Nvidia’s Big Test as AI Chip Race Heats Up

19:33Ed Ludlow:that demand ultimately came to bear. And that created a lot of disruption, a lot of default and issues in the credit markets. Okay. Thank you, Hyra, for talking tech. Coming up on the show, Gatic, the startup building autonomous freight raises$200 million in its latest funding round. Gaddick's CEO and co-founder Gautam Narang joins us to discuss. This is Bloomberg Tech.

19:58The 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. their 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 go 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.

20:48It'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. This is the Bloomberg Tech Minute brought to you by ChatGPT. Now with ChatGPT Work. I'm Carol Masser. Bill Gates Breakthrough Energy Ventures is betting on a cheaper way to produce clean jet fuel and made its first investment from a fund that is backed by Alaska Air, American, and several other carriers. The investment is aimed at accelerating the commercialization of clean aviation fuel.

21:30The recipient is Boston-based Lydian, which is developing lower carbon jet fuel made from hydrogen and carbon dioxide and is one of a growing number of companies developing next-generation clean jet fuel technology. Lydian says it can reduce capital expenses by more than 50 % compared with competing technologies. Lowering those costs is seen as crucial to making sustainable aviation fuel, or SAF, commercially viable, as it remains far more expensive than conventional jet fuel. Today, clean fuels represent a tiny fraction of the overall market. That's the Bloomberg Tech Minute brought to you by ChatGPT.

22:09Put ChatGPT to work on your most ambitious ideas and projects. Get started at ChatGPT.com today by selecting Work Mode. Available on Plus and Pro Plans. 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.

22:49With 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.

23:19Ed Ludlow:Autonomous trucking startup Gatic raised$200 million in its latest funding round. It's largest ever, led by deep-pocketed investors, including the Qatar Investment Authority and Coke Disruptive Technologies. Gaddix CEO and co-founder Gautam Larang joins us now. Really interesting list of participants, the most money you've ever raised. What's the biggest story, the capital needs, or those that decided to back you? Yeah, great to be back. The capital is mainly for expansion, right? So most recently, we announced our partnership with PepsiCo. Right. That partnership is multi-years. We started working with Pepsi back in 2022.

23:54Since then, we have scaled our operations to more than dozens of driverless trucks across multiple markets in Texas, Arkansas, and Arizona. And in addition to PepsiCo, we have long-term commitments from some of the largest retailers, grocers, and CPG companies. So this capital, our$200 million raise, is towards that growth.

24:15Ed Ludlow:What's the main focus? What's the priority? You just need more trucks? Well, scaling our operations, right? So when I was here earlier this year, I mentioned that we have over$600 million in contracted revenue. Right. That number is... It's basically a backlog. Backlog. That backlog has increased significantly since then. We're actively adding new customers as well. And yes, the capital is towards scaling our footprint. We expect to have hundreds of driverless trucks towards the end of this year. And the plan is to continue scaling that to thousands and beyond. The basics of it are trucks with a suite of sensors, local compute in part provided by NVIDIA, and then software that you guys write yourselves.

24:58Ed Ludlow:Operationally, it sounds like the challenge is just more getting out there. How much do you feel like you've solved the technology? That's fair. So it's an execution story from here. The core technology for highway driving and surface streets is already in place. Obviously, with the contracted revenue, we have the customer engagements in place as well. In addition to that, we have a very strong ecosystem of strategic partners. This is Isuzu, NVIDIA and others that are enabling that scale up. With this financing, you're right. Our focus is to keep our heads down and get more trucks on the road.

25:29Ed Ludlow:Qatar Investment Authority. You know, right now you're focused in U.S. cities, particularly U.S. states where regulation is supportive of this, right? Is that a signal that you guys will look to other markets like the Gulf overseas to expand? So, you know, QIA has been a great partner alongside K2T. And then we also have Kathy from ARK in the round. In my discussions with her and the team, they're super bullish on this. Our focus is currently on North America. The opportunity here is huge, right? We are talking about over$300 billion in the time that we are going after. Obviously, there is pull from international markets as well.

26:06But the current needs of the customers that we have locally, that's what we remain focused on.

26:10Ed Ludlow:Really interesting economic discussion to be had, right? On a dollar per mile basis, how does Gattic stack up against a human-driven? Fair. So when it comes to the business model, transportation is a service. One of the things that is the highlight of the business is each of the customers make multi-year take-or-pay commitments. The other aspect of the business is we are completely asset-light. So the base vehicle and the AV equipment is not on our balance sheet. But from the customer's perspective, they're faced with how much does it cost me to move this good? Exactly. Because for them, it's more about meeting that increased capacity.

26:45All the customers, they are looking to increase the capacity. The freight volume is expected to increase to over 50 % by 2050. So that's where we come in. That's the capacity needs that we are fulfilling for customers like PepsiCo and others.

