Anthropic’s Multi-Billion Revenue Share, Meta & Nvidia’s New Partnership, $10M Paydays for Data Center Executives

18 Feb 2026 · 41 min · 23 chapters

Ask about this episode

Ask anything about it. ChatGPT or Claude reads this page and answers with the times it was said.

Connect VO and ask about every podcast you hear, including the moments you saved. Add to ChatGPT · Add to Claude

In short

The Information's TITV Episode Summary

Episode Title

Anthropic’s Multi-Billion Revenue Share, Meta & Nvidia’s New Partnership, $10M Paydays for Data Center Executives

Host: Akash Pasricha Date: February 18, 2023 Special Guests: Sri Muppidi, Anissa Gardizi, Reed Duchscher, Stephanie Palazzolo

---

Episode Overview In this episode of TITV, the discussion revolves around the significant developments in the tech landscape, focusing on Anthropic's financial dealings, strategic partnerships between Meta and Nvidia, and the burgeoning creator economy. The episode also highlights the critical talent behind AI data centers and explores the emerging trend of "continual learning" in AI technology.

Key Topics Discussed

  1. Anthropic's Revenue Sharing Agreements
  2. Guest: Sri Muppidi
  3. Main Points:
  4. Anthropic is entering lucrative multi-billion dollar revenue sharing agreements with cloud providers like Amazon and Google.
  5. Anthropic’s revenue share with cloud partners has grown from $1.3 million in 2024 to an expected $1.9 billion this year, potentially reaching $6.4 billion next year.
  6. The revenue share model involves:
  7. Anthropic sharing 50% of gross profit with Amazon.
  8. Google receiving 20-30% of net revenue after infrastructure costs.
  9. Comparison to OpenAI, which shares 20% of total revenue with Microsoft.
  1. Meta and Nvidia Partnership
  2. Guest: Anissa Gardizi
  3. Main Points:
  4. Meta and Nvidia have announced a multi-year partnership, involving millions of Nvidia GPUs and collaboration on AI model refinement.
  5. This partnership solidifies their long-standing relationship amidst potential competition from Google.
  6. Discussion on the broader impacts of this partnership on Meta’s data center operations and chip use.
  1. Rise of Data Center Executives
  2. Guest: Anissa Gardizi
  3. Main Points:
  4. Highlighting key data center executives who are pivotal in the AI buildout.
  5. These individuals, often behind the scenes, are highly compensated (some exceeding $10 million) and are crucial as demand for data centers grows.
  1. The Creator Economy and Night Media
  2. Guest: Reed Duchscher
  3. Main Points:
  4. Night Media raised $70 million to expand its operations in the creator economy.
  5. Duchscher discusses the evolution of talent management and the shift towards owning live events and experiential marketing.
  6. The competitive landscape has intensified with traditional Hollywood agencies entering the creator space, challenging companies like Night Media.
  1. Continual Learning in AI
  2. Guest: Stephanie Palazzolo
  3. Main Points:
  4. Continual learning represents a shift towards AI that can adapt and learn similarly to humans, updating in real-time rather than through lengthy training processes.
  5. There are concerns about "fake" continual learning startups making exaggerated claims about their capabilities.
  6. Investors are cautious as many startups are not close to delivering true continual learning solutions.

---

Key Takeaways

  • Anthropic's Strategic Growth: The staggering growth in revenue sharing agreements signifies a competitive landscape where cloud providers play a crucial role in AI deployment.
  • Meta and Nvidia's Strengthened Ties: The partnership underlines a commitment to technological advancements, showcasing the importance of collaborative innovation in AI.
  • High Demand for Data Center Talent: The competitive compensation packages reflect the increasing value of skilled executives in a rapidly evolving tech environment.
  • Growth of the Creator Economy: The shift towards ownership of live events and diversified revenue streams indicates a maturation of the creator economy, where traditional boundaries are being redefined.
  • Future of AI Learning: While continual learning presents an enticing frontier for AI, the reality of achieving it remains complex, with significant skepticism regarding current startups' capabilities.

---

Additional Resources

  • [Anthropic's Revenue Sharing Agreements](https://www.theinformation.com/articles/anthropic-sweetens-deal-cloud-providers)
  • [Meta-Nvidia Partnership](https://www.theinformation.com/briefings/meta-nvidia-sign-strategic-partnership)
  • [Data Center Executives](https://www.theinformation.com/articles/meet-data-center-executives-leading-ai-buildout)
  • [AI Agenda Newsletter](https://www.theinformation.com/features/ai-agenda)

---

Follow The Information on Social Media:

  • [YouTube](https://www.youtube.com/@theinformation)
  • [X](https://x.com/theinformation)
  • [Instagram](https://www.instagram.com/theinformation/)
  • [TikTok](https://www.tiktok.com/@titv.theinformation)
  • [LinkedIn](https://www.linkedin.com/company/theinformation/)

Thank you for tuning in to this episode of TITV!

Written by AI. May contain mistakes. Listen to the episode to check what was said.

Chapters

Tap a time to open that second in VO

Anthropic Revenue Sharing Insights

0:45 to 1:25

Discussion on the multi-billion dollar revenue sharing agreements between Anthropic and cloud providers.

“And finally, we are taking a deep dive into continual learning the latest craze sweeping across Silicon Valley.”

