Nvidia Earnings This Week; Biggest Power Deal in History

18 May 2026 · 44 min · 27 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

Bloomberg Tech episode focused on Nvidia’s upcoming earnings, AI-driven power and compute infrastructure deals, and broader market/geopolitical impacts on tech stocks. It also covers SpaceX IPO plans, a new compute-power futures market, Apple’s privacy-focused Siri update, and AI compute access issues at Google.

Guests (and backgrounds)

  • Mandeep Singh (Bloomberg Intelligence): tracks semiconductors/AI infrastructure; analyzes Nvidia earnings drivers.
  • Liana Baker (Bloomberg/M&A): leads M&A coverage; discusses NextEra–Dominion.
  • Carmen Lee (Silicon Data / Compute Exchange CEO): runs compute spot/forward markets; launches compute futures with CME.
  • Mark Gurman (Bloomberg Apple editor): reports on Apple privacy and Siri.
  • Peter Elstrom (Bloomberg global tech editor): covers Samsung labor/memory market tensions.
  • Julia Love (Bloomberg Alphabet reporter): covers Google AI/compute allocation and researchers leaving.
  • Richard Sosha (Recursive CEO): AI lab building self-improving systems; ex-OpenAI/DeepMind ecosystem connections.
  • Dan Roberts (Iren co-founder/co-CEO): AI cloud/data center operator; builds power-constrained compute capacity.
  • Hemmer Palmer (Bloomberg): breaks down hedge-fund 13F filings.

Key claims + examples

  • Nvidia: strong EPS/top-line growth, but rising memory (HBM) prices may pressure mid-70% gross margins; shift from training to inference/agents could affect share (Anthropic/coding agents; Grok acquisition).
  • Geopolitics: US-China H200 supply-chain uncertainty after statements about China not wanting H200s.
  • Power deal: NextEra acquiring Dominion for $67B, framed as necessary for AI data center power buildout; regulatory approval expected within 12–18 months; consumer pricing credits mentioned.
  • Compute futures: CME + Silicon Data launching futures for AI computing power; compute treated like a hedgeable commodity; GPU indices (ex-China) used for normalization.
  • Apple Siri: auto-deletion options (30 days/1 year) and shorter “memory” retention via differential privacy.
  • Samsung: potential 18-day union strike could disrupt AI memory supply; AI industry depends on memory pricing.
  • Google: researchers leaving due to tighter compute access/prioritization for Gemini and paying cloud customers.
  • Recursive: aims for closed-loop, self-improving AI knowledge discovery; emphasizes safety via red-teaming.
  • Iren: biggest bottleneck is power/physical infrastructure timeline (steel/concrete/kilowatts), not software; partnership with Dell/Nvidia for gigawatt-scale compute.

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

Market Overview and Tech Insights

2:25 to 4:10

An overview of current market conditions and implications for tech stocks.

“For a second straight day, the Nasdaq 100 under pressure on Friday.”

NVIDIA's Upcoming Earnings

4:10 to 5:44

Discussing NVIDIA's earnings expectations and the AI landscape.

“just on the day, Ed, or a second straight day.”

Insights from Mandeep Singh on NVIDIA

5:44 to 8:20

Analyzing NVIDIA's growth and challenges ahead of earnings.

“But as we think ahead to Wednesday, NVIDIA's earnings out after the bell, I want to bring you what to expect with Mandeep Singh of Bloomberg Intelligence.”

Elon Musk's Plans for SpaceX IPO

8:20 to 11:40

Exploring Elon Musk's comments on the upcoming SpaceX IPO.

“Bloomberg Intelligence's Mandeep Singh, thank you very much.”

AI Data Center's Impact on Power Sector

11:40 to 14:02

Understanding the largest power deal in history driven by AI demand.

“Look, we talk about the need for energy within AI, but compute, of course, is officially becoming a financial asset class that we all care about too.”

Understanding the Compute Market Dynamics

14:02 to 14:44

Learn about the demand for GPUs and the impact of pricing volatility.

“so AI startups all the way to the new clouds.”

Transparency in Compute Access

14:44 to 15:55

Discover the challenges of transparency in accessing compute resources.

“I mean, for full transparency for our audience, you used to be part of the Bloomberg team.”

Pricing Trends and Supply-Demand Shifts

15:55 to 17:18

Explore current trends in compute pricing and their implications.

“What is it that you're seeing in your data about just the cost of compute at the moment?”

CME Collaboration and Regulatory Considerations

17:18 to 17:46

Understand the collaboration with CME and regulatory challenges ahead.

“We normalize the geolocation to one data point.”

Baidu and AI Business Shifts

20:43 to 21:46

Examine Baidu's strategic pivot towards AI and automated agents.

“And first up, Baidu's 1 % drop in sales was less than analysts had expected, void by its new AI businesses.”
Show all 27 chapters

Apple's Siri and Enhanced Privacy Features

21:46 to 24:18

Learn about Apple's privacy-focused updates to Siri and their significance.

“You bring us the scoops and push us towards, for many reasons, Apple wants to lean in on its privacy advantages.”

MIAX Exchange Growth and Retail Trading

24:18 to 26:11

Discover the growth of MIAX and its focus on retail trading opportunities.

“but not have it glued exactly to the user.”

Market Trends and Economic Sentiment

26:11 to 27:08

Explore current market trends and the impact of geopolitical factors.

“Nora, Melinda, thanks for bringing us and insights from your conversation.”

Samsung's Labor Relations and Memory Chip Market

27:08 to 28:00

Understand Samsung's labor challenges amid rising memory chip prices.

“Look, we've got a little bit of risk aversion creeping into the market, and it's largely geopolitically focused.”

Samsung's Labor Struggles Amid Profit Boom

28:00 to 29:04

Learn about Samsung's union tensions and the implications for the AI industry.

“But look, if we look across to Asia, not all was bearish.”

Debating AI Profits in South Korea

29:04 to 30:18

Explore the societal discussions in Korea regarding AI profits distribution.

“And just to set the scene, it is quite unusual for Samsung to have this kind of conflict with their union.”

Google's Leadership Challenges in AI

30:18 to 31:32

Understand the struggle within Google for accessing computing power in AI research.

“And also the government is beginning to ask this very important question.”

