In short
Bloomberg Tech Podcast Episode Notes
Episode Details
- Title: US Eases Path for Nvidia to Sell H200s to China
- Hosts: Caroline Hyde and Ed Ludlow
- Date: [Insert Date Here]
- Description: Discusses the US Commerce Department's changes to export regulations for Nvidia, insights from Skild AI CEO Deepak Pathak on their recent funding, and Netflix's strategic moves regarding Warner Bros. Discovery.
Key Topics Discussed
US Commerce Department and Nvidia
- New Regulations: The US Commerce Department will allow Nvidia to export its H200 AI chips to China under specific conditions, moving away from a prior presumption of denial for export licenses.
- Key Requirements for Export:
- Companies must demonstrate that exports will not create a supply shortage in the US.
- Manufacturing for China must not displace production intended for US customers.
- Implement strict "know-your-customer" procedures to avoid unauthorized distribution.
Market Reactions
- Stock Performance: Nvidia's shares saw a decline following the announcement, while AMD's performance fluctuated, reflecting market uncertainty.
- Future Outlook:
- Market analysts speculate a potential $50 billion annual revenue opportunity for Nvidia in China, although current estimates may be conservative.
- Companies like Alibaba and ByteDance have expressed interest in purchasing a significant number of chips.
Skild AI Interview
- Funding Round: Skild AI has raised $1.4 billion, tripling its valuation within a year.
- Core Technology: Skild AI's "omnibodied brain" aims to enable robots to learn and adapt across various environments and tasks.
- Training Methodology: Combines human video analysis with simulation data to train AI, addressing the current limitations in real-world data availability.
Netflix's Strategy
- Bid for Warner Bros. Discovery: Netflix is considering a revised, all-cash offer for Warner Bros. Discovery, which aims to streamline negotiations amid competition from Paramount.
- Market Context: The move is in response to shareholder preferences for immediate payment and to address complexities tied to stock offers.
Security in AI
- Depth First Funding: AI security startup Depth First has closed a $40 million Series A funding round, aiming to identify and remediate software vulnerabilities at an unprecedented scale.
- Performance Metrics: The platform reportedly identifies verified vulnerabilities eight times more effectively than previous leading solutions.
Other Notable Mentions
- Airbnb's New CTO: Announced plans to integrate AI across its platform to enhance user experience and operational efficiency.
- Spotify's Leadership Transition: Co-CEOs Gustav Soderstrom and Alex Nordstrom face challenges in navigating competition and user engagement strategies.
Key Takeaways
- Regulatory Changes Impacting AI Technology: Easing restrictions on chip exports to China could significantly alter Nvidia's market dynamics.
- Investment in Robotics and AI: The emphasis on creating adaptive AI systems signifies a shift towards more intelligent robotics solutions.
- Market Competition: Major players like Netflix and Spotify are adapting strategies in response to shareholder demands and evolving market conditions.
- Increased Focus on Cybersecurity: As AI technologies proliferate, the demand for robust cybersecurity solutions is more critical than ever.
Conclusion This episode of Bloomberg Tech provides critical insights into the evolving landscape of technology and investment, highlighting how regulatory changes, innovative startups, and strategic corporate maneuvers shape the future of the tech industry.
---
*For a complete listen of this episode, check out [Bloomberg Tech](https://www.bloomberglive.com/tech).*
Written by AI. May contain mistakes. Listen to the episode to check what was said.
Chapters
Tap a time to open that second in VOOverview of Today's Topics
0:45 to 1:46
An outline of the key stories including NVIDIA and the AI chip market.
“You can find new episodes of the Bloomberg Daybreak Europe podcast by 7am in Dublin or 8am in Brussels, Berlin and Paris.”
NVIDIA's Path to Selling H200s to China
1:46 to 3:40
Discussion on the U.S. Commerce Department's new export rules for NVIDIA and AMD.
“As you see, though, those names in the red and in general, the market on the downside.”
Implications of New Export Rules
3:40 to 5:51
Analysis of how the new rules affect NVIDIA's chip sales and market dynamics.
“that the companies must show that the exports to China will not create a supply shortage here.”
China's Strategy and NVIDIA's Market Position
5:51 to 7:30
Insights into the competitive landscape and China's pivot towards domestic chips.
“Let's stick with what this means for NVIDIA and more broadly for semiconductors.”
Global Supply Chain Challenges
7:30 to 11:15
Discussion about supply chain issues affecting data centers and power requirements.
“And that goes for China or even United States competitors.”
AMD and Intel's Competition in CPUs
11:15 to 12:09
Evaluation of AMD and Intel's positions in the CPU market and their strategies.
“No, I do not believe that will be the outcome.”
Legislation on Non-Consensual AI Images
12:09 to 14:01
Examination of a new bill allowing victims of AI-generated images to sue.
“Now coming up, anger grows over sexually explicit, non-consensual AI generated images on X.”
Deep Fake Legislation Discussion
14:01 to 16:44
Learn about the implications of the Defiance Act and the response from lawmakers regarding AI-generated content.
“any naked, underage images generated by Grok.”
Airbnb's New CTO and AI Strategy
18:06 to 21:26
Explore Airbnb's strategic shift towards integrating AI under their new CTO.
“I'm extremely excited, like Brian said, to bring craft to the art of deploying AI first features and products.”
Netflix's Bid for Warner Brothers Discovery
21:27 to 24:26
Understand Netflix's strategy and the implications of shifting to an all-cash offer for the acquisition.
“Let's turn our attention to entertainment now.”
Show all 17 chapters
Skilled AI's Funding and Vision
24:26 to 28:00
Learn about Skilled AI's innovative approach to robotics and their recent funding success.