Analyzing Cloud Partners' Dynamics

1:25 to 2:51

Exploration of the symbiotic relationships between Anthropic and its cloud partners.

“What did we learn about Anthropics revenue share agreements with the big cloud providers?”

OpenAI's Unique Revenue Agreement

2:51 to 6:29

Examination of OpenAI's revenue sharing model with Microsoft compared to Anthropic.

“Now, aside from the numbers, I want to understand the dynamic here between Anthropic and the cloud providers.”

Symbiotic Relationships in AI

6:29 to 7:45

Discussion on the mutual dependencies between AI companies and cloud providers.

“Who do you think has the leverage in this relationship?”

Meta & NVIDIA's Strategic Partnership

7:45 to 8:10

Overview of the new multi-year partnership between Meta and NVIDIA.

“especially because Anthropic and OpenAI, based on our analysis and our reporting, actually generate majority of their revenue from directly selling their business to customers.”

Implications of the Meta-NVIDIA Deal

8:10 to 11:10

Analysis of the potential impact of the Meta and NVIDIA partnership on chip usage and collaborations.

“Joining me now to help us understand the deal better is our cloud and compute reporter, Anissa Gardizi.”

Rising Talent in Data Center Sector

11:10 to 14:00

Discussion of key data center executives and emerging talent in the industry.

“Well, it relates nicely to a story that you published today.”

Data Center Executive Pay Trends

14:00 to 15:00

Explore how salaries for data center executives have surged recently.

“Some people actually did work at construction companies before, like a general contractor that worked on behalf of a hyperscaler.”

Rising Free Agents in Data Centers

15:00 to 16:10

Learn about key free agents in the data center sector and their potential impact.

“So I think this, you know, role that's always existed and always been important has just suddenly become so vital that, you know, they're getting paid the top dollar at their company.”

The Evolution of Knight Media

16:22 to 18:00

Dive into the history and growth of Knight Media since its inception.

“Night Media raised$70 million to build out its roster, its work in music and gaming, and also its live events portfolio.”
Show all 23 chapters

Talent Management's Central Role

18:00 to 19:18

Discover the significance of talent management in the digital age.

“You know, our venture studios got much bigger over the years, but the center of the business is still very much talent management.”

Competitive Landscape in Talent Management

19:18 to 20:58

Analyze the increasing competition among talent management agencies.

“And so creating experiences and moments, you know, I do very much live on the internet.”

Expanding into Live Events

20:58 to 22:28

Discuss the potential for Knight Media to enter the live events space.

“When I started the company 11 years ago, there wasn't a lot of people.”

Future of Creator Representation

22:28 to 23:49

Consider how the roles of agents and managers are evolving in the creator economy.

“I think this is like the, this will take place I think over the next five years, but like what is an agent and what is a manager?”

Challenges of Becoming a Major Creator

23:49 to 26:16

Examine the hurdles for new creators aiming for massive audiences.

“I don't know if I'll get back like into the day-to-day talent management business.”

Strategies for Success in the Algorithm Age

26:16 to 28:00

Learn strategies for creators to thrive amidst changing platform algorithms.

“And then you're able to launch like a product and a service.”

Monetization Trends in YouTube

28:00 to 29:30

Learn about how YouTube monetization has evolved over the years and its impact on creators.

“Thankfully, over the years, YouTube has figured out how to monetize better.”

Creators Transitioning to Hollywood

29:30 to 31:32

Explore the challenges creators face when breaking into Hollywood compared to established actors.

“about another article you wrote, which was about the intersection between the creator economy and Hollywood.”

The Role of Algorithms in Content Success

31:32 to 32:58

Understand how algorithms affect visibility and success for creators and the importance of consistency.

“Also, if you want to be an actor and an actress, these individuals go to school and they train to be seasoned actors.”

Interview with Reid Duxer

32:58 to 33:16

An engaging conversation with Reid Duxer on the creator economy and emerging trends.

“Like it's a very challenging career that takes immense focus and it's hard to do both of them at the same time.”

Understanding Continual Learning in AI

33:38 to 35:58

Learn how continual learning represents a shift in AI's ability to learn like humans and its implications.

“Well, most people refer to it as continual learning.”

Challenges of Continual Learning

35:58 to 37:55

Explore the obstacles startups face in developing true continual learning AI solutions.

“working, trying to fund startups working on this.”

The Future of AI and Continual Learning

37:55 to 40:34

Discuss the potential of continual learning to revolutionize AI and its impact on the future.

“So one startup I spoke with called Rider, which kind of develops enterprise AI tools, they basically came up with a new sort of model, you know, in late 2024, that is able to update itself with new information.”
Hear the part that matters, and keep it.Open this episode in VO. Double tap your headphones to save a moment as you listen.
Get VO free

Transcript

Automatic transcript. May contain errors.

0:13Welcome everyone to the information's TI TV my name is Akash Pasrich it is Wednesday February 18 Today, we are kicking off with some exclusive reporting on the multi-billion dollar revenue sharing details between Anthropic and the major cloud providers. We'll then turn to Meta and NVIDIA as the two companies deepen their relationship with a new multi-year partnership. We'll also dive into fresh reporting on the elite talent being drawn into the AI data center build out. We then got a conversation with Knight Media, the talent management company behind some of the Internet's biggest personalities and creators.