The Impact of Compute Shortages on AI Projects

31:32 to 33:05

Discover how compute shortages are affecting AI researchers' projects at Google.

“For more, Bloomberg's Julia Love, who covers Alphabet, joins us now.”

Recursive's Vision for Self-Improving AI

33:05 to 34:28

Learn about Recursive's approach to building self-improving AI and its market strategy.

“Yes, I think there's a few things happening at once.”

The Safety Concerns of AI Development

34:28 to 35:39

Understand the safety measures Recursive is considering in AI development.

“So we want to build recursive, self-improving superintelligence that automates knowledge discovery.”

Balancing Multiple Roles in AI Startups

35:39 to 37:04

Explore the challenges of managing multiple ventures in the AI landscape.

“And whenever in AI research we replace manual systems with learned systems, we see a massive improvement.”

Iron's Approach to AI Infrastructure

41:00 to 42:00

Learn about Iron's strategy for building AI infrastructure using renewable energy.

“Our own Ed Ludlow is standing by with Dell Technologies World Event upon you, Ed.”

The Compute Power Challenge

42:00 to 42:55

Explore the challenges of meeting the growing demand for AI compute power.

“but that's a small point in the grand scheme of things.”

Building Infrastructure for AI

42:55 to 44:28

Learn about the barriers to building data centers and AI compute facilities.

“Dan, I think I'm right in saying, just as with everyone else, you're in a position where demand for compute, your compute, is vastly outpacing your ability to supply it.”

Collaboration with Tech Giants

44:28 to 45:24

Discuss the importance of partnerships with companies like Dell and NVIDIA.

“I'd love to go back to basics for a second.”

Global Project Management in Tech

45:24 to 46:30

Understand the complexities of managing tech projects across different countries.

“Look, every geography is slightly different in terms of how they handle things, but the majority of our capacity is in Texas and they've been pretty good.”

Analyzing Hedge Fund Moves in Tech

46:54 to 48:41

Get insights into hedge fund repositioning around big tech and AI stocks.

“They're revealing how hedge funds are repositioning around big tech and the AI trade.”
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.

0:00The right technology can strengthen human judgment. That's why Deloitte brings together AI and data analytics with multidisciplinary teams who can help you connect the dots across your enterprise. From risk to operations to customer needs. So opportunities don't slip by and surprises don't spread. Because the smarter your systems, the sharper your instincts. That's how technology makes people better at what they do best. Deloitte. Together makes progress. Learn more at Deloitte.com slash Together Makes Progress.

0:58on Health AI gets you the right care fast. Healthcare just got less painful. Travel smarter, not harder at America's Best Value In by Sonesta with convenient locations from coast to coast and value pack comfort at every turn. It's a practical choice for road trips, quick getaways and everyday travel that keeps things simple without sacrificing comfort. And when you're a Sonesta Travel Pass member, staying at America's Best Value In means earning points toward free nights, upgrades, and more every time you stay. Go to Sonesta.com to book your stay and unlock the best rates with Sonesta Travel Pass.

1:31Here today, roam tomorrow. Join now at Sonesta.com. Terms and conditions apply. Bloomberg Audio Studios. Podcasts, radio, news.

1:49Bloomberg 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, all eyes on NVIDIA this week as investors anxiously await the company's earnings results. Plus, the AI data center boom triggers the largest power deal in history, with NextEra acquiring Dominion Energy for$67 billion. And Elon Musk says he's in Texas working on plans for the SpaceX IPO. The filing could come as soon as this week. But first, we check in on these markets, which again are sinking. For a second straight day, the Nasdaq 100 under pressure on Friday.

2:28It was around anxiety or underwhelm between the US-China relations. Now it's front and center back on geopolitics and the US-Iran conflict. We're off by five-tenths of a percent as oil pushes higher. But we're seeing the semiconductor index once again taking a pullback. We're down by 2.2%. In fact, we've lost about 6 % on this index in the last two trading days. As we saw a significant sell-off on Friday, it ramps up into another day. Now, all of this is why we still try to digest the reality of the AI demand and spend. And all eyes are going to be over in Las Vegas on Dell World today. This is as we think about how Dell is putting on its annual conference with developers.

3:04And we're trying to understand what new tech, new innovations, what new offerings are on board. But there's a very special conversation to be had a bit later. Ed Ludlow standing by. And look, you're going to speak to the men of the moment a little bit later.

3:19Yeah. And there's a lot that needs to be reconciled, right? This is the first time that the gentleman will be speaking, having come back from China. And what needs to be reconciled is that in March at NVIDIA's own conference, Jensen Wang said that they were ramping up their supply chain for H200. The U.S. had approved the licenses so they could do business in that market. the president on Friday aboard Air Force One said that he did discuss H200s with Xi Jinping but China does not want them and so now we're in this this holding pattern where we need to explain to the market what's going on because a big part of the trade Friday and a little bit into this morning's trade has been this idea that nothing concrete came out of President Trump's visit to China with the video kind of in the center of it but generally speaking so many supply chain questions post melt-up in the AI infrastructure trade.

4:09Yeah, maybe we take a breather in that melt-up just on the day, Ed, or a second straight day. But meanwhile, the enthusiasm doesn't seem to be particularly being curbed for NVIDIA if you're looking at some of the analysts' notes today. Yeah, it's fun to just hit FA Go on the Bloomberg Terminal and just go back to basics, right? When they have their print on Wednesday, EPS growth above 80%, top-line growth just below 80%. And what's so fascinating about being here at Dell Technologies World is that if you go back at history of how these companies used to do business, right, NVIDIA never really sort of had Jensen Wang fly around the world and deliver chips directly to the end user.

4:50Dell was the kind of key sales channel. And so in the era of more agentic AI, where more enterprises are investing to have the compute capacity they need, it's a really interesting lens. And the timing is curious, of course, right? Speaking Monday, NVIDIA reports earnings Wednesday. But they are very key partners in the state of play for AI servers, what they call AI factory, but data center for running AI workloads. And, you know, this is kind of his playbook, right, Carol? Let's be honest about it. You know, Jensen does tend to appear alongside his partners often. But I think we'll learn a lot today about the state of the world, the demand levers, and also the issues with supply.

