Qualcomm Signs Deal to Provide Amazon With Custom AI Chips

8 Sep 2026 · 44 min · 17 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

This Bloomberg Tech episode centers on AI infrastructure and chip deals, plus related AI safety/governance and funding news. Main topic: Qualcomm signs a long-term deal with Amazon to supply custom AI/data-center chips across multiple generations; Amazon can buy up to $4B of Qualcomm shares, and the chips are meant to support AWS AI infrastructure (starting with custom connectivity chips, later expanding to high-bandwidth connectivity and AI inference accelerators).

Key claims

the deal validates Qualcomm’s entry into data-center chips; it diversifies Amazon’s supplier/vendor stack (reducing reliance on NVIDIA); and it’s framed as risk-sharing/“circular financing” rather than problematic leverage.

Notable examples

Amazon’s RAMP; Qualcomm’s investor-day roadmap; comparisons to Google TPU; Anthropic’s failed $6B acquisition of Descartes AI; Mistral AI’s $3.5B funding led by Samsung; OpenAI chief scientist calls for “extreme caution.”

Guests

Isabel Lee (Bloomberg markets); Khunjan Sabhani (Bloomberg Intelligence); Joanne Feeney (Advisors Capital Management); Leanna Baker (Bloomberg); Arthur Mensch (Mistral AI); Ayoko Yoshioka (Wealth Enhancement Group); Sahir Jaggi (Forrest); Mark Gurman (Bloomberg).

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 Qualcomm Deal

2:29 to 4:23

Discussion about market trends and Qualcomm's deal with Amazon.

“Let's get a check on the markets with Bloomberg's Isabel Lee, who joins us now.”

Qualcomm's Custom Chips for Amazon

4:23 to 6:40

Exploration of the specifics of Qualcomm's chip deal with Amazon.

“Khunjan, tell us, what exactly is Amazon buying from Qualcomm as announced today?”

Market Reactions and AI Challenges

6:40 to 9:59

Analysis of market reactions to the Qualcomm deal and AI finance concerns.

“And Qualcomm shares have jumped on the news, Kunjan.”

Labor Market and Construction Concerns

9:59 to 12:20

Discussion on labor market challenges related to data center construction.

“Does this deal suggest that concentration around that company is softening as well?”

Geopolitical Update and EU-Canada Relations

12:20 to 13:31

Overview of EU-Canada relations in the context of global power dynamics.

“That could be a real concern for investors to be watching going forward.”

Anthropic's Abandoned Acquisition of Descartes

14:08 to 17:40

Discussion about Anthropic's decision to walk away from the Descartes acquisition and its implications.

“Anthropic is walking away from a potential$6 billion deal to buy AI startup Descartes.”

Mistral AI's Recent Funding and Strategy

17:40 to 22:34

Mistral AI's CEO discusses the funding round and the company's growth strategy in AI and manufacturing.

“Let's take a look at today's big number,$24.4 billion.”

OpenAI's Chief Scientist Calls for Caution

25:23 to 28:00

Discussion on the need for caution in AI development, including recent issues with social media algorithms.

“First up, OpenAI's chief scientist is warning that AI is evolving so rapidly that it is becoming increasingly difficult for humans to understand and control.”

NASDAQ 100 Rebalance Insights

28:00 to 28:31

Learn about the implications of SpaceX's share lockups on NASDAQ 100.

“as the NASDAQ 100 rebalances later this month.”

ASML's Semiconductor Production Gear

28:31 to 30:06

Discover how ASML's new machines will impact semiconductor production.

“Samsung and TSMC have agreed to begin adopting the Dutch firm's latest semiconductor production gear.”
Show all 17 chapters

Industry Collaboration on Chip Production

30:06 to 31:27

Uncover why major chip makers are collaborating on new production technologies.

“And that's also true on the memory chip side, where Samsung and SK Hynix and Micron have all been struggling with these technologies.”

Demand for AI Infrastructure

31:27 to 31:43

Examine the strong demand for AI infrastructure post-earnings reports.

“Bloomberg's Peter Elstrom, thank you for the update.”

AI Growth Expectations and Concerns

31:43 to 34:00

Discuss future growth expectations for tech companies in AI and market concerns.

“Here to discuss is Wealth Enhancement Group Portfolio Consulting Director, Ayoko Yoshioka.”

Anthropic's IPO and Market Implications

34:00 to 36:12

Explore the anticipated IPO of Anthropic and what metrics are crucial for evaluation.

“How is the market interpreting that caution, those checks and balances being put in place by the companies more so than the government?”

Forrest's Series C Funding and AI Impact

40:09 to 42:00

Learn about Forrest's recent funding and its implications for AI in healthcare.

“Explore the possibilities at 4imprint.com.”

Navigating Complex Treatments in Medicine

42:00 to 46:50

Learn how AI is improving patient access to complex medications.

“really helping doctors and patients in a select set of conditions and a select set of geographies.”

The Anticipation of Apple's Foldable iPhone

46:50 to 50:22

Explore the significance and history behind Apple's first foldable iPhone.

“Coming up, Apple is ready to unfold its next chapter with the first foldable iPhone expected to debut tomorrow.”
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 most effective people at work aren't working harder than everyone else. They're working smarter, inside better systems. Superhuman Go, from the makers of Grammarly, is the AI assistant that works inside every tool you already use, always ready, already aware of what you're working on. It's a teammate whose only job is to help you be better at yours. With Go working with you, you can show off what you do best. See what Superhuman Go can do at superhuman.com. Some people treat ChatGPT like some kind of smart search engine, and some use it to get work done. ChatGPT Work is a new way of working in ChatGPT that can take action across your apps and files, stay with a project for hours if needed, and turn a goal into finished work.