26:58Ed Ludlow:I know that we're having a very finance focused conversation, but what are the margins like for you? You know, once you've done the capital expenditures of getting the truck ready and it runs, how does that go? So one thing to highlight is every delivery that we have done is a paid delivery. You know, these are some very exciting numbers, right? So the fact that we have hundreds of millions of dollars in take or pay commitments, and that number keeps on increasing, the base fundamentals of the business are super strong, right? So with the scale that we have in mind, especially with the milestones of driverless operations at scale.

27:32Profitable strong? Very strong fundamentals, right? But so I do want to highlight the asset light aspect of the business model. So that's where the long-term sustainability comes in.

27:44Ed Ludlow:Gaddix CEO and co-founder Gautam Arang back on the program. Really great to have you. Thank you very much. Coming up, OpenAI is making some pretty spicy claims about the performance of its jalapeno AI chip. We're going to speak with the company's hardware VP, Richard Ho, about that next. It's half time. This is Bloomberg Tech.

28:12Ed Ludlow:Welcome back to Bloomberg Tech. Let's get back to that news that crossed earlier this hour. Canada announcing new tariffs ranging from 15 to 50 percent on$20 billion of U.S. goods. Bloomberg's Washington correspondent, Tyler Kendall, has the details. Hi, Tyler. Hi, Ed. Well, Canadian Prime Minister Mark Carney making good on his threat to go dollar for dollar in terms of tariffs and retaliation for the U.S. Moving ahead with those tariffs earlier this week against$20 billion worth of Canadian goods amid the stalled talks for a trade deal. Canada will now double its existing counter-tariff on steel and aluminum to 50%.

28:50We'll also see a new 50 % tariff levied against American-made furniture, clothing, and electronics. There's then these lower tiers of tariffs, including 25 % on a range of U.S. goods, including appliances, cheese, certain dairy and fish products, as well as importantly, derivatives of U.S. and steel and aluminum. That was something that we were watching really closely. These are going to take effect on September 8th. Perhaps that leaves a little bit of leeway here for negotiations, keeping in mind that this White House's next threat of 50 percent tariffs on autos is set to take effect on January 1st.

29:25We'll have to see where it happens, because today we've already heard from President Trump multiple times really criticizing Canada when it comes to these negotiations. Ed, I will also add that we've seen Canada now roll out an aid package to help those businesses expected to be impacted by these tariffs that does signal that the country could be in this for the long haul.

29:45Ed Ludlow:That's Tyler Kendall from the White House. Thank you very much indeed. OpenAI is out with performance results for its first custom inference chip called Jalapeno. The The ChatGPT maker says in TES, its chip was faster and more efficient than NVIDIA's GB300 system. Richard Ho, OpenAI's vice president of hardware, is with us here in San Francisco. Let's start with the basics of the testing and the performance benchmarks that you've published. What does it tell us about Jalapeno so far? First of all, thanks, Ed, for having me here. The benchmarks that we did were with an open source kind of benchmark called InferenceX.

30:20And we did that because it was a fair, kind of neutral benchmarking method. What we're seeing with Jalapeno is that we're able to get both high throughput and low latency with the same device, which is kind of a first in the industry. And why does this matter is because high throughput means that you can serve a lot of customers with a lot of tokens much cheaper than you could if you otherwise. Low latency gives you very fast response times. And so, you know, the agents will come back faster. you can do more coding for her. And that really, I think, is a real benefit for our products.

30:53Ed Ludlow:One of the questions that we got immediately was, why didn't you benchmark against Vera Rubin, NVIDIA's latest system that's currently in full production? I guess the answer is quite simple, that it's just come out. It's just come out. We wanted to use numbers that have been published, that have been optimized. Those were available, the GB numbers. The other chip makers also have the numbers in there. It just happens that GB is the best, indeed, product in the current benchmarks that are there. This is it. This is Jalapeno. You very graciously allowed me to get my hands on it. And again, the whole emphasis is on throughput and ultra-low latency.

31:30Ed Ludlow:And actually, in the ultra-low latency piece, you'd associate that with SRAM-based systems. This is both HBM and SRAM. Talk a little bit about its design and architecture, but why that ultra-low latency matters. Yeah. So it is an HBM-based design. It's HBM4. It's one of the first products that's going to go to volume with HBM4 right behind NVIDIA's products. And it is very low latency. The reason how we got it was really taking a blank slate approach to this design. So my team arrived at OpenAI, and we basically were able to work with the research team to understand where the bottlenecks were in large language models.

32:10And with a blank slate design, we figured out that you had to reduce your data movement a lot, and we figured out a way to do it. And so this architecture is different from GPUs, is different from TPUs and other chips in that nature. It really reduces the data movement. It optimizes the algorithm. And then it basically is able to get this very low latency, which matters for the agentic workloads, for example.