0:50And finally, we are taking a deep dive into continual learning the latest craze sweeping across Silicon Valley. It's going to be a fun show, so let's get right on into it. We have reported extensively at The Information on Anthropics' soaring revenue trajectory, But a new exclusive story from The Information Today gives a picture as to how much of that revenue is shared with cloud providers who help sell Anthropics tools to enterprises. I want to bring on the reporter behind that story. Shree Mupiti covers Anthropics and OpenAI for The Information. Shree, welcome back to the show. It's great to have you here.

1:25Excited to be here. What did we learn about Anthropics revenue share agreements with the big cloud providers? What we've learned is that Anthropic is significantly sharing a lot more money than we had previously known with its cloud partners. So we had previously known that Anthropic had shared about 1.3 million with its cloud partners in 2024. That number has now increased to about 400 million based on our analysis of the forecast that Anthropic had shared with potential investors as part of its prior round. And now this year, it's expected to be potentially$1.9 billion and even$6.4 billion by next year.

2:02And so this is specifically for cloud partners such as Amazon, Google, and Microsoft that resell Anthropik's models to their cloud customers. So a huge opportunity for Anthropik to be able to win over more enterprise customers. And then over time, that's expected to increase as well, according to what you found. Exactly. When the company had previously broken out partner profit share, those numbers were slightly lower than what we currently estimated to be. So, for example, last year over the summer, they expected it to be roughly$1.6 billion that they would bring in versus the$1.9 billion that they're expecting to bring in this year.

2:41And then next year, for example, it's$4.4 versus$6.4. for. And so you kind of really see this expectation that Anthropic could be bringing in much more money than they expected with their cloud partners and based on our analysis of Anthropic's forecasted figures. Now, aside from the numbers, I want to understand the dynamic here between Anthropic and the cloud providers. Help me understand why it is that the cloud providers are actually, in some ways, helping Anthropic sell their software because we know some of these companies are developing their own models that could be competitive to Anthropic.

3:19Why are they helping to sell Anthropic at all? It makes a lot of sense for these cloud partners because it mirrors a lot of software cloud partnerships that companies already have. For example, Microsoft and Databricks have a partnership where Microsoft would earn any money for Databricks software that they sell. And so it's a longstanding tradition, but specifically with the Anthropic relationships, what we have revealed in this reporting is that Anthropic actually shares 50 % of its gross profit back to Amazon for any revenue that Amazon sells to its customers. So it's a revenue share model that really works.

3:55And so gross profit meaning that any revenue minus any, for example, COGS, like inference costs, et cetera. And so that's a figure that Amazon likely would value. And then the relationship with Anthropic and Google is that Google typically shares between 20 to 30 % of any net revenue minus its infrastructure costs. And so we expect that to be similar for what they likely are sharing with Anthropic, sorry, Anthropic sharing with Google. But unfortunately, we weren't able to find out what specifically that revenue share model looks like for Microsoft. But you really see sort of the dynamics between these AI labs and the cloud partners of how much they're sharing.

4:38And so I just want to understand this. So we call this revenue sharing because it's sort of the easiest way to describe it. But what I'm hearing from you is there's revenue that the labs get from their customers. Then there's gross profit, which is after you subtract all of the inference costs, like you said. And then really, it's that gross profit. Then part of that is then what's paid out to the cloud providers. and so we can sort of see the margins getting slimmer and slimmer as this pie gets smaller and smaller. Is this unique to Anthropic? Is OpenAI in the same position? OpenAI does also have a revenue share, but with their long-standing partner, Microsoft.

5:21And so we had previously reported that OpenAI shares 20 % of its total revenue with Microsoft. And so rather than it just being, for example, any reselling model, like the way that Anthropic and Amazon, for example, have that relationship, it's actually just all total revenue, and that's because of the unique relationship OpenAI and Microsoft have since they were... OpenAI was first sort of getting started. So that's not a gross profit share. That's like a revenue revenue share. Exactly. It's 20 % of total revenue, and that's just a unique sort of relationship that both Microsoft and OpenAI have.

5:57Which is a good deal for Microsoft. I mean, they got there early, basically. Exactly, and they poured billions of dollars into the company but what we had reported in yesterday's story is that um what we had previously understood was that that deal lasts until 2030 but what we've revealed is that it actually goes till 2032 and those uh portions of the 20 revenue share will actually be weighted over those set of years but weighted more heavily towards the later years so it doesn't impact openai's cash flow as immediately now as it's continuing to burn cash. Who do you think has the leverage in this relationship?

6:36I mean, the cloud providers clearly need to be able to sell Anthropic and OpenAI's models to keep their offering compelling. On the other hand, the model companies need the cloud providers to run their businesses, and this revenue sharing agreement certainly shows how much loyalty they have to them. Who has the upper hand in this relationship, do you think? I'd say that it's a very symbiotic relationship right now because while Andropic is reselling, or sorry, Andropic has been, like the cloud providers are reselling Andropics models, the main area that Andropic is getting value from is actually getting access to compute and chips from its cloud partners.

7:19And same with, for example, OpenAI, where even though Microsoft might be selling OpenAI models, the majority of sort of value that OpenAI is gaining from Microsoft is, again, the cloud computing power. And so I would say that this is sort of a negotiation bundle across many aspects of its relationships, so cloud computing, revenue share, chips. and so when you look at it holistically, you really see that it is a symbiotic relationship, especially because Anthropic and OpenAI, based on our analysis and our reporting, actually generate majority of their revenue from directly selling their business to customers.