0:46It's designed to help you move from a chaotic starting point to a reviewable first version. So all the source materials, briefs, and scattered information that you have to grind through to turn into something useful can just become something useful. Put ChatGPT to work on your most ambitious ideas and projects. Get started at ChatGPT.com by selecting Work Mode, available on Plus and Pro plans.

1:11Riley Griffin:Healthcare doesn't always work great. If you've ever waited on a refill or couldn't schedule an appointment, you get it. That's the kind of stuff Optum is changing. They're using data and technology to integrate patient care, pharmacy, and everything else. So healthcare is connected, not complicated. What's that look like? Cheaper prescriptions that are easier to get and care that looks at the whole person. How you need it. Optum is helping make healthcare work as one for everyone. Learn more at business.optum.com. Bloomberg Audio Studios. Podcasts. Radio. News.

1:54Bloomberg Tech is live from the heart of Silicon Valley

1:58Riley Griffin:with Ed Ludlow in San Francisco. Coming up, Qualcomm will provide Amazon with custom AI chips in a deal that the company says will span multiple generations. Meanwhile, another deal falls apart after Anthropic decides against acquiring AI startup Descartes AI. And all eyes on Apple as the company kicks off its product launch tomorrow with its foldable iPhone, a decade in the making. This is Bloomberg Tech. I'm Riley Griffin, in for Ed Ludlow. Let's get a check on the markets with Bloomberg's Isabel Lee, who joins us now. Isabel, what are you looking at? Hey, Riley, let's get a quick look at the markets, especially on this holiday-shortened week.

2:42You're seeing the Nasdaq here drift lower and lower, down by, honestly, it's almost unchanged. but earlier it was down by a tenth of 1%. And of course, that comes as oil prices soar, flirting with$100 a barrel. It has paired some of those gains, but it's still higher by seven-tenths of 1%. This comes after news that Saudi Arabia reported that some of the attacks has caused some of their energy facilities to halt operations. And of course, that's why you see the Brent crude there really rising. We still have a lot of geopolitical uncertainty and not even the strength of the chipmaker. You see the stock here, the semiconductor index, pop by more than 2%.

3:15Not even that can offset the losses that we are seeing here. And a quick recap of what happened last week. We got a blowout job support that really shocked a lot of investors. And it really raised the stakes on whether we will see a rate hike in the September 16 meeting. And, of course, the key data to watch out for now is a Friday. CPI data. All eyes will be on that because we've heard a lot of Fed speak. And some of the Fed speakers have said that that will be the key data to watch out for. Let's take a look now at some of the individual movers. You mentioned it at the top of the show, Qualcomm popping by 5%.

3:47It pared back some of those gains, but earlier it soared as much as 9%, and if it holds that gain, it will be the highest since around June. So you see here, really green on your screen there. So to your point, Qualcomm signed a long-term data chip deal with Amazon, one that will span multiple generations. This gives Amazon the right to acquire$4 billion in shares, and the tie-up will create customized silicon to support the AI infrastructure of AWS. So you see a nice healthy gain there, and you see Amazon there down.

4:13Riley Griffin:by five-tenths of one percent, Riley. Bloomberg's Isabelle Lee, thanks so much. For more on Qualcomm and Amazon, Khunjan Sabhani of Bloomberg Intelligence joins us now. Khunjan, tell us, what exactly is Amazon buying from Qualcomm as announced today? Yeah, so this is a sort of portfolio of multiple different products or multiple different generations. As Qualcomm outlined recently in its investor day, which where they didn't mention the customer name, but we believe that first hyperscaler to RAMP was Amazon, clear from this deal now. But initially, the RAMP will be around its custom connectivity chips.

4:51So, they acquired a company recently where they got a portfolio of connectivity IPs, whether it's CERDs, whether it's 1.6 GTSPs. So, this is what initial RAMP is going to be. Eventually, it will broaden out into high bandwidth connectivity chips and eventually getting to AI inference accelerator chips. So, this will be a broad portfolio, multiple products over multiple generation over multiple years.

5:15Riley Griffin:Qualcomm is best known for its smartphone processors. I wonder if you could explain to us a little bit about the significance of this deal being for AI infrastructure broadly. Yeah, I mean, Qualcomm has been trying to really diversify their revenues away from handsets, and they publicly have announced since last year. Also had an investor day where they laid out, They're entering into the data center chip market. It's a big temp for them, a big opportunity. So this is really this deal validates their entry. There was a lot of doubts whether them coming in so late, competing with the likes of NVIDIA and AMD and Intel, would they be able to really gain some share?

5:55So this real deal puts out some numbers. It's a 60 billion dollar ceiling in terms of the products that Amazon can buy through them. Yes, it is over five to six years, but really validates that they are a serious contender here.

6:09Riley Griffin:And 60 billion is a big number. Do you expect Amazon to really reach that maximum? I mean, again, like I said, it's a ceiling or going sort of beyond 2030. So initially, the targets that Qualcomm has laid out, which is the$5 billion and the$15 billion in 28-29, I think could be reachable, especially given the backing of this deal. And this is not the only customer. There are more hyperscale customers that they have alluded to, which we could see the announcements coming sooner. And Qualcomm shares have jumped on the news, Kunjan. How is the market reacting for Amazon? Yeah, I mean, look, this is overall still, it won't impact Amazon significantly near term.

6:54Qualcomm chips are not something which are supply constraints. It wouldn't sort of help them deploy CapEx faster. But this does keep on boosting Amazon's stack, which eventually, which they are using. And eventually, just like what Google does with TPU, they plan to use that for external customers. So bringing in more suppliers, bringing in more vendor diversity definitely helps Amazon stack with that regards.