“Now, coming up on Bloomberg Tech, robotics startup, skilled AI, just closed a$1.4 billion funding round, a big valuation, tripling from a year.”
Exploring Robotics and Revenue Models
28:00 to 31:58
Learn about the unique revenue sources for robotics through AI and simulations.
“And soon after, many companies had their own LLM models.”
Depth First's Funding and AI Security
33:08 to 36:31
Discussion on Depth First's AI-based security solutions and funding insights.
“check it out for a complete wrap-up of your business week.”
Building Trust with Slackbot
36:31 to 42:01
Insights on Slack's new bot, security concerns, and productivity impacts.
“And you're seeing Palo Alto Networks, even ServiceNow making some big acquisitions in the space, trying to get ahead of the cyber risks around AI.”
The Role of Slackbot in Enhancing Productivity
42:01 to 43:28
Learn about how Slackbot enhances user productivity and the metrics used to measure its impact.
“It needs to be deeply trusted by the user, but also deeply trusted by the enterprise.”
Spotify's Leadership Transition and Challenges
43:28 to 45:16
Understand the challenges faced by Spotify's new co-CEOs and their strategies to tackle competition.
“Meanwhile, let's take a look at shares of Spotify right now.”
Co-CEOs of Spotify: Background and Dynamics
45:16 to 46:47
Discover the backgrounds of Spotify's co-CEOs and how their partnership influences the company.
“is that they're now actually focusing on fitness content.”
Transcript
Automatic transcript. May contain errors.0:00Hello, I'm Stephen Carroll. I'm in Brussels where many of Europe's biggest decisions get made. And I'm Caroline Hepker in London with the hosts of the Bloomberg Daybreak Europe podcast. We're up early every weekday keeping an eye on what's happening across Europe and around the world. We do it early so the news is fresh, not recycled and so you know what actually matters as the day gets going. From Brussels, I'm following the politics, policy and the people shaping the European Union right now. And from London, I'm looking at what all that means for markets, money and the wider economy. We've got reporters across Europe and around the globe feeding in as stories break.
0:37So whether it's geopolitics, energy, tech or markets, you're hearing it while it happens. It's smart, calm and to the point. And it fits into your morning. You can find new episodes of the Bloomberg Daybreak Europe podcast by 7am in Dublin or 8am in Brussels, Berlin and Paris. On Apple, Spotify, YouTube or wherever you get your podcasts.
1:02Bloomberg Audio Studios. Podcasts. Radio. News.
1:11Bloomberg Tech is live from coast to coast with Caroline Hyde in New York and Ed Ludlow in San Francisco. This is Bloomberg Tech. Coming up, the U.S. Commerce Department moves closer to allowing NVIDIA to sell its H200 AI chips to China. Plus, robotics startup Skilled AI just closed a$1.4 billion funding round, tripling its valuation in seven months. We'll discuss with the CEO, Deepak Pathak. And Netflix is working to revise its bid for Warner Brothers Discovery, shifting to an all-cash offer. As you see, though, those names in the red and in general, the market on the downside. We've still got geopolitical anxiety.
1:52We've still got earnings coming thick and fast. And we've still got a last-up 100 off by 1.5%. Not so in the world of commodities. A search for safety, a bid as the dollar falls. We see 2.8 % higher on the Bitcoin rally. So maybe a bit of catch up with digital gold where we see spot gold and a new record high. Up 7.10%. Copper, all important commodity to our world of AI and data centers, Ed. We're at a new record high on an intraday basis. We're up 5.10 % just coming off of that high on the day. But Ed, it's notable and shift of this year. Right, rules and requirements are changing in the world of AI chips.
2:30The US Commerce Department out with clarification, which impacts NVIDIA and its H200 that it wants to ship to China, but also impacts AMD as well. AMD has been trading all over the place this morning, at one point notably higher in the session, but NVIDIA markedly down, now trading at its lowest level in around a month. What are these new rules and what do they mean and why is this significant? Bloomberg Senior Tech Editor Mike Shepard joins us out of DC. lay out the new rules and requirements for us. But what we're saying in the Bloomberg story is this marks a significant moment for the companies in their efforts to sell them to China.
3:02Why? Well, this is a small but critical bureaucratic step that paves the way for NVIDIA and AMD to be able to sell H200 and comparable AI chips to China. What it does is it now says that the US government will review requests for export licenses on a case-by-case basis and move away from the past policy of a presumption of denial of any such request. And that had in the past been really an effective export ban. Now, these companies can't be there expecting a rubber stamp for their requests. The U.S. government is laying out some pretty specific requirements. One, that the companies must show that the exports to China will not create a supply shortage here.
3:46This is addressing in a way some of the concerns that lawmakers here in Washington have been raising with attempts to pass their own version of export controls. Two, they must show that any manufacturing done for China must not displace any production capacity intended for U.S. customers here. And three, the companies must implement strict know-your-customer procedures to ensure that these chips and the related technology don't get out to unauthorized users. It's interesting, Mike, that we are seeing NVIDIA down on the day and AMD. do we get any sense of how many they might be able to ship? Because it's not without its intricacies, as you say, 50 % of the overall made in the U.S.
4:26Well, it is the question of the moment surrounding this rule, just how this cap that it lays out would function. The rule says that the exports to China can go no more than 50 % of total product here in the U.S. But does that mean H200s produced from this day going forward? If that's the case, it would not be a very big set, and it would mean a far smaller number of exports to China. And that would certainly not go anywhere near meeting the huge demand that Jensen Wang has said that he is getting from Chinese customers. And we have reported that Alibaba and ByteDance are interested in buying as many as 200 ,000 chips each.