7:55And so you see sort of the value that they're trying to eke out between sort of all the relationships and aspects of the deals. Great, well Sri, I wanna thank you for coming on. That is Sri Mupiti, our OpenAI and Anthropic reporter here at The Information. Okay, Meta and NVIDIA are deepening their relationships, signing a multi-year strategic partnership where Meta will use millions of NVIDIA chips across its own data centers and through cloud providers. Joining me now to help us understand the deal better is our cloud and compute reporter, Anissa Gardizi. Anissa, welcome back to the show. It's great to have you here.

8:30Thanks, Akash. Explain to us what we need to know about this Meta and NVIDIA deal. So yesterday, NVIDIA and Meta announced what they call a multi-year strategic partnership. And there are lots of different prongs to the deal. But I think the main news of it was that Meta was really publicly committing to go deeper with NVIDIA as it relates to technology. They said that they're going to install millions of NVIDIA GPUs, the Blackwell GPUs that are out right now, and the Verirubin chips that are due out in a couple months. And Meta is going to use NVIDIA CPUs to run traditional workloads, use NVIDIA networking.

9:13NVIDIA engineers are going to help Meta refine its AI models. So there's a lot there. But I mean, the announcement was interesting because Meta and NVIDIA have long been partners. But for whatever reason, they wanted to get out this announcement yesterday and sort of confirm to the world that they will continue working together in the future. And I think this was part, it wasn't your prediction, but I think there were others at the Information who made the prediction that there could be a broader tie-up coming between NVIDIA and Meta. And so this was certainly something that was on our radar. Now, you reported late last year that Meta was also in talks with Google about potentially using their TPUs in their data centers.

9:59Do we have any reporting on if this deal would impact that, if one is to replace the other? Yeah. So especially in the cloud and chip world, we're seeing companies pursue lots of different options when it comes to which cloud providers they're using and which chips they're using. We reported, like you said, in late November, that Meta was in discussions with Google to use TPUs, both renting TPUs from Google and putting TPUs potentially in Meta's own data centers. And as far as we know, we don't know that much more about how the deal has progressed. But I would say from conversations and calls yesterday, people in the industry do not necessarily see this NVIDIA announcement as a sign that the Google deal is dead.

10:45But, you know, that was kind of everyone's initial reaction. So I think, you know, Jensen is obviously watching the market and who's using which chip and potentially seeing that Meta was considering using Google could have prompted them to say, hey, let's make sure that we have, you know, something in place with Meta. Now, we don't know the terms of the Meta NVIDIA deal and whether this means Meta is going to get discounts on NVIDIA gear. So lots more to find out. Well, it relates nicely to a story that you published today. We have talent trackers here at The Information, and the latest talent tracker that you put together was a list of the data center executives that should absolutely be on our audience's radar.

11:29As the data center story gets more and more heated, tell us a little bit about what your reflections were from this reporting, and then we'll get into who was actually on the list specifically. Sure. So I remember a couple months ago hearing about all the crazy stories in AI research land, you know, the direct phone calls from Mark Zuckerberg, the crazy offers. And over the past couple of months, it became very clear to me that a very similar dynamic was happening in the data center sector. You know, the types of people we highlighted on the list are not CEOs. They're not really people that you see out and about talking about their companies, but they are the, you know, I would say secret weapon maybe at their companies in terms of getting data center capacity online quickly and on budget.

12:16These are sort of the types of people that, you know, everyone is trying to poach and who are highly compensated at their firms to stay. So I thought it would make sense to highlight some of those people and get them on everyone's radar because these are people who are going to be very important as deadlines approach. You know, everything is going to kind of fall on them. Okay, so who's on the list? Give us a couple of the names. So a couple of the names. One is Chris Dolan from Crusoe. He's a very key figure in developing OpenAI's data center in Abilene, Texas. We have Rachel Peterson, a longtime meta employee who's vice president of their data centers.

12:55And we have some rising stars too, like FluidSax Corey Smith, who a lot of people worked with when he was at Oracle. So there are 16 people in total. Highly recommend checking it out, but those are a few people. What about the backgrounds of all these people? I mean, are these all people that share real estate experience in their history? Are they people that were former startup founders that then got acquired and took over the operations? Are there any common threads connecting this group at all? Yeah, it's a really great, great question. And as I talk to folks, you know, people really said you should look at everyone's background because that will kind of tell you everything you need to know about whether they are this key person that you're looking for.

13:39And I would say broadly, a lot of people on the list have been in the data center sector for 20 plus years. So they were working on data centers before they were cool, before they had anything to do with AI. and you'll see a lot of people with roles like head of design, engineering, and construction. Some people actually did work at construction companies before, like a general contractor that worked on behalf of a hyperscaler. You see some well-known data center names like Digital Realty, Equinix. So those are the types of people that really made the cut for the list. But it's interesting. One person on the list actually came out of retirement to get back into the sector.

14:23So, you know, there was one person we met. Who was that person who came out of retirement? That one was Chris Dolan, who was advising him so and then decided to join full-time. So he came back. And so are all these people getting paid pretty handsomely now, you know, more than they were 10 years ago when data centers were not as closely watched? Exactly. I mean, you can imagine if you come out of retirement or, you know, make a big shift, you know, 30 years into your career, you're likely getting heavily compensated for it. But yeah, like we said in the story, some of these pay packages are above$10 million, and that was not the case a couple years ago.