7:19Riley Griffin:Kunjan Sobani of Bloomberg Intelligence, thanks so much. Let's stick with Qualcomm. Advisors Capital Management Partner and Portfolio Manager Joanne Feeney joins us now. Joanne, there's been a lot of concern about circular financing amid this AI boom. Is this kind of deal going to raise those concerns as well? I, Riley, I think we're going to see always that conversation pop up whenever you see cross investments by companies, by buyers and their customers and their suppliers. So our view is that circular financing is really sort of an alternative way for companies to become more vertically integrated, to have more control over the inputs, the technology that they need to use to build up their capacity in the future.

8:03And this is another example of that. investors still have the opportunity to diversify risk and pick the areas in which they want to have most exposure. But these kinds of deals sort of combine the types of exposure an investor will get now with Amazon, because now owning Amazon means you're going to own a little bit of Qualcomm. But it's a really innovative idea for solving what is a complex financial markets problem, which is the limitations on borrowing, the different capacities of balance sheets by these companies. And it also creates long-term contracts. And it gives Amazon more skin in the game at the Qualcomm level, which gives Qualcomm the confidence to be able to build up this capability and to take the risk of creating big design teams and putting the resources into that.

8:52So it's a really effective way of sharing that kind of risk. And it means that innovation likely happens much more quickly than it would otherwise occur.

9:00Riley Griffin:And Joanne, what does Amazon get out of working with Qualcomm that it couldn't achieve itself? It's a behemoth, a company that is closely followed by anybody watching the tech sector. But what is it getting here? So it gets another supplier. It gets another designer, co-designer for chips. It gets more supply on the connectivity side and ultimately on the AI inference side. It's already working with Marvell, and perhaps that one partnership could have been seen as too risky. And so they get to diversify, which gives them better pricing power down the line for acquiring these chips. And it also gives them more protection against NVIDIA's price levels, which ultimately probably soften a little bit with this increase in competition among the chip designers.

9:48Broadcom is another player here that Amazon doesn't currently work with. But it is a way for Amazon to diversify and to get multiple sources for these chips in the future.

9:59Riley Griffin:I'm glad you mentioned NVIDIA. Does this deal suggest that concentration around that company is softening as well? Yeah, we should expect it to, Riley, which really doesn't diminish the attractiveness of NVIDIA as an investment. It is one of our holdings across a few of our different strategies. We've known it for a long time. The thing to think about is the size of the market, how much that is growing versus market share. You know, NVIDIA has been dominant. They've had massive and dominant market share. And we should expect that market share to fall as more players come in. But that doesn't mean that NVIDIA is doomed here.

10:34By contrast, the market itself, we believe, is growing so quickly that all of these players, at least the ones that we like, Broadcom, NVIDIA, AMD, and then the users like Amazon, Microsoft, Google, etc., we think they're playing into a very quickly growing market. We see the revenues coming up at places like Anthropic and OpenAI, at least the information we can get out of them, which suggests that this AI infrastructure build and deployment of applications remains a really powerful growth opportunity for investors with a long-term horizon.

11:09Riley Griffin:Amid this AI infrastructure boom, we have recently received new jobs numbers. I wonder, when you look at that data, what are you seeing about the data center build out? Yeah, so what we're seeing there is a few different things. One is obviously a very strong jobs report. There was some softness in some areas of infotech job creation and real strength in other areas. And that should be expected. Right now we're seeing labor being drawn into data center construction. There's always been the question of, you know, will AI ultimately, you know, reduce job opportunities in some places in the economy?

11:44There was some evidence of that, but there was an economist article out over the weekend that suggested that the, you know, AI space is not leading to massive job destruction. So right now, the AI data center build, the infrastructure build is creating a lot of job opportunities in a country where labor is actually in short supply in that area. So it's a tricky situation. We've seen those fees on the H-1B visas actually constraining applications. And that's actually the kind of labor we need more of in this country. So we'll see how the industry deals with that labor shortage. That could be a real concern for investors to be watching going forward.

12:24Riley Griffin:When you think about the labor shortage, there's also the workers who are physically building this footprint across the United States. Is there enough labor to support the physical build out? You know, Rodney, that's exactly the right area to point at, and that's where we may have real shortages. I mean, immigrant labor, we know, is a big part of the construction industry, both housing and commercial builds. We know it's a big part of travel and leisure industry. And so those shortages, I think, will be constraining. I've talked to folks who report that they can't get enough roofers to build all the houses they want, and so they've had to cut their housing building plans back by half this year.

13:03This is a client down in Texas, for example. So, you know, we are starting to see that. I think it is constraining the housing industry. And I think that the construction workers are being pulled into the data center builds and it's hurting actually other parts of the economy. So, so far, they're able to pay up to get the construction labor that they need. But it's certainly something that we need to keep top of mind.

13:25Riley Griffin:Joanne McFeeney with Advisors Capital Partner and Portfolio Manager. Thank you so much. Let's get to a headline that just crossed the terminal a few moments ago. The European Union and Canada are working towards a new relationship that will range from trade to security to offset the global power politics dominated by the U.S. and China. Sources say the European Commission president is expected to announce the plans at a State of the Union address in Strasbourg, France next Wednesday. Coming up, Mistral AI raises$3.5 billion in its latest funding round led by Samsung. We'll hear from the French startup's CEO and co-founder, Arthur Mensch, next.

14:07Riley Griffin:This is Bloomberg Tech.

14:25Riley Griffin:Anthropic is walking away from a potential$6 billion deal to buy AI startup Descartes. That's according to sources. Anthropic had been exploring a deal but ultimately walked away, though the companies might still find other ways to collaborate. Here with more is Bloomberg's Leanna Baker. Tell me a little bit about the details of this fallen apart deal. So on my team, the deals team, we love to report on transactions that are actually happening. But sometimes when you have these high profile names like Anthropic, which is on the cusp of its huge initial public offering, it decided to walk away from this acquisition.