5:11Now, if we are talking about all the H200s produced over all time, that would be a much bigger universe, and that would be a much happier outcome for NVIDIA, AMD, and their investors. So we will have to see exactly how that is interpreted, if that gets spelled out. Another big question, Carol, is how quickly the government can move to actually process these export requests. because we have been hearing over the past year and longer that they are simply running into capacity problems in the bureaucracy of commerce and being able to turn these around. Bottlenecks throughout the world of AI. Mike Shepard, thanks so much.
5:51Let's stick with what this means for NVIDIA and more broadly for semiconductors. Beth Kindig's with us, IO Fund lead tech analyst. So is this a$50 billion annual revenue opportunity? Is China that for NVIDIA? Hi, Caroline. And Jensen Wong has stated it's a$50 billion per year annual opportunity. When we left off, it was about$5 billion a quarter. That's$25 billion. So somewhere between$25 billion and$50 billion, which really means that as we look into analyst estimates, they are probably too low. And that has been something from the very beginning that we have found alpha in, is that analysts are really struggling to wrap their head around the opportunity of NVIDIA.
6:31And now here comes back the H200s. And where my firm is very focused is calendar year 2027. If you look at these estimates, they are too low as it is, let alone if China revenue resumes. What's really interesting, though, is China's response to all of this. You say it's too low how much might be the revenue for NVIDIA in China. But at the same time, we understand that deep seats driving new models, with seeing them use what they have in a different method that maybe Alibaba and Baidu don't need those 200 ,000 that they've already put efforts and offers in for. What do you make of the way in which China is going to pivot to domestic chips?
7:11I would argue that right now, the number I had was about 2 million H200s, and there's only about 700 ,000 available, which means that, once again, NVIDIA is oversubscribed. I have been asked before, who is the next NVIDIA? Is it something like Huawei? You know, is it Google's TPUs? And the answer is the next NVIDIA is NVIDIA. They are extremely hard to disrupt. And that goes for China or even United States competitors. Beth, taking into account the news story that Mike Shepard just brought us, Jensen Wong told us at CES that the$500 billion forecast for five fiscal quarters, it includes this one for Blackwell-Rubin, could get bigger because they could factor in some H200 sales in 26.
7:57I know 27 to a year, but have you modeled for that little update that Jensen Wong gave? Yes. So right now, the analyst estimates did actually catch up. They're about$320 billion this calendar year. Let's just go with 2026. I want to really emphasize that it's next calendar year that we're going to start to see analyst estimates too low. That is where the alpha is at right now. And it only takes one or two quarters for us to really see NVIDIA's price movement from all of the momentum we're seeing, not only from Blackwell Ultra, but then we've got Vera Rubin, second half of this year. Now we have the H200s coming back.
8:40Where that falls specifically in what quarter matters less to me. And what matters more is that those analyst estimates are far too low. Beth, the big story in global markets this morning is a blistering rally in metals, particularly copper. Copper, as S &P Global wrote January 8th, foundational to the expanding electricity supply that will power the data center build out. and it's now a big growth area in that market. How worried are you about that, copper in the supply chain, and how it relates to the power needs of the data centers that NVIDIA's chips go into? These are excellent questions, Ed.
9:19Across-the-board, resources, when it comes to this data center expansion, are extremely stretched, especially anything that supports the electrical grid, such as copper. What I would really emphasize is those alternative sources of energy is a great way to position going into 2026. One thing, too, my firm has dug up is that in order for us to get to Vera Rubin and then Rubin Ultra, power has to precede that. We have to really get this power to these data centers before those systems can really ship in volume and be installed. Things like Nat gas, even solid oxide fuel cells, those that do not need great dependency on the electrical grid is a wonderful way to position.
10:03And copper surging today is a reminder of that. What's interesting is we've hit NVIDIA. We're talking about the supply chain they're in. We're also getting the idea that CPUs is where it's at as well. And you just think about the moves that we've seen yesterday, today in Intel. I mean, where do you think the gluts or the bottlenecks are really going to be hitting hard? I mean, analysts, key banks saying Intel and AMD are basically sold out for 2026 on the CPU server side. Exactly. And what's even more interesting is that AMD was up more last year than NVIDIA. And they are such a strong CPU player as well as Intel.
10:38There are many bottlenecks, Caroline. Networking is another one that comes to mind. Of course, energy we just spoke about. Just keep in mind that it's not just GPUs, but there's almost an equal amount of money being spent on all of the components, all of the networking, all of the power. And we are no longer compute constrained. We're no longer GPU constrained. It really is becoming a power problem and a networking problem that needs to be solved, especially as we move into the inference market, which is only going to require more memory, more bandwidth, less latency, and more and more energy.
11:14Beth, do AMD and Intel just accept that it's the CPU market where they can do battle and just let the accelerator market go to NVIDIA and other specialists? No, I do not believe that will be the outcome. I think AMD is a strong contender, a strong number two, especially in the second half of this year. They've really been preparing to bring the rack scale architecture to market, which is called Helios. And Ed, what was, I've covered AMD very thoroughly and we could get into the specs all day long. But OpenAI giving them that six gigawatt deal is a massive nod. I mean, that is a big nod that AMD is a contender from the leading R &D firm.
11:57Helios AMD's first rack scale solution, but theirs in the world's first two nanometer chip inside, which they talked up in our interview last week. Beth Kindig of IO Fund, great to have you back on Bloomberg Tech. Thank you very much. Now coming up, anger grows over sexually explicit, non-consensual AI generated images on X. We have the latest. This is Bloomberg Tech.