5:30So much to learn. Ed Ludlow is going to be with us a little bit later in the show, but tune in later today. It's going to be an exclusive interview with the CEOs, as he says, of Dell, of NVIDIA, live from Dell Technologies World. That's at 2.30 p.m. Eastern, 11.30 a.m. Pacific time. But as we think ahead to Wednesday, NVIDIA's earnings out after the bell, I want to bring you what to expect with Mandeep Singh of Bloomberg Intelligence. And Mandeep, we just heard a little bit about the drivers, the trends from Ed, but get into the nitty-gritty on NVIDIA. Is there as much enthusiasm as to really what the CPUs are bringing to bear as well as the GPUs and the infrastructure?

6:06Yeah, look, I mean, when a company is growing top line the way NVIDIA is growing and that top line growth seems to be accelerating, it's just there's no doubt that, you know, they are capitalizing on the demand and the chips remain in short supply. The only thing I would caution on is, you know, the rising memory prices. So we still have to see the impact of that on NVIDIA's margins. And they have talked about, you know, maintaining those mid-70 % gross margins. And the other thing is there's been that big shift from training to more inferencing and reasoning. And we know NVIDIA almost had a monopoly when it comes to training workloads.

6:50But when it comes to inferencing and now with the rise of Anthropic, you know, and how much traction Anthropic has had with coding agents, it remains to be seen if NVIDIA still has that bulk of the share when it comes to the use of their chips. And I think that's something we'll try to ascertain with the earnings call. I mean, they've made moves in inference. I think about the acquisition of Grok with a Q, not a K. But, Mandi, talk to us about what we might hear about that partnership, about what they're building to serve the inference market. Absolutely. And you bring up a very good point that they have talked about, you know, Grok being a very big driver on the inferencing side to the tune, you know, north of$20 billion just from that acquisition, what they're doing, integrating that acquisition.

7:39And look, I mean, we know Anthropic has also partnered with NVIDIA now. and Anthropic is the one that has had the most traction, you know, this year with coding agents. So from that perspective, some of those inferencing workloads will be deployed on Brock. And to your point, you know, agenting AI does require a lot more CPUs, which is why we have seen, you know, the likes of Intel and AMD do very well. So it will be interesting to see how NVIDIA is positioned in all of this with, you know, CPUs also having their moment when it comes to the chip demand. That's looking towards Wednesday after the bell, all after that key conversation that Ed has later today.

8:22Bloomberg Intelligence's Mandeep Singh, thank you very much. Look, Elon Musk, he says he's back in Texas, working on plans for that SpaceX IPO as the rocket and AI company prepares what could be the largest listing ever. Now, Musk made the comments virtually at the Samsung International Smart Mobility Summit in Tel Aviv. Take a listen. This IPO, you know, we've got to get the SpaceX IPO going pretty soon, I think. Meanwhile, SpaceX has notified investors it's executing a five-for-one stock split. It's lowering the per share price for investors. In an email, shareholders were told the fair market value of each share was adjusted to about$105, roughly$527 following the split, according to sources.

9:04Now, coming up, CME and Silicon Data, they are teaming up to create a futures market for computing power. A key resource driving the AI boom, Silicon Data CEO Carmen Leigh is going to be joining us. This is Bloomberg Tech.

9:29Let's take a look at today's big number,$67 billion. This is the largest power deal in history, triggered by guess what? AI data center boom. Next era is acquiring Dominion Energy for$67 billion in stock, forging a giant utility that's going to span from Florida to Virginia's tech hubs. Blue Meg's Liana Baker, who leads the M &A team, is here with us. Liana, what is it that's the driving force here? Is it really all about the AI compute need? You said it, and I'm surprised it took this long to see a major deal in utility since demand for power is at levels not seen since, you know, after World War II.

10:06So it was really a matter of time for some of these giants to come together in utilities because they need to team up to be able to pay for the build out that's necessary for AI. They say it's about scale, but with scale comes eyes of regulators. How easy is this going to be to get through? From what we're hearing, it's not impossible. Now's the time under the Trump administration to go bold and go big on deal making. But also where Dominion, the target, is based, it's only in three states, apparently. You know, Virginia's data center alley is a really key part of this. But because only three states might be needed for regulatory approval, you know, this could really happen within a year to 18 months.

10:49Of course, there's other regulatory bodies that also regulate power, nuclear, et cetera. But it's not an uphill battle. analysts are saying, you know, it's not a trivial regulatory review, but looks like it's going to be OK. The timing is going to be key, though, in terms of there's a lot of political pressure, a lot of political focus on the cost of energy right now, in large part because of the AI build out. So how is that going to serve a customer base? Do we have any idea what impact that have on pricing? In the press release, they hit on this idea that there's going to be these credits going to Dominion customers.

11:24So they're very aware that they have to focus on this idea of cost and price to make sure that consumers don't think their utility bills are going to go up because of this deal. So that will be key. You're right. And time will tell if prices will go up. Leanna Baker, an extraordinary M &A Monday. We thank you for talking us through it. Look, we talk about the need for energy within AI, but compute, of course, is officially becoming a financial asset class that we all care about too. CME Group, Silicon Data, they're launching the world's first futures market for AI computing power later this year, pending regulatory review.

11:59Now joining us, I'm pleased to say in the studio, is Silicon Data CEO, Carmen Lee. You're also the CEO of Compute Exchange. So talk to us first and foremost, what the relationship with CME is going to be bringing us. How important is it to have a futures market in compute? It's incredibly important. I'm glad that I came in this week. It's an exciting week for the semi-industry as a whole. As you mentioned, energy spend, we're probably going to spend over $10 trillion this year. Think about the spot market, from the energy to co-location to the server itself. Everything is at a commodity level.

12:33You need financial infrastructure to hedge to manage your risk exposures. Think about the banks, right? They're underwriting trillions of dollars of loans. How do they manage your risk exposure? How do they hedge to get future volatility as compute going online. So from a commodity point of view, think about how all your futures, the wonderful job for people hedging all your fluctuation, especially during the Iran war time. So it's incredibly important to have futures. You mentioned the banks, but who else would be the players involved in such a market that need to be hedging right now? So think about people who naturally long GPUs, so the data center themselves, the Neocloud, design houses, fabs.