15:01Descartes was a company founded in Tel Aviv. It works in the AI space, and it was really sort of in demand, and the valuation would have been up to$6 billion. We're not exactly sure why Descartes balked from this, but they did perform due diligence on the company and then at some point decided not to pursue this. We see this a lot with acquisitions. To do successful M &A, it's all about timing, and maybe the timing wasn't right for Anthropic, but Descartes might have another buyer. We don't have the reporting to say who it is right now, but we'll definitely be

15:35Riley Griffin:chasing what company could step in for this startup. And for Anthropic's sake, the fact that they were exploring this deal in the first place, what was the technology they were seeking out of the deal. It would have fit in somewhere in its inference unit. I'm not super technical on what exactly Descartes does. They have a video engine. You can do video editing, but I've also read that it makes some of the AI models more efficient with video. So this is something that would have fit in somewhere. OpenAI, Anthropoc's competitor, has been a little more acquisitive when it comes to acquisitions, but Anthropoc hasn't been as busy buying companies.

16:14They've been sort of building and we'll see ahead of this IPO what happens. Once they're a public company, then they'll have a currency to do more deals. So this one probably won't come back,

16:25Riley Griffin:but it'll be interesting to see when Anthropic is public and what deals it could do then. We're all waiting with bated breath for the S1, that document that will reveal the insides of Anthropic's business ahead of the IPO. How does this failed deal fit into that picture? What kind of details are you looking for in that document when it comes to that? It's funny because when SpaceX filed their prospectus, there was details on the cursor acquisition in that filing. So we're going to be pouring over Anthropics Financials and to see if there's any mention of dealmaking. But this one, Descartes, will probably not be in the filing, I can predict, since they've decided to abandon this pursuit, which again happens in the world of M &A.

17:05Riley Griffin:And as I understand it, partnership is still possible even if it's not an acquisition. competition, is that something we anticipate learning more about? And would that keep, you know, the dollars and cents off the balance sheet there? Yeah, a commercial partnership is something that we said could happen because that just means, you know, Anthropic doesn't want to close the door on Descartes' technology. They want to, you know, maintain good relationships. And we've seen just so many partnerships across the board in AI and data centers. And this is one of them, but, you know, nothing to report right now.

17:37Riley Griffin:Bloomberg's Liana Raker, thanks for bringing us the latest. Let's take a look at today's big number,$24.4 billion. That's the latest valuation for French startup Mistral AI after raising$3.5 billion in its latest funding round, led by Samsung. Bloomberg's techs, Europe, Tom McKenzie spoke with Mistral AI CEO and co-founder Arthur Mensch yesterday. Well, it's of course an endorsement and I think a sign that the directions we've taken for the last three years has been the right one. Those 3 billion euros that we've just raised are going to be put at work to scale our training capacity, to further deploy the infrastructure that we need to serve our enterprise customers, to further invest in industrial artificial intelligence, which is what we have been doing with ASML and what we intend to be doing more and more in the semi-industry space.

18:29And it's also a way for us to continue to scale or go to market globally in that this fundraising is also a way for us to further expand in the US, in Asia, and of course, to double down on what we've been doing in Europe. And do you expect Compute to remain constrained through 2027? NVIDIA is an investor. You've now got Samsung, the high bandwidth memory. We know that's constrained. Does Compute continue to be constrained into next year? For how long does that remain a story? It will probably remain a story for a while. We do have good access. And so therefore, we've been able to build significant capacity for our customers.

19:03We also have started to run programs where our customers are actually committing to use the capacity we're building for them, which is a way for them to secure prices because we're in a world where the prices are going up because of the capacity constraints that we have on logic and that we have on memory. Everything that we do at Mistral is really meant to reduce the cost for our customers. the fact that we have ASML, NVIDIA and now Samsung at our cap table is also tying us further into the SOMI industry where we are a provider of technology we bring AI to make the SOMI industry more efficient and so that kind of collaboration I think is a way also for us to further guarantee to our customers that we are very strong at providing verticalized AI solutions for high-end manufacturing but also able to give good cost structure when it comes to doing every kind of AI deployment.

20:00And I'm wondering, your approach, full stack, that focus on industry, that focus on engineering, on design with the full stack, I'm wondering how that compares. It's a very different approach to OpenAI and Anthropic. And I'm just wondering if investors are looking in the wrong direction. Does it suggest that the Anthropic OpenAI Google approach is possibly not the right approach? Well, what we see today in the domain of AI is that there is a very strong shift toward open source models as a foundation for building critical workloads for using AI into critical processes. Why is that? Because when you're running a critical process in your factory, you want to have business continuity.

20:43You want to have cost control and you want to make sure that the IP that you've been building for decades is actually used to create assets that are uniquely your own. And so that has naturally brought a lot of our customers to build on open source foundations. And the fact that we have been a leader in releasing open source models and that we're able to serve them at scale on infrastructure that we own is indeed making a difference. So to be clear, longer term, your approach, the Mistral approach wins out over frontier labs like Anthropic and OpenAI. I mean, to be clear today, if you're an enterprise decision maker, almost, I would say 99 % of the use cases you have as an enterprise can be built on open source models.

21:26And so it means that the IP margin that you're paying when you're paying for closed models is not a margin you should be paying if you were to be efficient. And so the companies that are the most advanced and the companies that have the biggest requirements when it comes to controlling their data and the biggest requirements to have business continuity, and they happen to be in manufacturing, in financial services, in defense. Well, those companies have already decided that they will mostly build their technology on top of open source models. And I think this movement will continue to increase because structurally, AI is an enormous market.

22:02It's going to drive multiple digits of the global GDP in the next few years. And it is such a big market that any region cannot depend, cannot decide to buy the technology. They need to produce the technology. And if you want to produce AI, you need to use common assets and then customize them, deploy them on infrastructure that you control. And so you have macroeconomic reasons why open source AI is actually starting to end.