12:29The Supreme Court still hasn't ruled on challenges to President Trump's tariffs, leaving the world to wait until at least next week to learn the fate of his signature economic policy. No-Mex surveillance co-host Anne-Marie Horton joins us right here for more. So we delay once again, and that timing could be awkward. It could be awkward. So the Supreme Court punts it once again. They do not come out with a decision on whether or not the majority of actually a lot of this tariff revenue we're bringing in into the United States, more than 70%, is tethered to the president using IEPA, very flexible, a blunt tool that he has used since he came back into office.
13:04And why it could be so awkward is because the president will be in Davos next week, and he's going to be meeting with global leaders, the world elite. Last year, he joined on remote, really lambasted the global elite. And at this point, potentially, we could get the decision next Tuesday or Wednesday, because that's when the justices will be meeting again. That's when potentially we can get the decision. And he could potentially be facing a Supreme Court saying you cannot use that legal authority while he's talking to all these countries that he already did trade deals with. Bloomberg's Ann Marie Horden, thank you very much.
13:39Another top story. The U.S. Senate unanimously passed a bill allowing victims of sexually explicit AI images to sue the creators of that content. The move is in response to growing public anger after Bloomberg reported that Elon Musk's X has become a top site for generating pictures of people who have been non-consensually undressed by AI. Musk posted on X that he wasn't aware of any naked, underage images generated by Grok. That's XAI's generative AI tool. Let's bring in Bloomberg's Emily Birnbaum, who covers corporate lobbying government. The Defiance Act confers on American citizens a civil right to sue.
14:20Take that information and give the rest of the details, because that's the bit that's new here. Yes, that's exactly what it does. It essentially empowers victims of this non-consensual deep fake technology, people who have been non-consensually undressed by this AI tool, and it allows them to sue the people responsible for posting that. And it's a companion to a bill that was signed into law last year by President Trump that creates criminal penalties for distributing that kind of content. But it didn't go through the House last time, the Defiance Act, whereas Takedown did. And why would this time be any different?
15:03Is the pushback becoming enough? Because thus far, all we've really seen is the Department of Defense put Grok AI within its military system instead of pushing back on this being something that Grok or X needs to clamp down in some way. that's true there definitely is more of an appetite in congress to pass this legislation there have been six new co-sponsors that have come on board since the beginning of the year i think that the issue of non-consensual ai generated images has become so vivid to lawmakers many of them like alexandria ocasio-cortez are themselves the victims of this so i think the The appetite has really broadened this year.
15:49Emily, where does the company X and the platform X fit into this in a citizen's right to sue and those that distribute the content? So this, the Defiance Act, which passed the Senate this week, that probably would not impact Grok. That impacts the people who are responsible for distributing the images. But there is a lot of conversation right now about whether the Take It Down Act would empower attorneys general to sue Grok, XAI, Elon Musk over some level of responsibility for thousands of these images proliferating every hour. So far, no attorneys general have taken that up, but it's definitely under serious consideration.
16:35Certainly something that maybe Malaysia has been putting forward about whether or not it's XAI's responsibility in some way. Bloomberg's Emily Birnbaum, fantastic reporting. Thank you very much indeed. Now coming up, Airbnb has a new chief technology officer. We'll hear about his plans for the short-term rental company. That's next. This is Bloomberg Tech.
16:58I'm Barry Ritholtz, inviting you to join me for the Masters in Business podcast. Every week, we bring you fascinating conversations with the people who shape markets, investing, and business. CEOs, fund managers, billionaires, Nobel laureates, traders, analysts, economists, everybody that affects what's going on in the market, whether you own stocks, bonds, real estate, commodities, crypto, you really need to hear these conversations. Sometimes it's behaviorists like Dick Thaler or Bob Schiller. Sometimes it's fund managers like Peter Lynch, Bill Miller, Ray Dalio. Sometimes it's authors, Michael Lewis, author of The Big Short, and Moneyball.
17:44Regardless of the conversation, these are the folks that move markets each week. That's the Masters in Business podcast with me, Barry Ritholtz. Listen on Apple, Spotify, or wherever you get your podcasts.
18:06I'm extremely excited, like Brian said, to bring craft to the art of deploying AI first features and products. So I hope to be able to surprise the world with really beautiful features that they're going to love using every day. That was Airbnb's new CTO, Armand Aldale, weighing in on the short-term rental platform's AI efforts. Let's get more with our own Latterly-Lung, who spoke with Aldale and Airbnb CEO Brian Chesky. I think an interesting way to start is what is Airbnb trying to solve for here by bringing in a new CTO? What is it they want to change about the platform? So after a few years of sort of working on the plumbing of Airbnb's app, like revamping the app, they are ready to take it forward, adding AI across the app, including customer service, which they did last year.
18:52And now they're going to do more with search as well, like letting people discover listings and also cross market some of the new services like tours and experiences going forward. And I think you sort of put that exact question to Branceschi himself. Let's just take a listen to how he responded as to why they're pivoting and focusing all in on AI at this moment. Take a listen. We are the beginning of this incredibly exciting technological transformation with AI. It's going to be a journey. And I couldn't be more excited about Ahmed because he's one of the leading AI experts in the world. but he also brings a sense of craft to Airbnb, something we're really known for.
19:26And I think what you're going to see in Airbnb is not only AI search, but a very personalized AI experience where I think we're going to bring great design sensibility into the experience of Airbnb. Let's talk a little bit about how they're talking of the narrative of change, because they've said farewell to the previous CTO, who's there for about seven years, and brought in this expert from Meta. So who do they say farewell to and why is this meta executive the best place? So Ari Barlow, the previous CTO, came in during a time when Airbnb was going through hyper growth. That was more than seven years ago.