13:11Everyone long GPU exposure, they want a way to hedge it through short futures or port options. Whoever needs compute, just like you and me, like AI startups, they spend billions of dollars on GPUs. They need to hedge their cost as a big component to a line item. Traditionally, SaaS companies, they do not really have a cost line item, as I found people. Now they do, either from GPU level or token level, and it's a big line item coming up. So the whole push is about transparency and an ability to be able to put these sorts of assets on your overall balance sheet. But then when it comes to transparency, just about the cost of compute, you've already been doing that with the compute exchange.

13:51What is it that you built there and who's using that? It's a great question. So compute change is spot market for GPU resources. So we do forward contracts as well as reserve contracts for GPU resources. So a lot of participants, I would say all participants so far, are the underlying consumption layer. so AI startups all the way to the new clouds. So we're two really big sides of companies. We work with European new clouds, American new clouds, Southeast new new clouds. On the flip side, we have startups and enterprises globally. This is really fascinating. You see how people try to do longer term contracts because they won't lock in the rate.

14:25They don't want the 40 % daily volatility comes on demand GPU pricing. There's other way to hedge that through future options. So we want to make sure we have the right venue for the right use space. If you want to reserve GPUs, you want to physical deliveries, go to ComputeChange. You want to hedge through financial exposures, go to CME in the future. Tell us about what sparked this whole idea. I mean, for full transparency for our audience, you used to be part of the Bloomberg team. You used to work in this very building. What was it that you felt was missing in the market? Was there literally no transparency around access to compute from one AWS versus a Microsoft offering or a NeoCloud offering from CoreWeave, for example?

15:02I won't say it's no transparency, but very little. So unless you're in the ecosystem, you're actually in the NeoCloud transactions, you probably do not know what's the reserve price that people are paying for longer-term contracts. Think about term structure risk today. And then think about this lack of visibility with on-demand pricing all the way to reserve, which is three years, five years on deal. And also GPU is not homogeneous products. You need to normalize it. You can't just say, hey, I'm paying the H100 with certain HBM. You're paying different HBM, different geolocation. we can't just do a simple average.

15:35You need indices to normalize the underlying assets to produce as a data that's trackable and representative to the price we are paying today. Look, we're just talking to Manip Singh of Bloomberg Intelligence, who is saying that we need to get to the bottom of NVIDIA's numbers that come after the bell on Wednesday about how much HBM or memory costs are really starting to impact their margin. What is it that you're seeing in your data about just the cost of compute at the moment? Wait, the price is going up. The price is going up since December last year. It was pretty interesting. Last year, everyone was talking about, hey, shouldn't compute price always come down and come down to zero by year A?

16:09It's not really happening. Supply-demand curve shifts every single day. The term structure, the forward curve pattern shifts every single day because people have different expectations. There are different board nights come out all the time. Yesterday, last year is fab, and this year is memory, and going forward, maybe co-location space. So every time is different. So pricing pressure is there. For you, the pressure is on to get this deal solidified with the CME. You've announced it, but we did say there's regulatory oversight. What sort of hurdles do you need to go through? What sort of questions will be asked?

16:40So we are actively working with CME and a review with CFTC and other regulatory bodies. It is a very traditional data future product. We're not talking anything that's exotic, right? We have done that for oil, for gas, for any other asset classes. So I really don't see a major concern from anyone. And going global, how global is the offering that you currently have with CME or indeed the compute exchange? Where else are you likely to push in to more broad offerings worldwide? So we launched GPU indices at Bloomberg Terminal last year, the world's first GPU indices. We cover all prices ex-China.

17:16So think about the European from the U.S. We normalize the geolocation to one data point. So anything ex-China is kept to our indices. So ideally, around the world, you can hedge that, your exposure with our indices. Well, we'll have to see how this progresses. Come back on, I hope. Silicon Data CEO Carmen Leigh there on all things futures market for compute. But coming up, we're talking Apple's revamped Siri. But it's expected to introduce new privacy-focused features aimed at setting it apart from rival chatbots. More on that next. This is Bloomberg Tech.

17:54The right technology can strengthen human judgment. That's why Deloitte brings together AI and data analytics with multidisciplinary teams. People with deep industry experience who can challenge assumptions and help you connect the dots across your enterprise. From risk signals to operational pressure points to shifting customer needs, Deloitte helps you see what's coming sooner. So opportunities don't slip by and surprises don't spread. It's not just dashboards. It's real clarity in the moments your decisions are made. When models reveal patterns, people can ask better questions. When data and people are connected, leaders can move faster with confidence.

18:34And when your teams are aligned, smart choices can scale from the front line to the C-suite. Because the smarter your systems, the sharper your instincts. That's how technology makes people better at what they do best. Deloitte. Together makes progress. Learn more at Deloitte.com slash Together Makes Progress. Amazon Health AI presents Painful Thoughts. I, um, I can't stop scratching my downtown. Yeah, but I'm not itching to go downtown and tell a receptionist I'm here to talk about my downtown. Some things you'd rather type than say out loud. There's no question too embarrassing for Amazon Health AI.

19:16Chat your symptoms and get virtual care 24-7. Healthcare just got less painful.

19:46with high R &D spend, small cap stocks with improving operating margins, or the S &P 500 minus high debt companies. Chances are there isn't an ETF that fits your exact criteria. But on Public, you just type in a prompt and their AI screens thousands of stocks and builds a one-of-a-kind index. You can even backtest it against the S &P 500. Then you can invest in a few clicks. Go to public.com slash market and earn an uncapped 1 % bonus when you transfer your portfolio. That's public.com slash market.

20:42Time now for Talking Tech. And first up, Baidu's 1 % drop in sales was less than analysts had expected, void by its new AI businesses. The tech giant in China is making a strategic pivot, shifting towards automated agents and AI-powered cloud units and away from search. Plus, Publisys is acquiring online data broker LiveRamp for about$2.5 billion in cash. Publisys says it's paying a 30 % premium per share to supercharge its data assets and accelerates its AI-driven marketing capabilities. And shares of Japanese chipmaker Kyosha, they were flooded with buy orders after forecasting a massive$8.2 billion quarterly operating report.

