasts
  2. The Information's TITV
  3. How Microsoft’s New Office Chief Fixed Copilot, OpenAI’s Plan to Revive Sora
The Information's TITV

12 Oct 2026 · 44m · 17 chapters

Episode artwork

How Microsoft’s New Office Chief Fixed Copilot, OpenAI’s Plan to Revive Sora

In short

Microsoft’s new Maya AI chips to reduce NVIDIA dependence and cut Copilot costs; a profile of Charles LaMonna, who now oversees Copilot and Office; an unnamed startup aiming to revive OpenAI’s Sora for Hollywood via licensing; Silicon Valley “spare-room” server rooms running open-weight models; Biohub/Chan Zuckerberg’s $1.8B government-backed AI push toward curing diseases.

Guests and backgrounds

Aaron Holmes (The Information Microsoft reporter; previously covered Maya chip plans). Stephanie Palazzolo (The Information AI Agenda newsletter; co-reported the Sora-revival startup story with Corey Weinberg). Aaron Tilley (The Information reporter; wrote about home server rooms). Amy Doxler-Marcus (The Information health and science reporter; covered Biohub’s initiative).

Key claims

Maya 300 is “more efficient” than NVIDIA for Copilot; Microsoft wants “multiple gigawatts” of chips and may not yet offer Azure customers Maya rentals. LaMonna impressed Satya Nadella by pitching Anthropic models after Copilot was buggy and weak in Excel/PowerPoint; he pushes experimentation and a “no-jerks” culture. The Sora startup (Jeffrey Katzenberg, Bill Peebles, Sujay Jazwa) is in talks to license/tweak Sora tech for film studios. Home builders run open-weight models (e.g., Alibaba’s “Qwen”) locally to save money and for privacy. Biohub is the biomedical arm of CZI, partnering with government and big tech to AI-standardize data for drug discovery.

Notable examples

Copilot Excel/PowerPoint task failures; LaMonna’s Autopilot work inspired by “OpenClaw”; home setups in attics/garages; Qwen and other open-weight Chinese models; Biohub’s Virtual Biology Initiative; partners including Meta, Google DeepMind, and Isomorphic Labs.

Written by AI. May contain mistakes.

Chapters

0:40

Episode Overview

Overview of today's topics, including Microsoft's AI chips and OpenAI's Sora.

“Today on the show, we're going to dive into Microsoft's new Maya AI chips and an exclusive profile on the man overseeing Microsoft's co-pilot and his journey to Satya Nadella's inner circle.”
1:26

Microsoft's New Maya AI Chips

Discussion about the new Maya AI chips and their implications for Microsoft.

“Our Microsoft reporter, Aaron Holmes, previously reported on its plans to release the chip in part to reduce its reliance on NVIDIA.”
3:18

Profile of Charles LaManna

In-depth analysis of Charles LaManna's rise within Microsoft and his impact on Copilot.

“Any new details that we didn't know from your reporting?”
6:25

Innovation and Competition at Microsoft

Insights into LaManna's leadership style and Microsoft's competitive landscape.

“But this is something that Lamana and his team have sort of made a habit of doing.”
7:30

The Future of Leadership at Microsoft

Exploration of the potential future CEOs at Microsoft and internal dynamics.

“So definitely hearing that there's kind of this expectation that his teams will see what else is out on the market and see what could be applied to Microsoft.”
11:14

Closing Thoughts on Microsoft

Wrap-up discussion on the future of Microsoft and its leadership.

“It's not every day that people actually acknowledge that they're interested in doing it.”
13:14

Reviving OpenAI's Sora Technology

Overview of a new AI video startup's plan to leverage Sora technology.

“Open AI's popular video app, Sora, could be coming back, but not in the way that you might think.”
14:00

Sora's Revival and Hollywood's AI Video Landscape

Explore how a stealthy startup aims to revive OpenAI's Sora for Hollywood.

“Sujay Jazwa, who is the former CFO of Dropbox.”
18:05

OpenAI's Strategy Shift: Spreading Bets Again?

Discuss the implications of OpenAI reviving its diverse project strategy.

“Yeah, mostly because Hollywood, other than, well, other than Ben Affleck, I mean, I don't think people like AI that much.”
20:58

Thanking Stephanie Palazzolo

A brief thank you to guest Stephanie Palazzolo for her insights.

“I think all these side projects are going away again.”
Show all 17 chapters
21:44

Building AI Server Rooms at Home

Investigating the rise of personal AI server rooms among tech enthusiasts.

“People across Silicon Valley are using spare rooms in their homes to build server rooms as they try to optimize their own AI operations.”
27:38

Apple's Upcoming Smart Home Event

Previewing Apple's new smart home system and AI integration.

“Hey, Aaron, I want to ask you about this Apple event that is scheduled for this week.”
28:00

Apple's Home Hub and AI Integration

Discussion on Apple's upcoming home hub system and AI's role in it.

“But Apple will be coming out with a sort of home hub system and accessories for a sort of bigger smart home plan.”
30:28

Introducing Amy Doxler-Marcus

Introduction of Amy Doxler-Marcus to discuss Biohub and its initiatives.