20:03And now Airbnb is sort of, as Chesky said, going through a technological revolution of LLMs and chat agents. And now Airbnb sort of wants to use sort of the unique data they have on their platform to differentiate itself from other, you know, chatbots to provide better recommendations to customers on their platform. Airbnb, the company whose stock rose about 3 % last year, it's down a couple of percent so far this year. And Chesky was one of those people that was the early vocal behind founder mode. What's the health of Airbnb right now? Is this them saying we've got to do something to reaccelerate?
20:40Yeah, Brian was not satisfied with some of the below 10 % growth rate we've seen last year. And so he's been very active in wanting to pilot more businesses for this year. They're piloting boutique hotels in several cities. And they're trying to add grocery delivery for people to pre-stock their kitchens before they check in. And so they're trying to get their growth rate back up as sort of the core rental service matures. Natalie, you've also got a story about how the chief information officer has departed after the CTO shakeup. Is there more executive leadership changes to come? Yes, so they're still looking for the new CIO.
21:19We can expect that to come in the months and the year ahead. Thanks, Natalie Leung. Thank you very much indeed. Let's turn our attention to entertainment now. Netflix is working to revise its bid for Warner Brothers Discovery, potentially shifting to an all-cash offer according to sources. Now, the move is aimed at speeding up a deal that, look, has faced pushback from Washington and competition from rival bitter Paramount Skydance. For more, let's get to Bloomberg's Lucas Shaw, who covers media and entertainment. Hit this story hard. And Lucas, I mean, the all-cash narrative, why is that important?
21:53Why would that be the winning formula for Netflix? Because investors like cash now versus, you know, stock and Netflix that they could sell later, right? One of the knocks against the Netflix deal, as perpetuated at least by Paramount's guidance, the rival bidder, has been that Paramount's deal is all cash and that Netflix deal contains between$4 and$5 a share of stock. Now, Netflix stock is pretty valuable, something that you think people would want to have. But most people would rather just take the cash now and get paid. And so that, I think, Netflix hopes will accelerate on many fronts. It'll assuage any shareholder concerns about their bid and also eliminate any potential complexities as they continue to negotiate and pursue it.
22:38Lucas, when we were reporting this out, the Netflix response and the tit for tat, Paramount suing earlier this week and nominated our heads to the board. Do you get the sense that Netflix think this will work, that they now are back in sort of the lead in a tight race for the assets? Yes. Well, I think if you were to talk to the people at Netflix, they've never felt they were out of the lead, right? They are obviously facing a stiff challenge from Paramount, but they're the ones who have an approved deal. And they have yet to see kind of some – they have yet to see shareholders say en masse, we don't want the Netflix deal.
23:16They have yet to see the federal government say we're going to block this deal. There are certainly some kind of blinking yellow lights, but I don't think they've seen anything yet that makes them too concerned. They seem still pretty confident that they'll get this done. And in a week, January 21st, that's when the tender offer for certainly the Paramount Skydance side of the equation runs outright. I mean, do we get any signal for investors that they're at all interested in that side? You know, investors have been split. But some of the big investment funds, they don't say anything, but we have spoken with a number of the major shareholders.
23:50And I'd say their reactions range from we prefer Netflix to we prefer Paramount. And then a lot of people in the middle are basically saying, you know, tie right now goes to the winner, which is Netflix. But we hope slash expect that Paramount will increase its offer, which doesn't seem to be the Paramount move. Just very quickly remind us of the state of play, the difference between the two bids. Well, Netflix's bid is for just the studio and streaming service, so Warner Brothers and HBO. Paramount's bid is for the whole company, so that includes all the cable networks like CNN, TNT, TBS, and the like.
24:25Bloomberg's Lucas Shaw, who leads the team at Screen Time that has led the way on reporting this story. Thank you so much. Now, coming up on Bloomberg Tech, robotics startup, skilled AI, just closed a$1.4 billion funding round, a big valuation, tripling from a year. But this is next. This is Bloomberg Tech.
Read the full transcript
25:04Welcome back to Bloomberg Tech. Let's check in on the markets that are under pressure. Look, we've got a wall of worry as we come to the earnings season. geopolitical risks. And we see some drag down in big tech. We're up by 1.3 % on the NASDAQ. In fact, two straight days of losses on the S &P. We're coming off of record highs. NVIDIA, of course, confronting that news that maybe H200s could be getting closer to exporting to China. But at the moment, we're under pressure. It's one of the key drags on the downside. Bitcoin, though, up 3.4%. We're seeing maybe a bit of love for digital gold, not just gold, silver, and copper.
25:32We're also seeing maybe some shorts getting squeezed out. Move on, have a look at what's happening in individual stocks. I'm shining a light on Honeywell. Look, we're up 1.8%. Maybe an asset that it controls, Quantinum, could be looking for an IPO. We're understanding that they're potentially filing for an IPO with the SEC. We know that its last valuation was about $11 billion. We've got other key strategics, NVIDIA's in there, GP Morgan's in there as well. So all eyes on quantum computing as we look towards 2026. What are you looking at? Yeah, we also have a big deal in private markets. Robotics startup Skilled AI has just closed a$1.4 billion funding round.
26:07The valuation, more than$14 billion. And that's for a company only founded in 2023. Skilled is aiming to stand out in what's a crowded field by building an AI powered robotic brain. The goal is to develop a system that can adapt across environments and form factors and crucially learn from its own mistakes. Joining us to discuss is Skilled co-founder and CEO, Deepak Pathak. It's great to have you on Bloomberg Tech. What's interesting, let's start with the round. It's not just the volume. It's not just the valuation. Some of the strategics that are backing you, NVIDIA, others in consumer electronics, Samsung LG, and in the software sector as well, what are they recognizing as your core competence, do you think?