15:00So I think this, you know, role that's always existed and always been important has just suddenly become so vital that, you know, they're getting paid the top dollar at their company. Okay. And before you go, there were also a couple of free agents that you said that we should be watching. People that are not currently at companies that you think could get snapped up. Who are those folks? Yeah. One that we heard a lot was Joe Cava. He's a 17-year Google veteran and he retired last summer, but tons of people are retired. Yeah. As we know, tons of people retire now in data center land. They come back.

15:35Right. And people highly suspect that he will re-enter the industry. And then another one that we called out was someone who is now doing a lot of advising for private equity firms, Dan Madrigal. He was at Oracle and key to that Stargate data center in Texas. So if he ever wants to get back into the industry, I think that there will be plenty of firms calling him. Great. Well, to any of those free agents, we are inviting you to come on the show and tell us a little bit of what you're doing and who's pitching you offers right now because they certainly are names to watch. Anissa, I want to thank you for coming on.

16:11That is Anissa Gardizi, our cloud and compute reporter here at The Information. The talent management agency behind some of the internet's biggest creators and personalities has raised a new funding round. Night Media raised$70 million to build out its roster, its work in music and gaming, and also its live events portfolio. Joining me now is Reid Duxcher, founder and CEO of Knight Media. Reid, welcome to the show. It's great to have you here. Walk me through a little bit of the history of Knight. How long ago did you start the company and what did you start it as when it began? Yeah, started 11 years ago.

16:47I believe it was in July of 2015. That time, YouTube wasn't as prominent as it was today. I met Dude Perfect in 2014, left my job as a sports agent, got really obsessed with the platform and started spending a lot of time there, thought it was the right move and in 2015 I just couldn't get this out of my head that dude perfect wasn't going to be the last creator to build production company on top of youtube and so made that jump started the business been doing it ever since I'm more obsessed today than ever with the internet economy and let's just call it like the attention economy so it's been a fun ride so so it's always really been centered on the talent management concept yeah it has I think early on, the idea was represent the biggest creators in the world, build a venture studio that could eventually build companies and partner with them.

17:42That led to Beastables and some of the first things coming out of the venture studio. We went on to found 12 companies since then. But that was really the idea. The manager was the closest person to the talent. The manager usually ends up becoming the business partner. And so early on, that was the focus, still the focus today. You know, our venture studios got much bigger over the years, but the center of the business is still very much talent management. And so when people look at the structure of the business and what you're going to do with the$70 million now, I mean, do you still imagine that most of the revenue is going to come from management fees for managing this roster, or are you really trying to sort of shift the concentration towards, you know, having a bigger stake in some of these businesses and building that out?

18:27We have to continue to win the internet. You know, that's always been our through line is just be the most internet obsessed company, you know, be in the trenches of the internet. And I think talent management is a good way for us to do that. You know, talent gives us a wedge into a lot of the other things that we want to do. But the general focus of the company is continue to win internet, continue to grow talent management. You know, I think a lot of things are now transcending with the internet. You know, I think not even with talent, sports, comedy, podcasting, music, you know, everything is becoming so internet native, where you discover talent is on the internet.

19:01And so I think our through line is, you know, today, like we think that we are becoming more right every single day in our thesis. And so yes, we're going to continue to expand in talent management. We acquired an experiential marketing agency called Experiential Supply three months ago. That has been a big focus of ours as well. I'll just say it's, you know, my idea around people want to continue to touch grass. And so creating experiences and moments, you know, I do very much live on the internet. I'm a doom scroller. I have high screen time. But I do think that people value in-person experiences and live events.

19:34And so we've been really good at diversifying the revenue stream over the years. And so as we continue to grow, we'll look at things that we think use the internet or are transcended by the internet. And talent management just continues to be that piece. Who do you compete with now? I see all these sort of, quote unquote, old-fashioned agencies, older agencies that have been around Hollywood for a long time. they're getting more and more into the creator economy space. Do you see a lot more competition? And are you having to offer more in terms of deals? Yeah, the Hollywood agencies, UTA, WMECA, Gersh, they have entered the space probably about three or four years ago meaningfully.

20:16I think they were always trying to do digital, but it was always an afterthought at some of these agencies. UTA has probably been the one that's focused the most on the digital space. And so while we compete with them in some regards, there's a lot of things we don't do because we're a management company that they do, whether that is booking live tours, some of the stuff's around the edges. And so we've used them in the past. There's a lot more upstart small management companies coming into digital. I think the barrier to entry is relatively small. And so we're seeing a lot of two, three employee shops come up through the ecosystem.

20:52system, we've acquired a lot of those companies over the years and will continue to do so. And so, yeah, it's become a much more competitive space. When I started the company 11 years ago, there wasn't a lot of people. It was for the people that understand the world of the internet, it was the MCN era. And so it was Machinima, Fullscreen, Maker, Studio 71, Broadband TV. And so I was competing against all these MCNs that were signing like thousands of creators on a monthly basis and they couldn't really service them. And so I found a little bit of a niche just not representing that many people, but going after the top ones.