“But more frequency is definitely in their interest.”
30:40

Biohub's AI Initiative and Government Partnership

An overview of Biohub's efforts in AI to cure diseases, involving major partnerships.

“Biohub, the main philanthropic effort of Mark Zuckerberg and Priscilla Chan, announced a new$1.8 billion partnership with the government to use AI to work towards their mission of curing all diseases.”
36:59

Cultural Shift in Biomedical Research

Exploring the shift from academia to big tech in biomedical research and its implications.

“And I do believe there will be more to come.”
42:02

The Significance of Government and Tech Partnerships

Explore the implications of government partnerships with tech companies in driving innovation and discovery.

“You know, some big moment where there's resources, a huge amount of resources and new thinking.”
A man walking through an airport, asking his phone a question

Ask about this episode

Get the gist, explore an idea or ask about something said.

Claude

On The Information's TITV, what are the key points of this episode?

Searching your listening

Transcript

Automatic transcript. May contain errors.

0:12Stephanie Palazzolo: Welcome everyone to the Informations TITV. My name is Akash Pasricha. It is Monday, October 12th. A reminder that voting for the Signal Awards closes on Thursday and we are nominated in the most innovative video podcast category. We would love for you to vote for us. We're going to put the link in the show notes or you can vote at vote.signalaward.com. Appreciate your support. Today on the show, we're going to dive into Microsoft's new Maya AI chips and an exclusive profile on the man overseeing Microsoft's co-pilot and his journey to Satya Nadella's inner circle. We're also going to talk about the startup that could revive some of OpenAI's Sora technology.

0:59Stephanie Palazzolo: We'll then get into the latest craze that is Silicon Valley building its own server rooms in spare rooms across their homes. And to close out the show, we're going to break down the Chan Zuckerberg Initiative's massive$1.8 billion investment with the government to use AI to work towards curing all diseases. It's going to be a great show, so let's get right on into it. Microsoft has unveiled its new Maya AI chips. Our Microsoft reporter, Aaron Holmes, previously reported on its plans to release the chip in part to reduce its reliance on NVIDIA. I want to bring on Aaron, who joins me now for some more analysis.

1:42Stephanie Palazzolo: Aaron, welcome back to the show. It's great to have you here. Hello, happy to be here. So nice when your scoop gets confirmed by the company. And here we have it, the new Maya chips. What are they and why did Microsoft release them? Yeah, so like I reported back in August, the new Maya 300 chips, according to Microsoft, are more efficient than NVIDIA GPUs at running their co-pilot software. and Microsoft now wants to use these chips to rely less on NVIDIA for their own AI applications like Copilot. The company says that they want to scale up to multiple gigawatts of chips starting next year, which would be millions of chips across their data centers.

2:25And if they can successfully do that, they can potentially bring down the cost of running Copilot and help improve their margins and also help save more of the NVIDIA chips for customers that want to rent those on Azure.

2:38Stephanie Palazzolo: So today, they basically announced that they are now selling them that they're available, right, for purchase? So Microsoft primarily actually kind of framed this announcement around using them as like a first party chip. So the idea seems to be that they will use Maya to run their own Copilot applications, which, you know, support customers using Copilot. They didn't really say much about actually having customers rent Maya, which is interesting because, you know, we heard over the summer that they were trying to convince Anthropic to potentially sign on as a customer of Maya. But yeah, so far, no word of actual Azure customers renting Maya chips for their own workloads.

3:17Okay.

3:17Stephanie Palazzolo: And so was there anything new about the announcement today? Any new details that we didn't know from your reporting? I mean, primarily just the scale of their ambitions here. You know, they said that they want multiple gigawatts of chips. We reported over the summer that they were in talks to secure several hundred thousand chips for production from TSMC starting, you know, next year. And I think it's still kind of a question mark exactly how much of that manufacturing capacity they can secure, which will probably determine, you know, how successful they are in getting these chips set up more broadly.

3:52Right.

3:54Stephanie Palazzolo: Aaron, I want to pivot to talk about a story that you published. It was a profile of an executive overseeing another part of Microsoft's operations, Charles LaMonna. He is executive vice president overseeing co-pilot. Why did you decide to profile LaMonna? You know, I originally decided to do this profile a few weeks ago just because, you know, as I'm always asking people who they think might potentially one day be in the running to be next CEO. His name was coming up a lot, especially among, you know, senior Microsofties who have a lot of experience at the company. And then, you know, actually, after I started working on the profile just a couple of weeks ago, he was given even more oversight of its software.

4:41He now runs the Office Group as of earlier this month. So this profile is really kind of looking at how he has risen through the company. He started as an intern almost 20 years ago and now is top brass reporting directly to Saati Nadella with this kind of huge mandate of both Office and then also like the, you know, interns of Copilot, including which models power that AI software.