21:21Profit, it is the latest sign of meteoric demand for hardware for AI services and the resulting shortage in conventional memory chips. Let's talk elsewhere in the world of AI because Apple is set to debut its iOS 27 and this year's WWDC featuring a revamped Siri experience. Now, the update is expected to introduce new privacy-focused features that could help differentiate Siri in the crowded chatbot space. For more, Bloomberg Apple and Consumer Tech Managing Editor Mark Gurman brings us what is always the bread and butter. You bring us the scoops and push us towards, for many reasons, Apple wants to lean in on its privacy advantages.

21:56Is that what is going to make it distinct versus the other offerings already out there? Yeah, for Apple, privacy has been a differentiator now for going over a decade. They talk about privacy when they release any new product or any new operating system or any new software or service feature. For them, it's, you know, they're North Star, they like to say. Everything is built in from that perspective. And they really see it as a selling point that people want devices that have better encryption, better security, better privacy than the competition. And obviously, this is really important for everything you're doing on Apple devices.

22:30You have your financial data on there, your communications. You have your smart home information. all sorts of different features you want to keep private. And that's going to extend to the new Siri. We're in an age right now where everyone's got a chatbot, everyone's got AI features. And just like you've seen Apple do with features and comparing themselves to Meta and Google and whatnot, you'll see them do that with this chatbot that they're going to announce on June 8th at the developer conference, talking about theirs retaining less information and less so-called memory than other chatbots.

23:03And one feature that I reported over the weekend that they're planning for this, is the auto-deletion of chatbot conversations. So you can set your chats just like you can with the Messages app to save in your phone and your iCloud forever, or you can have them auto-delete after a year, or you can have them auto-delete after 30 days. So that's going to be a bit of a unique feature. Other chatbots have what they call an incognito mode, which is basically private browsing mode in a web browser, but private chatting in your chatbot. But this is going to be a little different. Talk to us about just how technologically that makes it stand out.

23:39If I have an auto-delete for 30 days, what does that actually mean to the memory that then me having had that discussion with the chatbot and it being useful going forward? Yeah, well, it's two separate things, right? The auto-delete of conversations means that that conversation, that chat, that's just gone forever. In terms of memory, right, you have chatbots that keep your memory forever, keep your memory for a finite amount of time. Apple is going to retain memory a shorter amount of time than a lot of the competition, which may make it useful in the short term, less useful in the long term.

24:14They're using techniques. There's one technique they have called differential privacy to be able to retain some memory, but not have it glued exactly to the user. And when it's stored, not glued towards the user, but you're able to store it on device. So it's able to come back and tell you what you've talked about or give you that context. So I think it is going to work fairly well. It is going to be fairly competitive, but Apple's really going to market it around privacy and security. As always, a must read with power on Bloomberg's Mark Gurman. Thank you very much indeed. Let's talk about Miami International Holdings, or MIAX.

Read the full transcript

24:48It has seen tremendous growth since it was founded over a decade ago, operating eight exchanges across options, equities and futures for disclosure purposes. As part of a collaboration with Bloomberg and MIAX, have launchpad futures and options. And Bloomberg has a dedicated studio on the exchange floor, where Bloomberg's Nora Melinda is joining us now. Hot off your interview with MIAX CEO Thomas Gallagher. Talk to us about what you've been discussing over there. Well, Caroline, I had the opportunity to chat with Tom Gallagher earlier this morning. Of course, we're here celebrating the launch of the Bloomberg 100 Equity Index Futures.

25:22Now, this is a product that the company is launching, hopefully geared at retail traders. We've been seeing an emphasis on a lot of companies, a lot of brokerages really being interested in bringing in the retail participation. This company offering the opportunity of commission free for retail traders as they are launching this product. The first of a series of products to also come. What was really interesting from my conversation earlier with Tom was the idea of Miami as the Wall Street South. We've been seeing so many big business leaders, Ken Griffin, a number of other names, talking about the tax environment being more favorable in states like Florida, southern states.

26:00And so you're certainly seeing MyEx really leveraging that, coming to Miami. Their headquarters, funnily enough, are actually in Princeton, New Jersey, but the trading floor just behind me. So a lot of activity here in some of these southern states and certainly MyEx looking toward further growth and development here. Fantastic. Nora, Melinda, thanks for bringing us and insights from your conversation. We appreciate it. Coming up, well, we've got plenty to be discussing with a recursive CEO, Richard Sosha. It's an AI company coming out of stealth. And already he's been running u.com as the CEO.

26:35He's already also got AIX, which is a venture fund. But now he's starting yet a new company all around the issues of AI and where we take large language model creation in the lab and the frontier lab area. We're looking, though, right now as AI not being the savior to these markets. NASDAQ 100 off by 0.7%. Look, it's the second straight day of losses. We've seen the semiconductor index as well off by 2.4%. That brings its losses in the last couple of days to more than 6%. We've, of course, got a lot of the eyes on NVIDIA for later this week. Wednesday after the bell, we get their numbers. We're off by percentage point for the world's most valuable company.

27:10This is Bloomberg Tech.

27:18Welcome back to Bloomberg Tech. Look, we've got a little bit of risk aversion creeping into the market, and it's largely geopolitically focused. We're worried once again about the conflict, US, Iran, and we're seeing oil prices on the rise. That is seeing yet a second day of pullback, particularly on the Nasdaq 100, or off by 8, 9 tenths of a percent, let's call it. We're seeing the chip stocks under pressure, in particular Micron and the like, really dragging in points perspective. But I'm also looking at what's happening in crypto. Look, risk assets in general taking a bit of a hammering. We're seeing down by 2.8%, but we're sub 77 ,000 now.

27:49It was 76 ,000 for Bitcoin. And there's been about 70 million in terms of the actual bearish bets coming back into the market. Bullish bets are being taken off. We've seen$700 million worth of those bullish bets being taken off in the last 24 hours alone, according to CoinGlass. So just keep an eye on risk sentiment. But look, if we look across to Asia, not all was bearish. I'm looking at Samsung shares in South Korea, up nearly, we'll call it 3%. Investors were actually encouraged by signs that the company's union is open to some negotiations, while a local court moved to limit the threat of potential strike.