Read the full transcript

22:27Riley Griffin:That was Mistral AI CEO and co-founder Arthur Mensch speaking with Bloomberg's Tom McKenzie. Coming up, OpenAI's top scientist is calling for extreme caution in AI development. We'll have that story next. This is Bloomberg Tech.

22:51The people who seem to get more done than everyone else. They're not working longer hours or running on more caffeine. They've just stopped wasting time on the stuff that doesn't move work forward. Switching apps, re-explaining context, hunting for files. Those aren't small inefficiencies. They're hours wasted every week. Superhuman Go gives you those hours back. From the makers of Grammarly, Go is an AI assistant that sits inside every tab and tool you already use. always available and already aware of what you're working on. Ask it to draft something, summarize a long thread, pull up a file, or prep you for a meeting.

23:32Go handles it without you ever leaving the page you're on. This is what it looks like when AI actually fits into your work instead of adding to it. It's like having a teammate whose only job is to help you be better at yours. Go keeps up so you can move forward. With Go working with you, you can show off what you do best. Superhuman Go. Find out more at superhuman.com. That's superhuman.com. So there's a lot of noise about AI, but time's too tight for more promises. So let's talk about results. At IBM, we work with our employees to integrate technology right into the systems they need. Now, a global workforce of 300 ,000 can use AI to fill their HR questions, resolving 94 % of common questions.

24:17not noise proof of how we can help companies get smarter by putting ai where it actually pays off deep in the work that moves the business let's create smarter business ibm it's time to plan ahead and make sure your brand is showing up in ways that can have an impact that's where 4imprint comes in 4imprint have promotional products that work as hard as you do durable useful and designed to make a lasting impression think quality apparel your team will wear again and again including popular and exclusive brands, drinkware that's enjoyed again and again, bags, notebooks, tools, and tech items that don't just look good, but actually get used.

24:56With thousands of customizable options, 4imprint makes it easy to find what fits your brand and your budget. You'll get expert help, free samples, and their 360-degree guarantee means you can be 4imprint certain your order shows up just right, right on time. Whether you're gearing up for fall events or simply planning ahead for the season, 4imprint can help your brand show up, stay useful, and make connections that last. Explore the possibilities at 4imprint.com. 4imprint. For certain.

25:30It's time now for Talking Tech. I'm Yahaira Anand. First up, OpenAI's chief scientist is warning that AI is evolving so rapidly that it is becoming increasingly difficult for humans to understand and control. In a post over the weekend, he says that this time Cook calls for, quote, extreme caution, adding that he expects AI labs to voluntarily slow development for safety reasons. Plus, Australia is considering another swing at social media platforms. A proposed bill would let users opt out of platforms algorithms and choose what appears in their feeds. Last year, the country instituted the world's first ban on social media for kids under 16.

26:13And Meta has run more than 300 ads on its platforms containing suspected AI-generated child sexual abuse material just this year. That's according to a new report by the non-profit Tech Transparency Project. The ads typically promoted deepfake nudify apps. A meta spokesperson said many of the ads had already been removed and the company continues to strengthen detection.

26:38Riley Griffin:Riley. Thanks, Yahaira. Coming up, SpaceX may be looking at a boost in its NASDAQ weighting following the index's quarterly rebalance. We have more on that next. This is Bloomberg Tech.

27:14Riley Griffin:Welcome back to Bloomberg Tech. I'm Riley Griffin in for Ed Ludlow. Let's take a look at the markets. Bloomberg's Carmen Reinicke joins us now. Carmen, what are you looking at? All right. Thanks, Riley. Let's take a quick look first at shares of Qualcomm, one of the top tech stock movers in the market today. So the stock is having its best day in more than a month after signing up Amazon as a data center chip customer. So this is a deal that will span multiple generations and allow Amazon to buy$4 billion of Qualcomm stocks. So we're seeing that higher today. Shares of Amazon at the same time down about half a percent.

27:51Let's go to SpaceX as well. Everyone's watching this stock all the time and we're certainly no exception. So a wave of buying set to hit here for SpaceX as the NASDAQ 100 rebalances later this month. The reason for this is that the company will have a larger weighting in the index due to share lockups that have expired since it was added to the index shortly after its IPO. So according to a JP Morgan estimate, this could lead to a wave of more than$12 billion in net passive buying for shares of SpaceX. one of the top things that will be on our radar, especially as there's another lockup expiration hitting later in the week.

28:31Riley Griffin:Back to you, Riley. Bloomberg's Carmen Reinecke, thank you. Another stock moving today, ASML. Take a look at its ADRs. Samsung and TSMC have agreed to begin adopting the Dutch firm's latest semiconductor production gear. And together, the companies will work towards a critical shift that should speed up chip production. Bloomberg's Peter Elstrom is here to give us the details. Peter, these are machines that can cost up to$400 million each. What do they actually do? So ASML makes the machines that help companies like TSMC, Samsung, and Intel produce semiconductors. They make them for NVIDIA and Apple and, of course, the memory chips for the companies themselves, too.

29:16And what ASML has been doing is has been trying to persuade customers to make this transition from their existing EUV technology to the high NA EUV machines that cost$400 million, as you mentioned. The current generation is maybe about half that cost. And there have been a couple of customers that have gone early. Intel is the leading example of that, where they've jumped into this high NA technology. But what we heard today is commitments from both Samsung and TSMC to make that transition. Samsung is going to begin using these new machines in 2028. TSMC is going to begin using these machines in 2030.

29:53And the goal is that they are going to move to more efficient, lower cost production of the most advanced chips on the market. We, of course, have been chip constrained in a number of different areas. We've heard this from NVIDIA and Jensen Wang himself, that if only they could get more production, they'd be able to satisfy the market's demand. And that's also true on the memory chip side, where Samsung and SK Hynix and Micron have all been struggling with these technologies. So this is part of a broader collaboration agreement where the companies want to move from photo masks, which is sort of a technical term.