5:05Stephanie Palazzolo: And so what is it that makes him so impressive to Nadella? Did you get a sense as to why he has grown so quickly in the organization? Yeah, so I heard that his influence really started to grow around a year ago when Charles LaManna made this internal pitch that Copilot was underperforming customers' expectations. At the time, they were fielding complaints that Copilot was buggy and that it wasn't very good at tasks in Excel or in PowerPoint. And at the time, Lamanna made this pitch to go out and start paying for Anthropik models in Copilot, which was kind of seen as heresy since, you know, Microsoft's deal with OpenAI already gave it free access to use OpenAI's technology.

5:49You know, why would it go out and pay for these other models? But he impressed Satya Nadella with demos showing that at the time, Anthropik's models were just way better at automating tasks in Copilot. And then, you know, just over six months later, there was a huge reorg and Nadella actually gave LaMana oversight of all of Copilot's backend engineering groups. And, you know, since then, he's taken over several other teams. And at this point, you know, most of that of the engineering groups building the inside of Copilot and, you know, the models, the power, how it work are now reporting directly to LaMana.

6:25Stephanie Palazzolo: And I want to zoom in here on this moment here where he convinced Nadella to use Anthropics models, because one of the things you pointed out in your story is that Microsoft doesn't actually have a culture of using outside tools often. But this is something that Lamana and his team have sort of made a habit of doing. Yeah, yeah. And so, you know, there was that decision to start using Anthropic a year ago. And I think even more recently, Lamana has been encouraging people to experiment with other tools and see, you know, how they could inspire innovation at Microsoft. So, for example, he has led work on Autopilot, which is this feature in Copilot that was kind of based on this viral popularity of OpenClaw last year.

7:11And, you know, more recently, I've heard he's encouraging engineers to try out things like Meta's Muse and the Instinct AI agent and think about how they could build similar features that would work in an enterprise context, which is a very different market, but obviously there's a lot of potential overlap there. So definitely hearing that there's kind of this expectation that his teams will see what else is out on the market and see what could be applied to Microsoft.

7:38Stephanie Palazzolo: Now, you had the opportunity to speak with him as well. So what were the big questions that you went into your interview with him with? Yeah, I mean, I primarily was interested in finding out how he is thinking about, you know, this delicate balance that Microsoft has to carry out, where it's partnering with OpenAI and Anthropic to power Copilot, but it's also competing with OpenAI and Anthropic. And, you know, that's coming at a time when both of those firms are in pricing war, where they're trying to essentially convince enterprises to use their own AI products and do so at like a better cost value equation than each other.

8:20And, you know, that was actually something that he said is a top priority of his right now. He wants Microsoft to be the Costco of AI, meaning like everyone from a broke college student to a millionaire would use Copilot because it's the best value that they can get out of AI without spending too much on tokens. and, you know, interesting to see how he will continue to thread that needle.

8:42Stephanie Palazzolo: That's kind of a staggering ambition. I mean, the Costco of AI, because, you know, it, I mean, it's trying, I guess. I mean, I'm sort of trying to liken it to what Microsoft became prior to AI, if it was the Costco of professional, you know, we had 365, we had Azure. So you could basically frame it also as not just trying to become the Costco AI, but remain the Costco of software because it did a good job. And now maybe it feels like that's slipping away a bit. Yeah, and that's a good point. I mean, I think that like the AI era is honestly making Microsoft face deeper competition than it has in decades.

9:25You know, for a while, it kind of had this productivity software market on lock with Windows and Office. and now suddenly, you know, in the AI assistant era or AI agent era, there are all of these other competitors essentially trying to do the same thing that Microsoft is doing with Copilot. And so I think it actually has to, you know, more than ever before, maybe think about pricing this in a way that is appealing and that can convince enterprises to spend more on AI and on traditional software year over year.

9:56Stephanie Palazzolo: And so in that conversation with LaMana, you talked about sort of the cost benefit being the major question. Was there anything he said that surprised you? And moreover, you know, what did you walk out of that conversation with in terms of further questions or further challenges that you think he's going to have to answer moving forward? Yeah, I mean, I think like he spoke at length about trying to do things differently than how has maybe been the norm at Microsoft. And that's also something that I heard from his staff is that, you know, his leadership felt like a break from the past. He, you know, talked about having a kind of no-jerk's policy, which I think is maybe a reference to how, you know, maybe a decade ago, Microsoft was notorious for having this kind of cutthroat stack-ranked culture where every org was competitive with one another inside the company.

10:46And I think that by, you know, unifying all of the co-pilot engineering groups, he's trying to somewhat put an end to that. I also, you know, had to ask him, like, you have risen from intern to executive vice president. Would you ever want to be a CEO? And he said, you know, for now, he is focused on just making Microsoft's software better and continuing to build products. But he said he isn't ruling out the potential of leading a company one day in the future. Oh, okay.

11:15Stephanie Palazzolo: Well, that's, that's fun. It's not every day that people actually acknowledge that they're interested in doing it. Normally, they say something like, wow, Satya's doing such a great job. And, you know, he's the captain, but he actually expressed interest in the role. That's great. Yeah. And I mean, the last time that Charles Delamana was a CEO was when he founded a startup about 15, 16 years ago, which was then acquired by Microsoft a year later. So, you know, he technically has been a chief executive in the past and it would be interesting to see if he ever wants to try to step into that role again.