28:21Bloomberg's global tech editor, Peter Elstrom, joins us. And just remind us of the context here. This is Samsung that has done phenomenally well in terms of share price, in terms of profitability. And now labor wants to see some upside from it. Yeah, these are very tense times in South Korea. Make no mistake about it. Samsung is squaring off against its union. The union has said it may go on strike as soon as Thursday for 18 days. That's a big deal for the company because, as you say, they're in the middle of this boom in memory chip prices, which has been great for their profits. They made a ton of money because of that.

28:54The whole AI industry needs more of the memory, both from Samsung and from SK Hynix at this point. But this labor disruption threatens to throw a wrench into that whole dynamic. And just to set the scene, it is quite unusual for Samsung to have this kind of conflict with their union. For decades, they never had any conflict at all. And then just a couple years ago, they had a small strike, but it wasn't that serious. So this is unusual for them to come this far. As you say, they've agreed to go back to the table after setting this possible strike date. They are talking again, but they haven't reached a resolution yet.

29:25The court took a little bit of the potential damage off the table by saying there are some limitations about what the union can do at this point. But there's still this possibility of a strike coming up later on this week, which would affect not just Samsung and the South Korean companies, but the AI industry more broadly. Yeah, everyone waiting with bated breath as we worry about memory prices yet further if there's less of supply. But Peter, I mean, it was just last week that we saw the South Korean market lose billions, hundreds of billions of dollars in just 90 minutes as people wondered how, politically speaking, we might see AI profits be shared with a nation writ large.

30:02We've already seen SK Hynix have to offer more of its profits to its workers. Just remind us of the tensions there and if they'll also spill out globally. Yeah, the Korean society right now is having this debate about what are you going to do with these enormous profits coming out of AI? And also the government is beginning to ask this very important question. How should governments prepare for this world that's coming at us with AI, with all these AI improvements, where perhaps you're not going to have the same kind of jobs that you had in the past? So as you're referring to, one of the policy advisors to the president posted something on his Facebook page where he talked about the possibility of a citizen dividend, where they would take some of the profits from these big companies and use those to establish a more stable foundation for the society.

30:47That could involve retraining, he said. It could involve support for people who want to start up companies. It could involve a number of different things. But his goal was really to kick off this conversation and try to ask the question, How do we as a government proactively start to think about these big issues? And if we're going to get this windfall of profits because these companies are making so much money, how do we use that cash to the best advantage for what's ahead of us? Pretty much, Peter Elstrom. Fascinating what's happening in Asia. We appreciate it. Let's look at what's happening with Google's leadership in AI right now.

31:17What's made computing power a really prized resource inside the company, with priority increasingly going to major projects like Gemini and paying customers. The bottleneck, well, it's frustrated some employees, prompting a number of AI researchers to leave for startups. For more, Bloomberg's Julia Love, who covers Alphabet, joins us now. And Julia, they're going and founding their own companies because they can't get what they want done internally at Google. How much of a change is that to be struggling to get access to compute? Yes. So I think it's just a fact of the AI industry that researchers would always like more compute.

31:54But the shortage has just become more acute in this day and age because large language models require such vast amounts of computing power. And so traditionally, AI research at Google, especially in Google Brain, one of the labs, was very egalitarian. Researchers got a certain amount of computing power and they could team up to try to do bigger projects. But now that pool still exists, but it contracts in times of high demand. and researchers just often can't get meaningful research done with their allocations. And so if their work isn't a priority at Google, that is the only avenue they can pursue, whereas in a startup, they have more doors to knock on.

32:41I mean, let's just talk about Andrew Dye, who you've focused on in this particular story. You've also focused on a female-founded company that's gone on to get VC backing. when they articulated to you what it used to be like, what it is now in terms of building businesses. But is it because Alphabet is too busy selling out its compute to its cloud customers? Or what is it that's making it so much more of a tussle? Yes, I think there's a few things happening at once. One is that generative AI has made, you know, Doodle's consumer products so popular and Doodle needs a lot of computing power to serve those.

33:17And then the cloud business is taking off. So Google wants to meet that demand. And then at the same time, AI research is more computing intensive than it once was. And so researchers need more compute to do a cutting edge experiment. And so there's a lot of mouths to feed. Everyone wants more. And if people run up against roadblocks at Google, that... Oh, what a tech issue on the tech show. Dan, we'll get back to Julia Love when we can, but we appreciate her insights there. Meanwhile, let's talk about where else computers being used and what we're seeing in the Frontier Labs. Because Recursive is a startup creating AI that conducts experiments on how to safely improve itself.

34:06And it just came out of stealth. The company is valued at$4.65 billion. The backers including Graycroft, NVIDIA, AMD Ventures and more. CEO Richard Socher joins us now for more. Many will recognize you because, Richard, you've joined on another company that you're also leading, U.com. You also happen to be busy in venture capital. So talk to us about recursive. What is it that you felt needed to be fixed? So we want to build recursive, self-improving superintelligence that automates knowledge discovery. You know, AI now is code and can code. And so you can have a closed loop where you replace human ingenuity and trying to come up with ideas, implement those ideas, and validate those ideas.

34:51And you can actually allow AI to do that on itself. The reason you're able to enter out of stealth with a$4 billion market capitalization is because you're a pretty well-known famed AI researcher, Richard. But there are many, well, shall we say, there are many very ambitious labs out there who also want to reach super intelligence. Why aren't they doing it right? What is it that you feel Recursive is going to gain an edge of rather than helping another lab already do what they're doing? Yeah, we have, we collected and combined an incredible team of co-founders, folks from Google, DeepMind, from OpenAI, from Salesforce Research, and many other incredible labs, Meta and so on.

35:34We are focused on open-endedness and really have the entire company try to allow AI to build the next better version of AI. And whenever in AI research we replace manual systems with learned systems, we see a massive improvement. And then on top of that, we can sit now on top of this whole new infrastructure where we built a company from the ground up by empowering AI to do its own experimentation. Whereas most of the other labs in the past basically started before that was possible. Okay, so you've got this focus from the very birth of the business, but you also need an awful lot of money, as we can see that you're raising, for compute to take on some of the juggernauts.