30:26It's the blueprint, essentially for a chip, as they print patterns onto silicon. They want to move from the 6-inch format to a 12-inch format. That way they're going to be able to reap greater efficiencies and lower costs from the production systems.

30:40Riley Griffin:And Peter, why are we seeing the world's biggest chip makers make this decision, this shift together right now? Well, it has been a challenge in the past. They've talked about this sort of transition from 6 inches to 12 inch photo mass in the past. Any one company trying to do this on their own would be very difficult. The companies are now doing this together, doing it in lockstep with each other to be able to have the benefit. So ASML can make the move with the other companies. They're going to work with TSMC, Samsung, Intel. SK Hynex came in late and said they're also looking at making this transition themselves, both to the high-end A machines and also to these 12-inch masks.

31:19So as a group, they figure they're able to make this transition a bit more smoothly than they could if they were trying to do it individually.

31:27Riley Griffin:Bloomberg's Peter Elstrom, thank you for the update. Recent earnings from NVIDIA, Broadcom, and Dell underscore just how strong demand for AI infrastructure remains. But investors are also getting more selective as the buildout runs into constraints around power, memory, and capacity. Here to discuss is Wealth Enhancement Group Portfolio Consulting Director, Ayoko Yoshioka. Ayoko, tell me, now that you've had some time to process the earnings, what are we looking for the back half of this year from these tech companies? Sure. Hi, Riley. So, you know, we think that just given the power of the AI infrastructure build out, we continue to expect earnings growth to remain at these elevated growth rates going into 2027.

32:12There's just too much demand and, you know, compute is supply constrained. And so we continue to expect these high growth rates out of many of these tech companies.

32:21Riley Griffin:And despite promising earnings results, tech companies haven't seen their valuations rise accordingly. Why is that? Yeah, you know, I think one of the things is that this time around, investors are very risk aware and very aware of that earnings growth second derivative. And we're waiting to see when that deceleration in that growth rate is going to happen. Do we go from 60 percent to 40 percent? It's still great growth. It's just that that growth rate is decelerating. And unlike the tech bubble in 2000, we don't have really elevated valuations this time around. We're still looking to see revenue come in for many of these AI companies.

33:04Riley Griffin:In the interim, a lot of attention has been on AGI, artificial general intelligence. Could you give us a state of play about where the companies stack up right now and what you're watching for? Sure. There's so much in regards to AGI or overall artificial general intelligence or intelligence that is really on par with human intelligence. And I think a lot of news over the weekend really talked about how AI is really improving upon itself. And this whole recursive self-improvement that you're seeing in AI, building better models. I think at the same time, you know, as you've seen with GBT Astra or, you know, Anthropics mythos, is that there is some sort of caution related to just the overall rollout and how quickly we can get this to as many people as possible.

33:56I think there's a little bit more restraint when it comes to the overall governance related to how powerful these models are.

34:02Riley Griffin:How is the market interpreting that caution, those checks and balances being put in place by the companies more so than the government? I think it's being viewed positively in that we need to make sure that, you know, the benefits for AI to just broader humanity are really highlighted. I think OpenAI has talked about that as well in terms of, you know, making sure that this is a benefit for all of us, that it's not the big, scary, you know, job loss that everybody expects that AI will take over everybody's jobs. And among the suite of AI companies, all eyes right now are on Anthropic as it prepares that S1, the document that will let people look under the hood and see what the company has been up to in advance of its IPO.

34:50Riley Griffin:What metrics will you be looking for when that document drops? Sure. I think from our perspective, just the overall business model and making sure, you know, what's the run rate on the cost? I think we've gotten a lot of information on the ARR and the overall revenue pace, but we want to know how much they're sort of burning in terms of, you know, cash and, you know, looking to see what the improvements are going forward, what their expectations are. And as they proceed developing these models, we're also thinking about the physical infrastructure of this AI boom. What bottlenecks are you watching for as the buildout continues?

35:33Sure. There's so many. I mean, we have some bottlenecks when it comes to advanced packaging. We've got bottlenecks that are now really being focused on the overall power generation for these data centers and just the data centers themselves. You know, there's so much backlash, I think, in certain communities in regards to the build out here. And so we really have to watch out whether or not the timing of the physical build out of AI, it creates a mismatch in terms of investor expectations since these stocks trade every second.

36:06Riley Griffin:Bloomberg today reported that Anthropic did not move forward with a deal for Descartes AI, a company that was seeking to make its compute work more efficiently. This question of efficient compute is an important one. What do you make of that deal not moving forward in advance of the IPO? Sure. You know, I think when it comes to whatever happened with the M &A, I'm sure there are reasons on both sides why this wasn't completed. But when it comes to the efficiency of AI, I think it's something that everybody's paying attention to. We're looking at the cost per token, which has declined substantially over the last several months.

36:45But the rental sort of prices for GPUs when it comes to model training has increased even for some of the older NVIDIA chips. And so I think it's a balance in terms of what the cost is per megawatt for a job to be completed. And then you can start to do the analysis as to whether or not, you know, a human task can be completed at a efficient rate relative to an AI task.

37:12Riley Griffin:Amazing. What a breadth of knowledge. Thank you so much, Ayako Yoshioka of Wealth Enhancement Group. Appreciate it. Coming up, we'll speak to the CEO of Forrest about the AI healthcare startup's latest funding round. This is Bloomberg Tech.

37:41the people who seem to get more done than everyone else they're not working longer hours or running on more caffeine they've just stopped wasting time on the stuff that doesn't move work forward Switching apps, re-explaining context, hunting for files. Those aren't small inefficiencies. They're hours wasted every week. Superhuman Go gives you those hours back. From the makers of Grammarly, Go is an AI assistant that sits inside every tab and tool you already use. Always available and already aware of what you're working on. Ask it to draft something, summarize a long thread, pull up a file, or prep you for a meeting.