11:47Stephanie Palazzolo: So let me ask you one one more question, Aaron, because, you know, you did a profile of Asha Sharma, the head of Xbox. Now you got Charles under your belt. There's a couple other key EVP characters that I imagine maybe we might see some more reporting from you in that camp. Is it, does it feel like a bit of a competition internally between these key characters for who could be the CEO? Yeah, I mean, it definitely kind of looks like that, at least from the outside. I think it's been pretty clear that Nadella has elevated this kind of new generation of leadership over the last year or so and, you know, has been appointing people who have been at the company for decades, as well as people like Asha Sharma, who just joined in the last couple of years and filling his senior ranks with these, you know, younger leaders after a year of pretty heavy executive turnover.

12:44So I think it will be interesting to see who of the young EVPs, as they're called sometimes, will stay in the running and how that will shake out in the year ahead.

12:53Stephanie Palazzolo: And then inevitably, once one of them gets a CEO job, the rest of them are just going to go leave and be CEOs of other big companies, I imagine. Although I don't know. That's just typically how it goes. Aaron, I want to thank you for coming on. That is Aaron Holmes, our Microsoft reporter here at The Information. Open AI's popular video app, Sora, could be coming back, but not in the way that you might think. A new buzzy startup is in conversations with Open AI to leverage some of the core Sora technology. I want to bring on Stephanie Palazzolo, who reported that story with Corey Weinberg. Stephanie, welcome back to the show.

13:33Stephanie Palazzolo: It's great to have you here. Hey, it's great to be back. Okay, so let's talk about the startup that is at the center of this story. Who is this company? Yeah, so the company doesn't yet have a name, or we haven't reported the name yet, but Corey and I first wrote about this company about a month back. It's a new AI video startup that's founded by former Hollywood mogul Jeffrey Katzenberg, Bill Peoples, who previously led the SOAR team at OpenAI, which was their very well-known video model, and Sujay Jazwa, who is the former CFO of Dropbox. And essentially, they are looking to develop AI video technology and sell it to Hollywood film studios so that those studios can then use that tech to make movies.

14:25Stephanie Palazzolo: So it feels like this is almost exactly what Sora was supposed to be, minus the whole social networking component, if I'm recalling correctly. yeah so definitely the startup seems super early and i think there's still a lot of kinks being worked out in terms of like are they going to be selling the model itself to hollywood are they going to be sent like you know using the model to create movies and selling those movies to the film studios but you're right it it doesn't sound like there's any aspect of like a kind of consumer facing version or like a social you know social version of this which obviously sort did did have Right.

15:03Stephanie Palazzolo: And so Bill Peebles, he was the head of Sora at OpenAI. He left. The interesting part here about the story is that they're actually having conversations with OpenAI about some of the technology. What are these conversations about? Yes, exactly. So what's interesting here is that in recent weeks and months, this unnamed stealthy startup has been having conversations with OpenAI about essentially licensing out the Sora model and then tweaking it and selling some version of that to Hollywood film studios. And so just taking a step back here, you know, Sora was one of the kind of many different product lines that OpenAI had tried to go after.

15:47Obviously, there's, you know, ChatGPT enterprise, but it really was like spreading its bets quite wide at some point, including the role of AI video and social media. And so basically OpenAI announced earlier this year that it wanted to cut back basically on Sora and its connected social media app because they wanted to, you know, really focus their bets on ChatGPT and Enterprise and kind of cut back on these side projects. So this is, I think, a way for OpenAI to make some sort of return and not really waste all the time and effort that went into making Sora, because now this other startup will be able to use the model and kind of revive it back from the dead in some way.

16:30Stephanie Palazzolo: It is interesting to me that if I think about the AI video startups that have positioned themselves towards Hollywood, I mean, Runway, Chris Valenzuela, he had a film background for an interest in film at the very least. And he had always very much positioned that company as a tool that he saw Hollywood studios using. And I think they might have even used the software for some stuff. They've now pivoted to enterprise, I think, a bit because of the fact that Hollywood is hard. I'm always so curious about the decision. I mean, Katzenberg, he's a Hollywood guy, right? I mean, he's had every job under the sun in that industry.

17:15Stephanie Palazzolo: like it's always interesting to me why these people start new company why not just take runway um i you know but i guess i don't know maybe there's the open ai angle i mean you got bill peebles in the mix here but why not just join runway i don't know like this is such an interesting question to me yeah i mean i think their thinking is probably like hey we have you know this huge hollywood guy in jeffrey katzenberg we have literally the person that made the sore model Bill Peoples, I think they're kind of thinking of it as like, we can, you know, kind of form this dream team that can connect with Hollywood and kind of sell this product to them in a way that hasn't been done before with companies like Runway, which obviously is doing quite well, but maybe doesn't have those kind of deep Hollywood ties that the way that this startup would have.