21:28And so I think it'll continue to get more competitive over the years, but we've done a good job of just being like the biggest dog in the space. But I mean, I'm sort of looking at what those agencies have today that you don't offer. I mean, the live touring thing, for example, I mean, wouldn't that be an example of a space that you would sort of naturally see yourself having to go into because of how lucrative events can be and also just where the business can go from a growth perspective it's a good question i don't know if we'll ever push into live events just the booking of live events i would rather own the live events uh than just book so like a single like a single singular event like one one type of thing i mean we're looking at this right now with the experiential business that we acquired they they run an event called haunt that is in LA and New Jersey around Halloween.

22:18Like we're going to continue to expand upon owned and operated live events. That to me is a much more interesting long-term play than us just being on the booking side of live. Now we'll look at it. I think this is like the, this will take place I think over the next five years, but like what is an agent and what is a manager? I think more and more they're becoming the same thing. And so if I just think ahead five, six, seven years from now, I'm of the mindset that it's just single representation companies. And so, you know, WME will own a management company. It'll be single representation. And so I think in that world of talent representation, yes, we will do booking.

22:57We will do non-scripted, scripted television. Like I think we may end up getting there. But again, ruthless obsession with the internet. I don't know. That just continues to be where we're focused. Who do you want to sign? Who are you after right now? I don't know if I can tell. You can make your pitch right here. Right now, we'll post it. You can make me elevator pitch. Yeah, I've been lucky. In my career, I went from Dude Perfect to Mr. Beast, represented Mr. Beast for almost seven years. And we accomplished a lot in that tenure that I represented him. So who's going to be the next Mr. Beast?

23:35That's the question. i don't know if it's possible in today's youtube to break through like jimmy did you know i think the algorithms don't allow someone to break through all the noise they they kind of feed content the type of content that people want to watch where jimmy and i figured out how to get hundreds of millions of views on a single video so i i think the next mr beast is not going to come from youtube you know i think maybe it comes from twitch maybe it comes from some other place kai sinat obviously has done a very good job on twitch and you know we've seen that career grow there's not someone that is like top of my mind right now i'm pretty focused on like how do we continue to hire the best people retain the best people what businesses do we want to look at acquiring you know i think if i was to get back into representing talent it would probably be something in the music space um either within hip-hop dance or country but that's yeah i don't know.

24:31I don't know if I'll get back like into the day-to-day talent management business. But it sort of leads me to the broader question here, which is, you know, the question around, will there be another Mr. Beast, which is exactly what you're hitting on. You know, you've written a little bit about this in your newsletter, this idea that we're now transitioning into, I think what you call sort of the micro giants sort of sphere. And I want your sort of view on how the micro giants sort of differ from micro influencers, because a couple of years ago, you know, I remember a time when, hey, it was the influencer actually with 10 to 20 ,000 people on Instagram that I sort of had a relationship with.

25:11You know, that's what people said were the future. Now we've sort of come into this middle ground where I feel like people are saying, well, there's not going to be a Mr. Beast. You need more than 20 ,000. But I don't know where that leaves us because, I mean, in my mind, like, you know, if I don't have, if someone doesn't have 400 million subscribers, fine, you know, that's an astronomical number, but you still have to put in the same amount of work to get, you know, the two or three million subscribers. And so I feel like the goal is still kind of where we were initially, right? Jimmy and I talked about this a lot early on in his career is, you know, PewDiePie dominated YouTube for a long period of time.

25:50And we always thought that people used to say that there would never be another PewDiePie. No one would ever surpass him in subscribers. Obviously, we proved that very wrong. And we not only passed him, but I believe Jimmy's closing in on 500 million subscribers. So it's possible. I think what I've written about and what I've seen happen with the social platforms, TikTok, Instagram, and YouTube is that they don't want to incentivize creators to get too big. they want to push everyone down to this like middle layer that you kind of just said which is like 1 million to 5 million subscribers pulling hopefully 500 ,000 views per video on long form videos the the beauty of that I think with that world is you can have high American audience you can have highly engaged audience you're not getting tens of millions of views and a lot of those views are from different countries or a lot of those views are just like bought it you know you're focused on a specific area, whether that's like outdoors, you're like reviewing trains, or you're playing video games, and you have a very specific audience to that niche.

26:51And then you're able to launch like a product and a service. You know, when Jimmy and I thought about doing chocolate, we wanted something that was very broad, that could be globally appealing, that everyone buys, that could be a daily use item. You know, I think in the new world of creators, if you're a channel that focuses on farming in rural Iowa, you have to think about your product a little bit different because your fan base is probably just people centered in that area. And so I don't know, I'm pretty bullish about the future of internet. I just don't think that the platforms incentivize someone to break out like a Jimmy, PewDiePie, Logan Paul.

Read the full transcript

27:30It's just much more challenging because their algorithms have gotten so good at feeding people the type of content that they only want to watch. So how do you combat them? As a management company? Yeah, yeah. I mean, what's your advice then? I mean, it's you against the algorithm, really. I mean, my advice is use short form as a metric for discoverability. Long form continues to be the gold standard of building a business on the internet. And so you have to have a long form strategy on YouTube. Thankfully, over the years, YouTube has figured out how to monetize better. So the ad units or the RPMs on videos have gone up, especially in spaces like automotive, cooking, technology.