26:52So if you think about robotics, right, it's one of the oldest areas in technology, like in AI. AI was built for robotics. We have seen amazing demos in robotics for the last 70 years, not 4 or 5 years, 70 years, but where are robots around us? And this is what people often get wrong. When you think of robots in your mind, you imagine hardware. That comes upon your mind first because that's how Hollywood has primed us. But the main thing behind not having robots around us today is the brain is missing. And if you have the brain for robots, you can really enable them around in a variety of areas, variety of applications.
27:29And that's what exactly Scaled is way. We are building what we call omni-bodied brain. Any robot, any task, one brain. I'm interested in how you're building it. The limitation that a lot of people are talking about right now is real-world data and training. There are different approaches. Digital twins, trying to simulate physics in digital worlds. This is very much what NVIDIA is focused on, right? Is that the approach you've taken, simulated or synthetic data? So, I think I will draw analogy to LLMs, like chat GPT-like models. like if you see OpenHeadChat GPT. And soon after, many companies had their own LLM models.
28:04And the reason for that is data exists on the Internet for language or vision. But robotics, there is no Internet of robotics today. So what we do is we look for alternate sources. One such source is human videos. We look at how humans operate in their scenarios, in your kitchen, while working in factories, and combine that with training from simulation, which companies like NVIDIA invest a lot on, like Omniverse and all those. So these two things, watching people and then practicing in simulation, that's how we combine and scale these models to large scale. Your software can be put within the GPU.
28:43And what's interesting is you're already going from what we understand, zero to tens of millions of dollars, Deepak, in revenue in just a few months in 2025. Where's that revenue coming from? Who is using your software already? Yeah. So the revenue has mostly been from the enterprise applications. And the idea is, instead of going for, our end goal is a consumer home-like application. But right now, since it's the beginning of the field, we are taking this model and we are deploying it in a variety of applications. Like we are deploying robots in sectors like point-to-point delivery applications, security applications, data centers, and manufacturing and warehouse.
29:32So the revenue is coming from most of these applications. You were born out of academia and you're still Professor Carnegie Mellon. But what's interesting is you're not without some competition out there. I think about physical intelligence, for example. You're all trying to attempt to get robots to become intelligent and situationally aware. Do you think you can take on? Are you worried about competition? Well, competition is always good. And it turns out the AI community, born from very small circles, so they're all our friends and colleagues from earlier. But what's unique about what we are building is this omnibodied brain.
30:08Like any robot, any task, one brain. Now, if you really think deeply about this, it sounds absurd. Like a humanoid robot with a human-like body, a dog robot, and they share the same brain. And that's what you see in the results we release. The difference in approach is fascinating. It is a crowded field. You know, Burnt from 1X was on the show yesterday. They have a world model that essentially allows Neo to carry out a task it's never seen before. and that largely you know i'm not an engineer right but what you all have in common is vlm those are the inputs distinguish your your approach to one x's as an example yeah so in our case uh the robot learns by watching people right now imagine i have to pick up this cup in front of me do i have to exactly imagine scary by the way watching me well you learn when you are a kid you learn by watching your parents you learn by watching others now if you think about how you operate every time you pick up a cup do you imagine in your head a picture of a cup and then putting your finger in the handle no, it's all merged in it's consciousness inside your body and then you imagine in some abstract space and then do it and that is the main idea here that's how we do it but just like being said watching alone is not enough because if it was enough I could play like Federer I keep watching his games all day and that's where practice comes into play.
31:35Now, practicing in the real world, making mistakes is expensive and that's where we combine simulation. So videos alone is not enough, sim alone is not enough, but put them together and boom, you have a magic recipe to really figure out how to scale robotics. A magic recipe that values$14 billion thus far, Deepak Pathak, co-founder, CEO, Skilled AI. Thanks so much for joining us today. Coming up, more funding round news, Accel partner Sarah Ittleston joins us to talk about her latest investment in AI security startup, Depth First. This is Bloomberg Tech.
32:31all over global business, finance, tech news, all as it is happening in real time. And we've got complete coverage of the US market close. Gotta say, basically, if it impacts financial markets, if it impacts companies, if it's impacting trends and narratives that are out there, we are on it. We also have a lot of fun doing it. Bloomberg Business Week also brings you the analysis behind the headlines through conversations with our expert guests. And we are doing this all live each weekday. And then we bring you the best analysis in our daily podcast. Search for Bloomberg Business Week on YouTube, Apple, Spotify, or anywhere else you listen.
33:03Check it out on your way home from work to catch up on the conversations that you miss during the business day. And on the weekend, check it out for a complete wrap-up of your business week. That's the Bloomberg Business Week Daily Podcast. I'm Carol Masser. And I'm Tim Stenevec. Subscribe today wherever you get your podcasts.
33:26security startup debt first has closed a 40 million dollar series a funding round the company uses ai to identify and harden software vulnerabilities the round led by excel joining us to discuss excel partner sarah illison debt first this was a big theme actually at aws reinvent the idea that particularly sort of swarm based agents can go on the offense, not just defense. So a sensible place to start with why you made the investment in general security intelligence is what can it do that existing AppSec and other sort of static analysis tools just can't? I think the main thing is performance and speed and the ratio of signal to noise.
34:07So I think what we have to realize in this moment is as we're all so excited about how AI is making it possible to ship software so much more quickly, that same AI tool is being used by the bad guys to ship attacks more quickly. And so the historical ratio of signal to noise just won't stand in the current era. And so the security platform that Debt First has made is AI native. And what they're doing is continuously, agentically scanning a company's code. They're looking at its architecture. They're looking at its business logic. And they're finding verified vulnerabilities eight times more than the previous best in class.