18:01And so, but you're totally right. It does feel like there have been several attempts at this kind of like AI video startup for film in Hollywood that, you know, I think they've all had a little bit of success, but there hasn't been, you know, this kind of like huge blowup success, you know, totally AI generated movie.

18:19Stephanie Palazzolo: Yeah, mostly because Hollywood, other than, well, other than Ben Affleck, I mean, I don't think people like AI that much. Yeah, I think, I think between the actors and, you know, the producers and everybody, I think there's definitely a very strong feeling of, you know, we don't feel super great about this AI stuff and it's taking our job. So, I mean, very understandable. I think similar fears that people in every other industry are having. Well, let me ask you one different question, though. So one thing you highlighted in the newsletter this morning with Corey is that this is a bit of a return for OpenAI to their spray and play, do everything strategy, which they took a step back from at least a year ago, I want to say.

19:06Stephanie Palazzolo: So why do you think they're coming back to this? I mean, didn't this come back to bite them initially? Yeah, so it's a very interesting kind of question to think about. I think on one hand, people might see this as a positive sign, where it's like, okay, this company's making a very smart move where it is able to kind of offload this product and not really hopefully spend that much time on it anymore, but still get a return on it potentially by, you know, getting equity in this new startup. So if that startup does really well, OpenAI can make some money from this thing that it spent a lot of compute and resources on.

19:45So on one hand, some people might see that as a kind of, you know, a continuation of the like cutting the side projects bet. But at the same time, we're increasingly seeing more signs that OpenAI is really starting to spread its bets wide again, right? We have things like, you know, it's really revived its efforts in robotics. it recently hired a Patreon co-founder to lead this new creator product division. It's obviously been pushing very hard on this personal device that, you know, we've been reporting on that they've been working on for years at this point. So at the same time, it does kind of feel like, you know, they are kind of starting to inch back towards having those other bets again.

20:25So it's a bit of this like contradiction. I do think part of this is that the person that really pushed to cut these side projects, Fiji Simo obviously had, you know, left OpenAI earlier this summer. And so maybe in her absence, there feels like there's a little bit of a power vacuum and people can kind of push again to, you know, really push for different projects that they work on. Or maybe the company just thinks that it over-rotated too much on the, like, we need to focus on, like, two things entirely. And now it's trying to find a better middle ground where they still are spreading its bets wider.

20:57Stephanie Palazzolo: So, look, I think we can make, I'll make the prediction. I think all these side projects are going away again. Like, I think there is no way they don't because, I mean, look, they're going to, you know, and they're obviously going to go public. I think they're going to have a better path to profitability. I think investors are going to start to ask the question, like$20 billion short of the revenue you thought that we were. What's your advantage here in profitability? I think they're all going away, but that just leaves more room for you to do some great reporting. Stephanie, I want to thank you for coming on.

21:34Stephanie Palazzolo: And sorry to add to your plate because I know that means that you're going to have more work to do. That is Stephanie Palazzolo, author of our AI Agenda newsletter here at The Information. People across Silicon Valley are using spare rooms in their homes to build server rooms as they try to optimize their own AI operations. I want to bring on Aaron Tilly, our reporter who wrote a story about that to share more of his reporting. Aaron, welcome back to the show. It's great to have you here. Hey Akash, thanks for having me. So do you have a server room in your house? I don't. I'm still a cloud user of AI.

22:10Yet. Yet. They're trying to convince me to buy some hardware.

22:14Stephanie Palazzolo: Great. Okay. What's going on here, Aaron? First of all, what rooms are people using? What are the servers they're building? Are they putting GPUs in there? What's the story here? Yeah, so there's this growing community of technologists and AI enthusiasts, geeks, nerds who are buying significant amounts of compute GPU, Mac minis, Mac studios, AMD GPUs to build out their own AI systems to use in their houses. They put them in their attics. They put them in their garages. They put them in their basements. These are dying computers that they have to sometimes retrofit their home electricity to put more power to these systems.

23:04And they are kind of creating their own AI labs inside their houses.

23:12Stephanie Palazzolo: And what happened to the cloud? well i think i think it's it's it's kind of a uh there's many facets to why this is happening um for a lot of these people it comes down to cost the cost of buying your own gpus and your own uh compute is a much cheaper to than renting the cloud using api costs to to send to the remote cloud and spend on monthly or even on these token per token costs. And it's just much cheaper to run it locally. So that's one big facet. Another is, of course, privacy and security. They are paranoid that these companies are using their data to train their their next models, and they are very they're very concerned about that.

24:04If they're building a product, they don't want these companies to be using that for their next models, or it's very private information that they want to use, that they don't want these big model companies to have.

24:18Stephanie Palazzolo: And I just want to make sure that I sort of understand this correctly. So you mentioned the hardware that is being used, but the models themselves, I mean, if they're so paranoid about the privacy element to this, are they downloading open weight models and using those models? Or is there a way to use the closed source models on your own server such that they won't train on it? What model are they actually using here in most cases? Yeah, no, this is almost 100 % open weight models of the Chinese models. And now some of the Western models coming out now, A very popular model is like the Queen, the Alibaba Queen model that's come out.