28:12You know, the RPMs are much higher than they were eight years ago when Jimmy and I were starting this YouTube business. And so, you know, I'm bullish on it. I think the longer your videos, the more ads you can put inside the videos. And so creators on a relative basis are actually making more money per video than they ever have. um and so that's where we've been focused is like how do you just continue to help creators grow build production companies hire editors hire cameramen make greater videos i just think the days of a creator getting 15 to 20 million views per video on youtube is so difficult and if you just look back three years ago there's a lot of creators pulling 10 million views a video you know markiplier doesn't do that anymore even mark rober for the most part like struggles to pull the amount of viewership that he used to.

29:00But I don't think that's a bad thing. I think that this stuff just ebb and flows. You have to react to it. And as a management company, we've taken a little earlier risks on people that we may have not signed in the past because they've been too small for night. And now we've taken some early risks. And I think we've hit on a lot of people. I mean, we found Outdoor Boys very early. We found the Rizzler very early. I still feel like I have a great eye for talent. Now I just have to look much earlier. Let me ask you one quick question about another article you wrote, which was about the intersection between the creator economy and Hollywood.

29:37And look, we have seen these worlds mix more and more. We've seen Hollywood celebrities get more and more into podcasting, which is a space that you're very closely involved in. We've also seen creators, as you've written about, get more and more into the movie scene. And you You have creators now producing and filming their own movies. I wonder how you think about, you know, which one is the harder transition? And I think the answer really is creators getting into Hollywood because, I mean, Hollywood stars, they launch podcasts, you know, every month it seems, and they really catch on. And yet it still feels like creators are fighting an uphill battle to get legitimacy in Hollywood.

30:17And I wonder what the challenges are you think that you're seeing? Well, creators breaking into Hollywood is challenging because they're still gatekeepers, right? The beauty about YouTube with a traditional actor, actress, musician crossing over the internet is there's no gatekeepers. You can create your own channel, own your IP. You have creative final say. You can do everything you want to do. You don't have to sell something to a studio. You don't have to take a casting director out to 15 dinners to get them to put you in their next series. You know, there's not any of that. And so I think the transitions, I don't know which one is more difficult.

30:54I just know they're very different. Like a lot of creators now are signing Netflix deals. They're, you know, they're getting announced every single week. It seems like Netflix is working with a new creator. But it's taken a long time for us to get here. Netflix has never taken YouTube creators seriously. They always kind of thought it was like a downgraded, sloppy style of content, and they didn't think that it fit on their platform. But now we're seeing deals with Mark Rober, Salish Matter, a lot more things, I think, to be announced over the coming months. But I do think it's a harder transition to get into Hollywood just because of that nature of you have to impress people.

31:29There are gatekeepers. It's a much more challenging world to get in. Also, if you want to be an actor and an actress, these individuals go to school and they train to be seasoned actors. I think a lot of creators in the past had thought that they could just make that transition easily because they were content creator. It's not as easy as you think. It's very hard, but you are right. If you're Jason Bateman, you can start a podcast. You're already a very entertaining individual. You can put that up. But the beauty of the internet is like no one's telling him no. He can do whatever he wants. He can talk about whatever he wants.

32:03That's what I think is beautiful about the internet and why a lot of Hollywood talent are now crossing over into this genre because they had to answer to someone. I will say, going back to what you said earlier, there are no gatekeepers on the internet, but the algorithms in some ways are kind of a gatekeeper, right, to who succeeds in some ways. Yeah, in some ways. I think in podcasting less, there's really no discoverability on Apple and Spotify. You have to promote and then people have to go watch it. But yeah, you have to understand the algorithm to make a good video. You have to understand retention.

32:34You have to understand thumbnails. It is a craft that they all have to learn. And I think a lot of individuals, you know, The Rock had a YouTube channel at one given point in time. And, you know, I think they just struggled with cadence and consistency. I actually don't think it was a problem of like coming up with ideas and making good videos. Like everyone watched the videos that The Rock uploaded. It's just a grind. Like it's the same thing with being an actress or an actor or a YouTuber. Like it's a very challenging career that takes immense focus and it's hard to do both of them at the same time.

33:07Great. Well, Reid, I want to thank you for coming on. That is Reid Duxer, CEO and founder of Night Media here on TI TV. Venture capitalists are increasingly looking to fund a new pocket of AI technology called continual learning. It is the latest buzzword that represents the way that companies are trying to improve their AI models. My colleague Stephanie Palazzolo wrote about that trend in her AI Agenda newsletter today. I want to bring her on to talk all about it. Steph, welcome back to the show. It's great to have you here. Hey, it's great to be back. Okay. Is it continual learning? Continuous learning?

33:44Which one is it? Well, most people refer to it as continual learning. Continual learning. Okay. So what is it? Yeah. So basically, continual learning is a kind of newer buzzword that's really caught on in the the AI industry in the last, you know, six months or so. But basically it's idea of AI that's able to learn more like the way humans do. So, you know, learning on the fly from real world experience, kind of trial and error in real life versus the way that AI models learn today, which is, you know, going through these very long and very expensive kind of formal training processes where they're trained on tons and tons of pages of data from the internet and other places using billions of dollars that are spent on chips.