34:45And I think the key there is verified vulnerabilities. They're not waving their hands about things that aren't really a problem. They're finding real problems. And then they're also serving up a solution. You can press one button and say, fix it. And folks are doing that in mass. I was looking at the round. It's a sizable series A. In the last couple of years, the market has been somewhat distorted, depending on the company's type. But what you just said, eight times best in class, how critical is a data point like that to you when you decide, do I want to be in this round at all or not? Evidence.
35:21It's really critical. And I think particularly in the case of security, there's just a societal imperative that we get this right. And I think Excel, we've been investing for 43 years. Throughout that time, security has been a key category of investment for us, folks like CrowdStrike or Netscope that IPO'd just last year. And what we've seen is in these major innovation transitions, these major cycles of transition, the security apparatus has to be replaced. It has to evolve to that new era. And that's why we're so excited about companies like depth first, but you're exactly right. It has to work.
36:00It's imperative that it work. And so those stats are really important. They're important to the CISOs. They're important to the Fortune 50 that are evaluating these companies. And so if we're going to put ourselves and our funds behind a company, it's imperative that we believe that the thing works and not just today, but that you have the team that can drive over this next era of transition, that you have the talent that can sort of weather this moment. The talent is highly sought after. And you're seeing Palo Alto Networks, even ServiceNow making some big acquisitions in the space, trying to get ahead of the cyber risks around AI.
36:40Do you think that this is a long-term bet? Will this company go public? Or do you think it will be an M &A story? You know, I think this is absolutely a long-term bet. That is always our focus when we do our investing. And I think that this is a team that's really uniquely qualified to go that distance. So this is technical leaders from companies like Databricks, from Google DeepMind, from FAIR. These are folks who have built large, durable companies and confronted these security problems at an incredible scale. But marrying that cyber experience with the AI experience, very few people have had sort of the experience of building at a place like Google DeepMind and can now apply that to this context.
37:29So it's a really interdisciplinary team, a highly technical team that we believe is very unique in their ability to tackle this problem. Sarah, you were head of strategic partnerships at FAIR. So when you're thinking about who are the talent you want to go for, is it your own network you lean upon? Is it that you're hearing these companies being born out of these big deep minds even before they're really created? How are you tapping into a really early stage business that needs financing? Yeah, I think it's all of the above. And then, you know, we hope to spend a lot of time with folks during the early days when they're really, you know, considering starting a company or the early definition of the company.
38:11You know, our goal is to be along the journey for decades, right? The generational companies that are going to solve these security problems will last that long. And so we do. We try and find them early, and it can be through a lot of things. But certainly one of the greatest signals is when we see incredibly talented people that we know to be excellent aggregating around an opportunity. And I think what really emerged about this team is that they're incredibly motivated by this mission. And if you look around, all of our critical systems in society are powered by software, software that people can write increasingly quickly with AI.
38:49And the other dimension that's changing is increasingly we're giving this software more autonomy. We're giving it more access. We're giving it the ability to reason. And that's really exciting. But it's also something that would be concerning if those things aren't secure. We just watched your segment talking about the robots, right? We want to make sure if we're giving these devices access to our homes, to our businesses, that they're secure. And I think that that is a belief that's shared by the Depth First team, that this is sort of an imperative for society that we get this right, that we ensure that all of this new AI-generated code that's being shipped at increasing rates is all secure.
39:30What's being shipped by a load of your portfolio companies? Cursor, Lovable, Anthropic, among some of the bets over at Accel. We really appreciate your time, Accel partner, Sarah Ittleston. Now coming up, Spotify's new co-CEOs. They face a host of challenges. More on the audio giant's plans next. This is Bloomberg Tech.
39:53Salesforce is looking to bolster AI capabilities for Slack users. The company is releasing an upgraded Slack bot that uses Anthropics Claude model. Joining us to discuss Rob Seaman, Slack interim CEO and chief product officer. You know, I have not yet got access to this updated Slack bot, but as a Slack user, very interested to see where this goes, Rob. The general idea, right, was that Mark Benioff, the parent company's CEO, was using ChatGPT. And you guys are like, we want to have an internal tool. So has it worked? Is he using it for? What are you using it for? Absolutely. So one, I can't wait for you to try it.
40:33So we're using it internally. Mark is certainly using it. I'm using it. I'll give you a few examples of how I use it. I used it recently. I just said, hey, go find the most recent all hands deck that we're working on. One of the things that we do is actually introduce every single new hire every month. And I said, tell me how to pronounce the names of all the new hires. And that goes off and finds the presentation without me saying, reads the presentation, figures out where the new hires are, loops through all of them and figures out their highest probability nationality and tells me how to pronounce them.
41:04I don't have to go bug people to figure that out. It's awesome. It saves me time. In the first instance, it's underpinned by Anthropics Claude, right? And you're going to tell me, well, this is interim and we're going to look at all kinds of different model options. We're model agnostic, but it is an interesting and the latest example of how Claude helps people in the enterprise get to market quick. You know, what was your experience of that, Rob? The biggest thing for us was being able to take an LLM, pick it up and set it down in our infrastructure in a way that our customers' data never left our security boundary and wasn't used for the training of the model.
41:36And that's what we got with Anthropic and Cloud. And on top of that, we've done a bunch of sophisticated prompt and context engineering that makes Slack bot deeply personal and a deeply personal agent to every single person that uses it. Robert, how are you thinking about security? I mean, it's paramount for us. It's been something that's been the very, very top of our priorities since the foundation of Slack. And so one, Slackbot is a personal agent. It needs to be deeply trusted by the user, but also deeply trusted by the enterprise. And so what we do is the user, the Slackbot is acting on behalf of the user.