25:06Right.

25:07Stephanie Palazzolo: Okay. So, and by the way, we should say one thing you noted in your story is the electricity costs are pretty significant. You said they're retrograding their houses to fit it. I mean, why is this effective from a cost-benefit analysis to them? um you're still not paying cloud companies there i mean the the margin that they're trying to make there you know there's there's there's certainly a lot of money you have to invest in the the compute hardware and then the ongoing energy costs but compared to what you're spending in the margin that cloud companies and ai models want to make um selling their services their ai services to you, this is still, I mean, the amount of compute these people are consuming, it's still much cheaper.

25:55It's still much more cost effective. Okay.

25:58Stephanie Palazzolo: And where do you see this story going, Aaron? Is this, like, are we going to get back to having giant personal servers, do you think? Or are cloud companies going to find some way to offer this in a more secure fashion? yeah i think i think this will be a date debate in coming years um you know this certainly connects to an earlier era of the internet uh in the late 90s 2000s there are a lot of like technology enthusiasts that would buy their own home servers to store for digital media um i think the difference with now is that like this is uh the the the kind of information and the the stuff you're doing with AI is so much more proprietary than just owning a piece of, you know, digital media.

26:46This is, this is, you know, your, your products, this is your private life. And I think people just want a lot more control in this Dextera. I mean, I don't think everyone will, but I think there will be a market for people who want control and access all the time to their own AI compute. You know, there's just concerns that these AI cloud companies will revoke access or throttle or mess with the AI in ways that these people don't like and they won't control that. So I think I think this is more permanent than maybe prior eras of like sort of home servers. But it certainly won't be for everyone.

27:25I mean, there is just, you know, not everyone wants to be a IT, have, you know, run their own IT, run their own computer systems. So, you know, there's room for both, but I think this will last. Right.

27:39Stephanie Palazzolo: Hey, Aaron, I want to ask you about this Apple event that is scheduled for this week. Can you give us a bit of a primer? What should we be expecting from this event? Yeah, so Apple will be holding tomorrow on October 13th. It's a smart home unveiling kind of plan. This is many years in the works and maybe many years behind their competitors like Google and Amazon. But Apple will be coming out with a sort of home hub system and accessories for a sort of bigger smart home plan. And, you know, this is somewhat of a long time coming, but AI will be at the center of this system. and it took a long time for Apple to get Siri to where it's at, even though it's not even very advanced compared to what else we see out there.

28:28But now it's good enough to do some sort of centralized home system that you could control accessories and run your life through. So that will be unveiling this week.

28:42Stephanie Palazzolo: Can I ask, I don't have any data here, and this might be anecdotal. I feel like Apple's been doing more of these events in the past year and a half compared to the year and a half prior. Like, you can say that this is totally wrong, but is that the case? Because the theory I have here, hypothesis, maybe, I don't know, you see these AI companies now doing all these flashy releases, and Apple obviously doesn't have its own models to give biweekly updates on. I wonder if we could see Apple try to make a little bit of noise here with its even marginal product releases to stay ahead of the attention cycle.

29:25Stephanie Palazzolo: What do you think? Yeah, I think for sure. So Apple has a very predictable cadence for when it makes major product launches. and they are trying to ramp up this, like do it more with greater frequency, make product splashes more often throughout the year because it's been a very sort of like stayed product launch, a very traditional product launches, their fall iPhone releases, software summer releases at WWDC. And so with AI, like there's just so much news, so much new products emerging all the time every week. And they're just no longer in the same, like, same relevancy with all this happening.

30:11So I think they do aim to kind of increase frequency with how they announce and to talk about their products. And, yeah, we'll see if this is, like, if they can capture attention with this, with the new world. But more frequency is definitely in their interest. Great.

30:33Stephanie Palazzolo: Aaron, I want to thank you for coming on. That is Aaron Tilley, our Apple reporter here at The Information. Biohub, the main philanthropic effort of Mark Zuckerberg and Priscilla Chan, announced a new$1.8 billion partnership with the government to use AI to work towards their mission of curing all diseases. I want to bring on Amy Doxler-Marcus, our health and science reporter, for more on that. Amy, welcome back to the show. It's great to have you here. Hello. So Amy, set the stage for us here. How involved has Biohub been in the field of medicine? And also just give us a bit of a primer on what Biohub actually is.

31:14Yeah. Well, Mark Zuckerberg and Priscilla Chan have been putting a lot of their philanthropic efforts into trying to cure diseases. And that's been going on for around 10 years. Biohub is sort of the culmination, the latest iteration of what's been evolving. They announced a virtual biology initiative in April through Biohub with a$500 million commitment. And I think that they are trying to figure out how to use AI to advance biology and science to lead to their goal of curing diseases. And the announcement last week was sort of a call to other partners. And with the government and other tech companies answering that call, they've got a$1.8 billion commitment now.