38:22Go handles it without you ever leaving the page you're on. This is what it looks like when AI actually fits into your work, instead of adding to it. It's like having a teammate whose only job is to help you be better at yours. Go keeps up so you can move forward. With Go working with you, you can show off what you do best. Superhuman Go. Find out more at superhuman.com. That's superhuman.com. The thing about AI for business? it may not automatically fit the way your business works. At IBM, we've seen this firsthand. But by embedding AI across HR, IT, and procurement processes, we've reduced costs by millions, slash repetitive tasks, and freed thousands of hours for strategic work.

39:08Now we're helping companies get smarter by putting AI where it actually pays off, deep in the work that moves the business. Let's create smarter business, IBM. It's time to plan ahead. and make sure your brand is showing up in ways that can have an impact. That's where 4imprint comes in. 4imprint have promotional products that work as hard as you do. Durable, useful, and designed to make a lasting impression. Think quality apparel your team will wear again and again, including popular and exclusive brands. Drinkware that's enjoyed again and again. Bags, notebooks, tools, and tech items that don't just look good, but actually get used.

39:46With thousands of customizable options, 4imprint makes it easy to find what fits your brand and your budget. You'll get expert help, free samples, and their 360-degree guarantee means you can be 4imprint certain your order shows up just right, right on time. Whether you're gearing up for fall events or simply planning ahead for the season, 4imprint can help your brand show up, stay useful, and make connections that last. Explore the possibilities at 4imprint.com. 4imprint, 4certain.

40:20Riley Griffin:Healthcare automation startup Forrest has raised$150 million in a Series C funding round, tripling its valuation to$3 billion just months after completing its last fundraise. It's the latest deal to show investors rising interest in AI tools aimed at speeding up patients' access to medicines. Forrest CEO, Sahir Jaggi joins us now. Sahir, Tell me, this wasn't initially planned as a fundraise. Why did you move forward with it? Yeah, I mean, so the reason we're building our company is because you're seeing this huge explosion in AI accelerating drug discovery, but especially for complex conditions like autoimmune diseases and cancers, but it still takes decades and billions of dollars to turn a molecule into a production medicine.

41:07And once those medicines are in market, doctors and patients often spend weeks or months working through insurance coverage, affordability, supply chain to get their medicine, with over one in three patients never even getting their first dose. And the reason that exists is because the system is so complicated that any individual physician or patient can't really navigate through it without support. And what we built over the last couple years is this platform where we're able to help individuals get their medicine on time and affordably without any of the burden falling their shoulders.

41:36Riley Griffin:And what does the new funding allow you to do that you couldn't have done six months ago? So we have been working with real doctors and patients for a couple years now. We are being used across all 50 states, reaching patients across 85 % of zip codes. And we have thousands of providers who are using the platform every day to help their patients get on therapy. What we've been doing with this has been focused initially on a first set of specialties, really helping doctors and patients in a select set of conditions and a select set of geographies. And with this capital, it's really going to allow us to expand across all different specialties of medicine, all types of care.

42:11So ranging from clinics to hospital systems to infusion centers and making sure that we can make our platform available to every doctor's office in the country so every disease is treated with the best medicine available.

42:21Riley Griffin:And what specialties have you started initially focusing on? What learnings are you taking away from those specialties? Yeah, so we started in the specialties focused around autoimmune diseases, right? These are medications and conditions where individual patients are seeing flare-ups and issues that are not easily treatable by traditional generic medications and require more complex drugs and a lot more care around kind of the diagnosis and treatment decisions. And so those are specialties like dermatology, gastroenterology, rheumatology, allergy, places where you're really trying to navigate what specific issues are underlying someone's symptoms and treating them with more and more complex treatments until you find success.

42:59And we felt that those areas were really underserved by the market and places where both the physicians and the patients felt like they had a lot of need. What we've learned from those spaces is that there's not just complexity for the medications that are most expensive, but really all medications that have some sort of additional process around the insurance coverage process, around financial assistance and affordability, around supply chain and navigating fulfillment. And as we've kind of gone through those specialties, you see different kind of situations where patients are not just taking medications that they pop in as a pill at home, but even medications that they have to go into the clinic and get as an IV in their arm.

43:38And working through all of that has really prepared us now to expand into other specialties like oncology, cancer centers, neurology, you know, treating patients with Alzheimer's and multiple sclerosis. These cases where the diseases are even more acute have more complexity around them.

43:51Riley Griffin:With these complicated drugs, these biologics, companies have an incentive to keep patients on them or get them on them in the first place. Revenue comes to this company from the pharmaceutical companies. How do you build trust with patients who are skeptical about big pharma and the gaming of the system to ensure that they get on those drugs? Yeah. Well, our first focus for most is ensuring that people get the therapy that they need. What we learned when we started spending time initially in the space and meeting folks was that many patients with these conditions are actually desperate for help.

44:24They've been with these conditions for years. They've been struggling to get treatment. And when their doctor finally provides them with something that could be a cure or something that could bring their condition into remission, it's so challenging to get it that it's actually really hard for them to stay on it through their full course of treatment. You have people who are going through the process on an annual basis, having to renew their access to medication. And every time they change jobs, every time they move, every time their insurance company tries to change policies, they have this full process they have to go through again from scratch with totally different rules.

44:56And so they're so thrilled when there's some product or tool that can help them do that easier and faster. And we're able to make it so that they can get this medication without that complexity bleeding through to them. Now, as you mentioned, we partner with life sciences companies to help them develop and bring new therapies to market faster and more efficiently. And on one side, you might think, you know, as you mentioned, there's some fear around big pharma. But what we really see is that the patients with these conditions are the people most, you know, most excited about these companies being involved in more medicines.

45:28In fact, their primary goal is actually I hope that more companies are investing in more drugs to make more progress for people like me. Right. Because if you have this disease, sure, you might have some things you've read in the news. But most important is that the science is actually accelerating towards a point where your condition can be treated, where you can actually get the cure and treatment that you need.