42:13So it can access the information that a user can, their direct messages, their private channels, and the public channels within a particular Slack instance. And as I mentioned before, from a company perspective, you can rest assured that your data is not being used to train the model, nor is it leaving the security boundary. So it's all adhering to your existing security policies in Slack. This is all about return on AI investment. It's all about making it clear to me, to you, to the users that this stuff saves you time, makes you more productive. How are you measuring that role? We don't measure individual productivity, but we do measure the usage of Slackbot.
42:50So we measure the penetration. So the number of users that actually use Slackbot over the number of users that have access to it, but week-over-week retention and then message intensity. And that allows us to approximate the value that's being created. And the week-over-week retention for Slackbot is higher than any other feature we've ever had in Slack. And the message intensity keeps growing. And to us, that's a quantitative signal of qualitative value that's being derived from our users for time savings. But also we're seeing, like, through our qualitative surveys, people use this for brainstorming and creative pursuits that we didn't even expect.
43:22Rob Seaman. Slack interim CEO, been great speaking with you about Slackbot. We'll see how it catches on. Meanwhile, let's take a look at shares of Spotify right now. And indeed, over the start of the year. Well, off by 9%. That's, of course, when the new co-CEOs have joined the company. Gustav Solderstrom, Alex Nordstrom took over as from the Spotify founder, Daniel Ek. Now, they've both been at the business for a long time, but Ek himself grew the platform from money suck It's one of the most powerful businesses in the music industry. And the new CEOs have to steer that powerful ship around some pretty tough headwinds.
43:58AI, angry creators, user fatigue to turn around that share drop. Bloomberg's actually come and has been writing about all of this, about the audio giant's leadership transition. It's in Businessweek. And there are a lot of headwinds. What is front and center for them? How are they taking on the algos and the AI story? Right. They're coming with a lot of different products that they're hoping will kind of buy them some time and also maybe address some of what users want to do. So for example, they're integrating Spotify into JackChipity. They've made a partnership with Meta on its glasses. They are letting you enter prompts to get a playlist in return using even like online context so that you can get the top rated albums of the year from Pitchfork or something like that.
44:43So they really want to give people both this personalization, more options, showing up where people are, podcasts are coming to Netflix. They're really trying to meet people where they are rather than forcing them to just come to Spotify. YouTube, not a new issue, but the biggest in front of them, it seems like. You know, it's very detailed in the Business Week story, this competition for eyeballs. And YouTube is the thing on both of those CEOs' desks right now, Ashley. Yes, YouTube, Spotify has moved into video. One of the scoops from our story is that they're now actually focusing on fitness content.
45:20So as a result of them pushing into video podcasts, creators who we see on YouTube that actually put out guided workouts have started uploading their content to Spotify and Spotify is seeing that people really enjoy that content. So instead of just using playlists to soundtrack their runs, they're actually watching yoga on Spotify, for example. So this is an area they're pretty excited about and seem to be investing in. We know a lot about Daniel Ek. We've sort of grown up alongside him as he's taken this European success story global. What do we know about the co-CEOs? Who are they? What are their descendants?
45:54What are they about? So both of them are Swedish. Alex is of half Chinese, half Swedish descent. His mother was a single mother. She ran a Chinese restaurant growing up. He's really kind of the culture guy. He gets a playlist regularly that keeps him on top of trends. It seems like he's really sort of the social person who's doing all these partnership deals with the labels and audiobooks and such. Gustav, formerly really focused on product, now co-CEO, also Swedish. He sold a startup to Yahoo back in 2006. He did a stint in the Swedish military. And he's really tech philosophy focused. He was really excited about a book he read from the inventor of quantum computing.
46:31So he really gets deep on that. And supposedly the two of them together are really like a yin and yang. They go well together. A fun tidbit. Nordstrom means North Stream. Soderstrom means South Stream. So it's written in the universe, I guess. This really is a must read on Bloomberg online or in a magazine coming to you very soon in Business Week. Bloomberg is actually Carmen with the deep dive on Spotify. Thank you very much. Caro, that does it for another edition of Bloomberg Tech. Markets still continuing to start the year. Pretty hot. pretty hot. Well, we're under pressure if you're looking at big tech, but they're hot if you're looking at commodities.
47:07And that too is an AI story. I mean, the joke of 2022 and onwards, everything's an AI story. Don't forget to check out all of them on our podcast. Find it on The Terminal as well as online on Apple, on Spotify, and on iHeart. This is Bloomberg.
47:33Every weekday, we bring you the top headlines from the world's biggest tech companies. From finance to defence, AI to entertainment, and from startups to the magnificent seven. We highlight the latest stories of the people and companies pushing the tech sector to new frontiers and the politics that shape global tech markets. We do this all every weekday, then bring you the most important conversations and analysis in our podcast. Search for Bloomberg Tech on YouTube, Apple, Spotify, or anywhere else you listen. Join us every afternoon on your commute home and stay ahead of the tech news cycle.
48:05That's the Bloomberg Tech Podcast. I'm Caroline Hyde in New York. And I'm Ed Ludlow in San Francisco. Subscribe today, wherever you get your podcasts.
From the publisher
Bloomberg’s Caroline Hyde and Ed Ludlow discuss the US Commerce Department’s move making it easier for Nvidia to sell H200 AI chips to China. Plus, fresh off its $1.4 billion founding round, Skild AI CEO Deepak Pathak discusses how the company’s robot “brain” works across form factors. And, Netflix is working to revise its bid for Warner Bros. Discovery, shifting to an all-cash offer.
See omnystudio.com/listener for privacy information.