34:38Okay. So how does that differ from the existing way of doing it? I mean, we hear about reinforcement learning, human feedback. It's different. Yeah. So kind of the way that AI models work today is, you know, if something new happens in like the real world, so like maybe like a sports game happens, or some sort of like world event happens, the AI models actually don't have that new knowledge. And so kind of like a way to work around this that AI companies have used is allowing AI models to look up information on the internet. But that's different than kind of an AI model, kind of in real time like you know uh world events that are happening or uh other updates to its knowledge that like a human might kind of know um and so with continual learning there's this idea that uh this new type of ai will be able to update itself in real time um and then won't have to go through this very long training process every single time it wants to get like new skills or learn new information about things that are happening in the real world.

35:47So investors are starting to fund a lot of startups in this category? Yeah. So it's kind of a bit of a weird case here where, yes, investors are trying to find startups working, trying to fund startups working on this. But they're also telling me too that increasingly they're actually getting pitched by startups that are claiming that they have solved continual learning or are close to solving it whenever that's really just not the case. And so that's kind of what I wrote about in my column, this idea that investors are increasingly getting pitched by, quote unquote, kind of fake continual learning startups.

36:21And so I guess fake continual learning is really startups over promising on what it is they're actually doing and investors actually having a little bit of hesitation about how accurate the results are in some cases. Yeah, exactly. And so in some of these cases, these startups, you know, are basically kind of finding clever tricks and workarounds and trying to frame them as, quote unquote, continual learning whenever that isn't really the case. And so one example is, you know, basically giving models this kind of like scratch pad that they can use to take notes on and update with new information.

37:00So like, let's say that like a sports game happens, so they can take a note on this quote unquote, like metaphorical scratch pad and say, okay, you're like the result of the sports game, or maybe this quarter, maybe this election happened, here's the results of the election. And then whenever they're answering questions from users, they can kind of reference that scratch pad. but you know that's kind of a more of a kind of clever workaround and I think most people whenever they imagine super intelligence you're not really thinking of an AI model that's kind of like looking at a cheat sheet basically to update itself with new information and then also at some point you know the model could theoretically like run out of room on that cheat sheet or kind of like reach the end of its context window another example that I kind of heard of, you know, there definitely are cases where startups are doing interesting research here, but I think most of them are nowhere near solving continual learning.

37:55So one startup I spoke with called Rider, which kind of develops enterprise AI tools, they basically came up with a new sort of model, you know, in late 2024, that is able to update itself with new information. But one problem that they ran into is that the model couldn't actually distinguish when it's given new information, like what is actually new knowledge versus just made up information that somebody might be trying to feed to it to trick it. So for instance, I could tell the model, okay, you know, so-and-so just won the presidential election in the United States. And that could be true. Or I could tell it, hi, I'm the CEO of this company and I deserve to have all the salary information of every single person that works here.

38:43That's obviously made up. And so the issue that they ran into is like, how do you tell the model or how do you help the model figure out kind of what's real and what's fake? And so they were able to find a solution for that, but that doesn't really scale to models that are working with, you know, hundreds of millions of consumers like the ones powering, you know, ChatGPT, for instance. So, Steph, if someone were to ask you, I mean, is this the next big thing? What would you say? Is this sort of the term that we're going to start hearing more and more of? You know, another example of this is, for example, we had somebody on the show a couple of weeks ago talking about context graphs.

39:22You know, that was another term that people were starting to throw around. And maybe that was a pocket of startups that we may hear more about. I mean, how much confidence do you have that continual learning will actually be the thing that becomes, you know, game changing for AI? I mean, I think a lot of researchers do think that it is the kind of, you know, next major kind of holy grail that we need to solve in order to kind of like reach AI that's able to be super intelligent and kind of perform better than humans on most sorts of tasks. So I think it's definitely a given that every single AI lab right now probably has researchers working on this problem.

40:10I'm not sure if it's entirely clear yet that this is going to end up, you know, kind of leading to super intelligent AI or like being the one thing that stands between us and super intelligent AI. But it's definitely an area where a ton of research is going on. And I think all the major labs and several newer Neo labs have risen up just specifically to work on this problem. Great. Well, Steph, I want to thank you for coming on. That is Stephanie Palazzolo, our AI 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.

40:46Eastern. I want to thank you all for tuning in. We really do appreciate your viewership. Make sure to subscribe to the information on YouTube, X, Instagram, TikTok, and check us out wherever you get your podcasts. Have a great rest of your Wednesday. Bye-bye for now.

From the publisher

Sri Muppidi talks with TITV Host Akash Pasricha about Anthropic's multi-billion dollar revenue sharing agreements with Amazon and Google. Anissa Gardizy covers the strategic Meta-Nvidia partnership and the elite executives leading the AI data center buildout. Plus, we talk with Reed Duchscher about Night Media’s $70 million funding round and the future of the creator economy, and we get into the "continual learning" trend with Stephanie Palazzolo.

Articles discussed on this episode: 

https://www.theinformation.com/articles/anthropic-sweetens-deal-cloud-providers 

https://www.theinformation.com/briefings/meta-nvidia-sign-strategic-partnership 

https://www.theinformation.com/articles/meet-data-center-executives-leading-ai-buildout 

https://www.theinformation.com/newsletters/ai-agenda/get-tricked-fake-continual-learning 

Subscribe: 

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/


More from The Information's TITV

All 304 episodes
Anthropic’s Multi-Billion Revenue Share, Meta & Nvidia’s New Partnership, $10M Paydays for Data Center ExecutivesThe Information's TITV · 41 min
Listen in VO