32:02Stephanie Palazzolo: Okay. And help me understand, Biohub, it's an organization on its own? Does it work with the Chan Zuckerberg Initiative? How are these initiatives all connected? They are connected. They're all sort of together. Chan Zuckerberg Initiative is kind of the head, and Biohub is now the biomedical arm, really, that's leading the biomedical research for them. And Biohub really has a focus now of AI and bio. I think that's what the big difference is, this feeling that this is the future and that they want to go all in on this effort in particular. Okay. And there are other tech companies getting involved in this particular initiative, right?

32:48Stephanie Palazzolo: The Virtual Biology Initiative, I think is the name? Yes, that is correct. Yes. There's a lot of names, Amy. I mean, gosh, like how many different ways can you say I'm donating money to medical research? I don't know. Yes, I agree. I think, you know, listen, they want to get everyone excited. No, it's good. It's a very great, of all the things you can donate to, this is the one to do. I just, maybe they could make it simpler. Absolutely. And Mark Zuckerberg was kind of funny because Meta is also a new partner for this whole initiative, as is Google DeepMind and as is Isomorphic Labs, which is a spin out from Google DeepMind.

33:29And Mark, at one point when he was announcing this, he said, I'm wearing my Meta hat now. I'm wearing my Biohub hat now. So I think he would sympathize with some of your questions here. But, you know, look, this is like really, to me, I wanted to write about this, not just because of the announcement. I mean,$1.8 billion is impressive, but also for what the moment means, you know, for science. I feel that there's really like a cultural change that this represents. And that's what I wanted to focus on in my piece.

34:05Stephanie Palazzolo: Tell me more about that. What do you mean? Well, you know, after World War II, President, you know, Franklin Delano Roosevelt asked his head of science to like give set a vision for what science was going to look like to use all of the wartime resources, but devoted into peacetime efforts. And really, since World War II, the main engine of biomedical innovation in the United States has primarily been universities, particularly the big, major academic elite institutions that we know about. And now I think with this kind of announcement, especially with the government getting involved, I think that there's a sort of signal that, yes, universities are obviously still going to be important.

34:53But if we're envisioning this as a world of AI and AI medicine and AI science, that it's clear the government wants to partner with tech, you know, especially big tech, like the people that are getting involved with this initiative.

35:09Stephanie Palazzolo: And is this strictly because of the money that tech companies can offer? Or does it say something about the way that tech entrepreneurs move quickly to build their companies, to scale their companies? I wonder if this is sort of a cultural difference that maybe people feel like, you know, academia is slow and maybe tech companies move faster with this? I think part of it is that these big tech companies, they bring their models, their AI models, their frontier models. And I think one of the things they're bringing to the table is they're saying, you know, let's get all of this amazing data that's been collected for decades by the government and by universities.

35:52Let's get it AI ready. that's part of the money in this initiative is going to like go towards standardizing data sets so that AI can hopefully, you know, you know, find things that were harder to find before. So I think that that's part of it. I think part of it is that big tech has been courting the government, has been courting President Trump and I think are recognized for that. And I do think that, you know, the universities are, it's not that they're going away, not by a long shot. They're very, very important. They have the labs as well. But I do think that this symbolizes that the big tech companies, for a variety of reasons that we've been chronicling this past year, they're doing IPOs.

36:37They want to show that AI can be useful for humanity. They don't want everyone to think that it's going to kill us all. Maybe it'll save our lives. They have a lot of money. They want to partner. I think for all those reasons, there has been a willingness by the government to partner with them. And this announcement, I think, is a symbol of the change that's already underway. And I do believe there will be more to come.

37:03Stephanie Palazzolo: And Amy, I just want to make sure I understand your point here that you're making here. The shift from academia to big tech here in terms of AI research associated with the medical field. I mean, is this an observation that you've seen just with AI or are we talking about medical discovery and, you know, how much money gets poured into the research for cures, broadly speaking? Look, I think that this is – the tech companies have not yet proven that AI is going to discover drugs that haven't been, you know, found before. They are starting to show that they're moving money, they're moving efforts, they're moving research into drug discovery.

37:53Big Pharma is also getting involved, partnering sometimes with other tech companies, the big models, or developing their own models on top of the existing models to also try to accelerate drug discovery. I think everybody is aligned on the idea that the hope is AI is going to find new cures for diseases and is going to do it more rapidly. The jury is still out on whether they have done that. They are still that. We don't know that yet, but that is certainly the mission. And this kind of money going into that is further evidence that they're trying to do that. And they did tell me, the Biohub folks did tell me, that they hope to expand this partnership to get other groups involved, including pharma companies.

38:43So I think you could see this alliance expanding further as we move forward.

38:51Stephanie Palazzolo: And so if we go with that, Amy, if we see this expand further, I'm sort of curious what you think the repercussions or the ripple effects could be of that. Because where my head is going, and I don't know how much of an impact this has, but academia, universities, when they do research, they have these tech transfer groups, right? The groups that are in charge of taking research that happens in academic institutions and then sometimes spinning off companies, turning them into commercial entities, licensing the discoveries to these companies. Here we have sort of a slightly different model in that we have tech companies that have sort of created these philanthropic organizations or partnerships or whatever they are legally.