45:45Riley Griffin:And you've come on today to Bloomberg Tech. Tell us a little bit about the AI implementation that's taking place in your business. How is AI facilitating this better process? So the way that we actually provide the product that we do to patients and physicians is by giving every prescription its own AI agent that navigates through the process on their behalf. So think of this almost like an AI case manager, right? It's taking into account each individual patient's medical history, their financial situation, their insurance coverage, and using all that to govern how it navigates through the supply chain for them, right?

46:19These are situations where every single patient, every payer, every drug, every pharmacy, has different constraints, different factors you have to understand. And so it really requires the complexity that an agent can handle, dipping into all sorts of tools, all sorts of data sets, and relying on clinical models, specialized subagents, and millions of previous cases to navigate through each individual new process correctly.

46:45Riley Griffin:Sahir Jaggi, CEO of Forrest, big day for you. Thank you so much. Thank you for having me. Coming up, Apple is ready to unfold its next chapter with the first foldable iPhone expected to debut tomorrow. More on that next. This is Bloomberg Tech.

47:17Riley Griffin:Apple is gearing up for one of its biggest product launches in years. Tomorrow, the company is expected to unveil its first foldable iPhone, a device that has been a decade in the making and which could cost as much as$2 ,199. It also marks the biggest design change to the iPhone in nearly 20 years. Bloomberg's Apple reporter Mark Gurman joins us now. Mark, I wonder if you could walk us through some of the history behind the design of this new product. Yeah, thank you for having me. This is a really big deal for Apple. It's a really big deal for the industry. Foldable phones have been around on the market since around 2019 when Samsung launched the first Galaxy Fold.

47:58But Apple's work on the product actually predates that. About a decade ago, their core technology teams, these are sort of research groups within the hardware team at Apple that looks at future technologies, started playing with foldable screens. And the original idea was an iPad mini that could unfold to be a larger iPad. But Tim Cook goes to Asia around 2020, goes to China, goes to Korea. He sees Huawei. He sees Samsung. He sees everyone using these phones. And he comes back inspired and enthusiastic, unusually so, I may add, about foldable phones and asks Apple, what's up? Where's our foldable phone?

48:36And so they pivot from the iPad to a foldable phone. And around 2023, they had their aha moment, which was this short and wider design, more interesting, better experience for watching video. It's an aspect ratio known as four by three. It's basically the same aspect ratio you get with an iPad in landscape mode versus traditional foldable phones had a more squarish design. And so now this is what Apple is going to introduce tomorrow on Wednesday at its Surprise and Shine event in Cupertino. And we'll also see new Apple Watches and likely new AirPods as well. So this should be a pretty blockbuster introduction for new CEO John Ternus.

49:17Riley Griffin:And are you anticipating high consumer demand for a product like this, given the broader market that you just outlined? You know, Apple is probably going to make 7 to 10 million units this year. And the expectation of essentially everyone is they're going to be able to sell out of every unit they can make. These are extremely complex products to build and manufacture. The display technology is very advanced. The hinge is very advanced. Getting the batteries and cameras and everything in there is a very complex process, much more complicated than building your traditional slab or candy bar form factor phone, which, by the way, there'll be new ones of those as well with the 18 Pro and 18 Pro Max.

49:54But in terms of what percentage of all new iPhones will this phone be, it's going to be slim. The foldable market is about 2 % of the United States at this point. But in China, it's actually double that. And so this is going to be a very important phone for the Chinese market, which has obviously been a little bit of a roller coaster for Apple right now. It's on a bit of a high, which is a good thing. And the hope for the company is that the foldable will keep them there.

50:19Riley Griffin:Bloomberg's Mark Gurman with all the latest. We're watching tomorrow closely for the Apple event. And stay with Bloomberg Tech tomorrow. Ed Ludlow returns live from Apple headquarters. He will be joined by Carolina Milanese of Creative Strategies, Nabila Popal of IDC, and more. That does it for this edition of Bloomberg Tech. Don't forget to check out our podcast. You can find it on the terminal as well as online on Apple, Spotify, and iHeart. This is Bloomberg. If you listen to financial news, you know a lot of time is spent thinking about what's next. The next opportunity. The next investment.

50:59The next move. But sometimes what matters most is being ready for what you never saw coming. For more than 75 years, Cincinnati Insurance has worked with independent agents to help protect businesses, homes, valuables, and more. Because planning for the future isn't only about knowing what's next. It's about making sure you're ready for what you can't predict. Let Cincinnati Insurance make your bad day better. Find an independent agent at CINFIN.com. game night rush, or any night of the week, really, Genius keeps every order moving. From online ordering to your kitchen to the front counter. Big League reliability for any business.

51:41That's Genius. Before you sign off, you tuned in for ways to help teams move faster, make sharper decisions, and turn scattered context into work they can use. ChatGPT for Business can help. ChatGPT for Business gives teams a shared workspace with admin controls, permissions, and access to work and codecs in ChatGPT. This means your business can move from question to answer and code to rollout quicker. Join over 10 million business and enterprise users worldwide already using ChatGPT for work. Download the ChatGPT desktop app or contact sales to learn more.

From the publisher

Bloomberg’s Riley Griffin breaks down Qualcomm's deal signing up Amazon as a data center chip customer, an agreement that will span "multiple generations." Meanwhile, another deal falls apart after Anthropic decides against acquiring AI startup Decart AI; and all eyes are on Apple as the company kicks off its product launch tomorrow, with its foldable iPhone a decade in the making.

See omnystudio.com/listener for privacy information.

More from Bloomberg Tech

All 342 episodes
Qualcomm Signs Deal to Provide Amazon With Custom AI ChipsBloomberg Tech · 44 min
Listen in VO