36:17How much is it a costly business? How are you thinking of reframing the demand for GPUs and the like? You're absolutely right. A big new scaling law we're trying to prove is that the more compute results in more inventions, more improvements. And so compute is one of the largest costs for us, and that's why we're really excited to have the folks like NVIDIA and AMD be part of this round. Just the open-ended nature of driving in endless innovation. You say you want to do that safely. What are the worries and limits of it not being done safely? You're absolutely right. Safety is a huge concern. That's why we have some of our co-founders who've done research in rainbow teaming, for instance, where AI improves the safety of large language models like Tim Rock-Teschel and his collaborators worked on.

37:07We take safety very seriously. And we believe, though, that this is the best and fastest way to superintelligence, which will then allow humanity to flourish much, much more by inventing, building ultimately a Eureka machine, one that allows us to invent everything else thereafter. after. How are you going to divide your time, Richard? You're CEO of U.com and you're also at AIX, the venture firm that's just lost its co-founder. So how do you divvy this all up? I work all the time and I have incredible teams around me that enable this. Of AI agents? Also AI agents, yeah, but also some really incredible humans.

37:51Okay. Incredible humans cost a lot of money. So are they incentivized by equity, by the sheer ambition of the task? Or how are you going to be hiring yet further talent? Because they're also getting to a significant expense line compared to GPUs. That's absolutely right, yeah. I think as a founder these days, you want to think more about the overall potential of the pie rather than your size and amount of pie that you have. As in, we do share equity with our employees. and everyone. And I think that encourages and excites everyone. I think in general, the more you have ownership of what you're creating in this AI world, the more you love AI because then you just care about the outputs.

38:34And so I think it's important to share that equity and ultimately grow something amazing together. Richard Sosha, great to have some time with you. CEO, co-founder of Recursive. We appreciate your time today. Coming up, we're going to go back to Las Vegas. It's Dell World. And we're speaking with Daniel Roberts from the AI cloud provider, Iren. That's next. This is Bluebird Tech.

38:58Amazon Health AI presents Painful Thoughts. I, um, I can't stop scratching my downtown. Yeah, but I'm not itching to go downtown and tell a receptionist I'm here to talk about my downtown. Some things you'd rather type than say out loud. There's no question too embarrassing for Amazon Health AI. Chat your symptoms and get virtual care 24-7. Healthcare just got less painful. Support for the show comes from Public. Lately, it feels like there are two types of investing platforms. Some are traditional brokerages that haven't changed much in decades, and others feel less like investing and more like a game.

39:40Public is positioned differently. It's an investing platform for people who are serious about building their wealth. On Public, you can build a portfolio of stocks, options, bonds, crypto without all the bugs or the confetti. Retirement accounts, yep. High-yield cash, yes again. They even have direct indexing. Public has modern design, powerful tools, and customer support that actually helps. Go to public.com slash market and earn an uncapped 1 % bonus when you transfer your portfolio. That's public.com slash market. And paid for by Public Holdings. Brokered services by Public Investing. member FINRA SIPC.

40:16Advisory services by Public Advisors, SEC Registered Advisor, crypto services by ZeroHash. All investing involves risk of loss. See complete disclosures at public.com slash disclosures. Travel smarter, not harder at America's Best Value In by Sinesta with convenient locations from coast to coast and value pack comfort at every turn. It's a practical choice for road trips, quick getaways and everyday travel that keeps things simple without sacrificing comfort. And when you're a Sinesta Travel Pass member, Staying at America's Best Value In means earning points toward free nights, upgrades, and more every time you stay.

40:50Go to Sonesta.com to book your stay and unlock the best rates with Sonesta Travel Pass. Here today, roam tomorrow. Join now at Sonesta.com. Terms and conditions apply.

41:07Let's get back over to Las Vegas. Our own Ed Ludlow is standing by with Dell Technologies World Event upon you, Ed.

41:16Yeah, and we've got a really great case study. Iron is an AI cloud that is building infrastructure all over the world, not just the United States. In all these geographies, Dell is its key server partner, right? It wants the NVIDIA GPUs. Couldn't think of a better conversation to have. It's kind of kick off the day. Dan Roberts, co-founder, co-CEO. You've been a busy guy as well on the M &A side. You've acquired a marketing or creative agency. I mean, basically why? But there's also this conversation, Caroline and I have it all the time, about the public's perception of what's really happening with data center and how it impacts them.

41:54Yeah, look, brand awareness, we're going global, we're growing our customer base. It's really important to us, but that's a small point in the grand scheme of things. It's all about power and time to compute. And the issue is you can't implement a software update to bring on power. The tech industry is very brilliant at going really fast, but you can't code your way to an AI gigawatt of compute. And that's what we're focused on. In the United States, at least, there is some sort of public backlash, right? In zip codes where data centers are getting built, the public feels the pressure of higher energy prices.

42:28What's that like for Iron? And what are you confronting in that respect? Eight years ago, when we started this business, we had a differentiated strategy. We never went to metropolitan areas. we went to regional communities where there was an abundance of renewable energy that creates a cost, a structural advantage over time. And we've reinvigorated local communities across British Columbia, Canada, Texas. And it's an advantage for us because we are located where people want us to do business. Dan, I think I'm right in saying, just as with everyone else, you're in a position where demand for compute, your compute, is vastly outpacing your ability to supply it.

43:05what is the biggest barrier to you getting more capacity online, getting more stuff built? Absolutely. I mean, we're still in the dial-up era of AI. When you step back, the amount of friction you've got, waiting for prompts, you've got to craft your prompt really carefully. You know, complex prompts take 15 to 40 seconds to actually spit out an answer. As that speeds up, demand accelerates, which drives more compute and more use cases. And for us, it's been a focus on what is the end choke point and that's power it's steel it's concrete it's actually building the physical infrastructure that takes years and years if you wanted to start today and build a gigawatt ai factory you are looking 2030s before you get the first computer online let's get a bit deeper on that so if you wanted to say is in other words you choose a site you get permissions you're not going to get it online for six years four years it'll take you 18 to 24 months to get the attention of a utility to work out whether your block of dirt even has power or not.

44:04That's the starting point. Now we're getting to it. What are the utilities not doing that you need them to do? Well, utilities, they're doing the best they can. Like they're running complex networks with a lot of generation on one side, a lot of power demand on the other. And at the end of the day, when you connect to them, this is why grid connected capacity is so valuable. When you connect to them, they're underwriting 24-7 guaranteed power. So they're naturally risk adverse and they're naturally dealing with an inundation of requests at the moment. I'd love to go back to basics for a second.