39:38Stephanie Palazzolo: um the point is you know it's it feels a little more corporate i guess insofar as how these discoveries are going to get made and so i wonder if you think there will be ripple effects insofar as maybe the commercial um implications of all this and it's all for a great cause by the way we should just say that at the outright you know like these are important things to focus on i'm just sort of trying to think about where how the money changes here yeah no look these are all great questions, and I'm not sure we have all the answers yet, but I do think that the pharma companies aren't necessarily going to want always to partner with the tech companies precisely because of what you're saying.

40:19They have their own business model. They want to keep going. I think that you're still going to have universities trying to spin out their own companies, but we've already seen signs that some of these big companies want to do their own drug discovery and want to set up wet labs. Anthropic did say that they have their own wet lab. It's small right now, but it could get bigger. And who knows what's going to happen with the other companies like Meta. Isomorphic Labs, the spin out from Google DeepMind, is an AI-driven drug discovery company. So you have so much blending. I think that you don't have these clear lines anymore.

40:57And I think that's the most fascinating part of this. It's not clear how it's going to spin out. But I do think that we are on the cusp of something that's different and that it merits closer attention for that very reason.

41:10Stephanie Palazzolo: And, I mean, you sort of, I imagine, have a bit of this race, silent race going. I mean, Anthropica is established as a wet lab, but I do imagine that for the PR public perception of these AI companies, I mean, the first lab, AI lab, to maybe have any kind of a breakthrough in medicine or biology are the things that I think people are really hoping AI can merit in the end. And I mean, that would be a huge win from a public perception perspective as well. Absolutely. And, you know, we wrote in other stories, we've written about challenges that are happening now to try to use AI to speed up, you know, discovery.

41:57Like, we're in a moment here is what I guess I'm trying to say. This is really, there are often these big paradigm shifts and they can come like out of World War II. You know, some big moment where there's resources, a huge amount of resources and new thinking. And I think when you see the government partnering with tech companies in this way, that it signals something that a moment is happening right now and will continue. you, obviously they are going to have to prove that they are engines of discovery, that medicines are coming out of this, that this is more than just, you know, a really nice way to spend your dollars, you know, your philanthropic dollars.

42:40But, you know, I think that it bears watching, but I think the symbolism of this is what I'm trying to get at, that we should be paying attention to this. There's a reason this is happening. Great.

42:53Stephanie Palazzolo: Well, Amy, I want to thank Thank you for coming on. That is Amy Doxler-Marcus, our health and science reporter here at The Information. That does it for today's show. A reminder, we are on this stream Monday through Friday at 10 a.m. Pacific, 1 p.m. Eastern. If you can't make it then, episodes are available on theinformation.com, on our YouTube channel, or wherever you get your podcasts. Make sure to follow us on social media, on X, on Instagram, on TikTok, and on LinkedIn. I am already excited for our next show tomorrow. Have a great rest of your Monday. Bye-bye for now.

From the publisher

The Information’s Microsoft Reporter Aaron Holmes talks with TITV Host Akash Pasricha about Microsoft’s new Maia AI chips and the executive behind Copilot’s overhaul. We also talk with Stephanie Palazzolo about the startup that could revive OpenAI’s Sora and Aaron Tilley about home AI computing and Apple’s smart home plans. Lastly, Amy Dockser Marcus discusses Mark Zuckerberg’s $1.8 billion biomedical partnership and big tech’s growing role in medical research.


Articles discussed in today’s show:

- How Microsoft’s New Office Chief Fixed Copilot and Won Over Nadella

https://www.theinformation.com/articles/microsofts-new-office-chief-fixed-copilot-won-nadella

- OpenAI’s Plan To Revive Sora

https://www.theinformation.com/newsletters/ai-agenda/openais-plan-revive-sora

- GPUs in the Wine Cellar: Why Techies Are Hoarding AI Compute in Their Homes

https://www.theinformation.com/articles/gpus-wine-cellar-techies-hoarding-ai-compute-homes/

- Zuckerberg’s Big Biotech Moment

https://www.theinformation.com/newsletters/the-weekend/zuckerbergs-big-biotech-moment


Subscribe:

- YouTube: https://www.youtube.com/@theinformation

- The Information: https://www.theinformation.com/subscribe_h

Sign up for the AI Agenda newsletter: https://www.theinformation.com/features/ai-agenda


TITV airs weekdays on YouTube, X and LinkedIn at 10AM PT / 1PM ET. Or check us out wherever you get your podcasts.


Follow us:

X: https://x.com/theinformation

IG: https://www.instagram.com/theinformation/

TikTok: https://www.tiktok.com/@titv.theinformation

LinkedIn: https://www.linkedin.com/company/theinformation/


Chapters:

01:13 - Introduction

02:45 - Microsoft’s new Maia AI chips

04:53 - The executive behind Copilot’s overhaul

14:14 - OpenAI’s possible Sora revival

22:44 - Building AI labs at home

28:38 - Apple’s smart home plans

31:39 - Zuckerberg’s $1.8 billion biotech partnership


More from The Information's TITV

All episodes