44:33You build data centers, and those data centers have servers inside them, and those servers are for AI workloads. What's it like working with a Dell? What is it that you rely on Dell for? And how do you get hold of NVIDIA GPUs, right? There is this idea that you can just phone Jensen Wong and say, hey, I need some of those. But in practice, it probably doesn't really work like that. Yeah, look, we've got a great relationship with both Dell and NVIDIA. We announced a partnership with NVIDIA for 5 gigawatts of compute. Are you also being an investment? Yes, so they're exposed to upside in us as we grow and deploy their GPUs, and we're working on their reference architecture for a gigawatt factory in Sweetwater.

45:13But the opportunity with those partners is to work closely and build up the whole ecosystem, all the way from land, steel, concrete, down to the end users of this compute. Is there a geography somewhere in the world that is the best place to build a data center right now? Look, every geography is slightly different in terms of how they handle things, but the majority of our capacity is in Texas and they've been pretty good. If you could ask Mr. Wong and Mr. Dell one thing right now, or ask something of them, what would it be? Oh, look, I think the whole industry is moving from if silicon is no longer the choke point, How do we solve steel?

45:51How do we solve concrete? How do we solve the kilowatts? And this president and this administration, they have an answer for that? I think everyone's trying to work it out. It's been an interesting evolution for your company, right? Now public in the United States, Sydney, Australia is your home. What's that like trying to work across borders at the moment and trying to get projects done at the same time in different jurisdictions? Oh, look, the reality is we've got projects in North America. We've got projects in Europe. We've got projects in Australia. Yes, my family is in Sydney and I get back for the weekends.

46:25But, you know, when people ask where I live, my typical answer is on a plane. Dan Roberts, co-founder, co-CEO of IREN, really critical AI cloud. And like your previous guests, I think everyone's just working all the time right now, Caro. Largely on planes. Ed, thanks so much. Look, tune in later today as Ed sits down with the CEOs of Dell, of NVIDIA, for an exclusive interview live from Dell World. You don't want to miss it, 2.30 p.m. Eastern, 11.30 p.m. Pacific. Coming up, new 13F filings. They're revealing how hedge funds are repositioning around big tech and the AI trade. We'll break down the biggest moves next.

47:01This is Blueberg Tech.

47:15Fresh regulatory filings, known as 13F filings, are giving investors a close look at how hedge funds and major money managers are positioning around big tech and the AI trade. Joining us now to break it all down is Bloomberg's Hemmer Palmer. It's always so much happening each quarter at this time. Hemmer, pass through, for example, what's happened with Microsoft, because certainly the Bill Ackman news caught our attention. Where else has there been moves in that company? Absolutely. Bill Ackman taking a pretty sizable stake in Microsoft, more than$2 billion worth, making it his fourth biggest holding.

47:44And this is a stock that fell a great deal in the first quarter. So he may have been buying up these shares at attractive prices. But he believes that Microsoft is stronger and a lot more resilient than investors are out there believing. He thinks Microsoft's 365 suite with Word and Excel are two of the most valuable franchises in enterprise tech. So really bullish sentiment from Bill on Microsoft. But that's not what everyone's thinking. Yeah, is that at odds with some of the others, the Tiger Hubs? Absolutely. So it's really interesting when you see these divergences. So Tiger Global and Co2 slashed their stake in Microsoft by more than half.

48:21Wow. Lone Pine got out of it entirely. So you see really different moves from these firms that really do spend a lot of time investigating these companies. Now, the stock fell 23 % in the first quarter. It has rebounded some of that since. What about just in general, whether Microsoft is the old one out here in terms of everyone either taking one view or the other or the big moves in other companies? So we did see a decent amount of selling in the first quarter from across the Tiger Cubs. We look at them because they tend to focus on tech. They tend to be directional and intentional. Keep in mind the first quarter, incredibly volatile.

49:00March was a wild month. A lot of funds lost a lot of money. That could be why we saw some firms trying to sell. We don't know when in the quarter funds were selling just because of the nature of 13Fs. But we did see some other trends when we look at popular sells. Amazon, this is a stock that sunk about 10 % in the first quarter. Tiger Global trimmed its stake. Code 2 cut Amazon by about a fifth. And Lone Pine sold out of Amazon entirely. Berkshire Hathaway also exited that company too. So you do see some big moves in some of these localized companies, for sure. Fascinating, because some of them have since ridden up higher on the back of that move.

49:40Hemipalma, always breaking it down for us. We so appreciate all the moves from the regulatory filings. Meanwhile, that does it for this edition of Bloomberg Tech. Tune in later today. Again, do not miss this exclusive interview. We can't tell you about it enough. CEOs of Dell, of NVIDIA, they are live from Dell World. It's at 2.30 p.m. Eastern, 11.30 a.m. if you're over on the Pacific side. Don't forget to check out the podcast too. You can find it on the terminal as well as online on Apple, Spotify, and iHeart. From New York and from Las Vegas today, this is Bloomberg Tech.

50:22If you follow markets, you know the value of long-term thinking. You plan, you diversify, you prepare for volatility. But even the best strategies can't prevent every bad day. For more than 75 years, Cincinnati Insurance has helped individuals and businesses navigate tough moments with expertise, personal attention, and independent agents who focus on relationships, not transactions. The Cincinnati Insurance Companies. Let them make your bad day better. Find an agent at c-i-n-f-i-n dot com.

51:23products like premium apparel, drinkware, tech, and more, making it easy to create a gift that feels meaningful and on-brand. And with 4imprint's expert support and their 360-degree guarantee, you can be 4imprint certain your order will arrive exactly as intended. Explore gifting with confidence at 4imprint.com. 4imprint. 4certain.

From the publisher

Bloomberg’s Caroline Hyde discusses what to expect from Nvidia this week as investors anxiously await the company's earnings results. Plus, the AI data center boom triggers the largest power deal in history, the filing for SpaceX's IPO could come as soon as this week, and Ed Ludlow brings the latest from Dell Technologies World in Las Vegas.

See omnystudio.com/listener for privacy information.

More from Bloomberg Tech

All 343 episodes
Nvidia Earnings This Week; Biggest Power Deal in HistoryBloomberg Tech · 44 min
Listen in VO