32:33Ed Ludlow:We're getting to why a frontier lab would look to a custom silicon program, the why. What is it that you can optimize for with Jalapeno or any custom ASIC or accelerator that you just can't with an NVIDIA-based system, for example? It's the full stack control that we're really looking for, all the way from the models, through the software, all the way down to the silicon. By having that control, we can make trade-offs. We can make trade-offs and optimize it really well. And that's something that's unique from being inside an AI frontier lab, that this trade-off can actually be made all the way down into the silicon.

33:08And with that, right, so I think that the key thing here is that this is quite a novel architecture. I don't think other people have done it this way before, where there's affinity between the memory and the cores in a very integrated way. One of the worries has always been, is this programmable? and new architecture is sometimes it's hard to get the software up but we're able to show because of how fast we were able to bring these three very different models that were in the benchmark up and running on this new hardware which we only got a couple of months ago is indication of one how good the programming model is and two how good our models our AI models have gotten in optimizing the firmware, the kernels in order to make this performance possible here

33:54Ed Ludlow:This is an inference chip. It is. So very simply, why focus on inference, not on the training side with your custom silicon program? Well, I think that if you look at it, right, OpenAI has this enormous need of compute. And where that growth of compute is, is really on the inference side. We need a lot of training and we have a really strong partner in NVIDIA who has really got a great training chip. So we focused on where we thought we needed to increase the amount of compute that OpenAI has available. With the growth in our users, with our weekly active users growing very, very fast, we saw the need to continue to have a wide range of inference products.

34:29And this is going to be part of our fleet, along with our existing partners who are providing us other chips.

34:36Ed Ludlow:How do you decide which workload goes to Jalapeno versus NVIDIA or Cerebris, for example? You have options, but how do you make that decision? It's a great question, right? I think with this optionality that we're now creating for the company, I think that we will be able to provide users with a range of models that have different cost benefits and different latency capabilities. We will map that ourselves. We will map out which SKU, which chip is best for which one of these models. But the whole goal of this is to deliver more intelligence to our users at a lower inference cost, at a lower infrastructure cost.

35:15And that's what this is part of, that entire strategy that, you know, we've been working on for a while now.

35:20Ed Ludlow:You know, lots of people struggle with the benchmarks. You know, to the layperson, it's difficult to decipher. So in very simple terms, on a dollar per token basis, how does jalapeno stack up against the other options that are out there? How has this changed the economics for you? It should drop. It's a really good chip. It should drop it by a lot. The performance per watt that we're seeing, which is a good proxy for how much token will cost, is somewhere in the region between 1.8 to 4x better than other existing chips right now. And so we should see that ultimately when this rolls out into production, customer token prices should come down.

36:01That's a big drive that we've had at OpenAI for a little while now. And this really is part of that strategy. And again, it goes back to attacking not only the software, but attacking all the way down into the hardware and even into the data centers as well.

36:14Ed Ludlow:You are the VP of hardware, not just a single chip. Talk about the challenge of scaling Jalapeno. You do this in partnership with Broadcom, right? You know, there are still bottlenecks in high bandwidth memory. Ramping output takes time on the fab side. You know, what's that like? Yeah, totally. I mean, yes, Broadcom was our big partner in the chip design. Seleska is our system partner who basically are doing the racks with us. This is the big challenge today, right? Memory is highly constrained. Wafers are highly constrained. We rely very heavily on our partners. We also take control of that and have direct relationships with many of the vendors that we operate with.

36:53And I think it's one of those things where I think we have to work together to basically be able to ramp up the volume and get us into high volume production. Richard, two rapid fire questions.

37:01Ed Ludlow:versus percentage, how much content does OpenAI own of the server design? We did the entire design. Broadcom has done the physical design, which is like after you take the logic design, you turn it into the masks that TSMC would manufacture, for example. So that's where the separation is. On the system side, we did the design, and Celeste co-designed it with us. I ask because these days, NVIDIA is pushing 55 % of the ownership of the content of the server design. In the future, how much proportionally of your inference workloads do you see being run on custom silicon versus third parties? I think if we can ramp it and if the cost profile looks good, if it's good as I say, I think we'll see a lot more of it coming up.

37:46And I think the other thing I want to highlight is it's a multi-generational roadmap. So we're already deep into development of our second generation of this device, and we're already starting concept work on our third. So we see this as a long-term bet to lower the overall infrastructure cost for OpenAI.

38:01Ed Ludlow:Richard Ho, Vice President of Hardware OpenAI. First time on Bloomberg Tech. I believe your first time on television. Thank you so much for being here. Thank you so much, Ed. Bringing Jalapeno with you. Again, you know, NVIDIA reports quarterly earnings after the market close on Wednesday. The stock is rebounding. It's off-session highs right now, but it's had seven straight days of declines, matching its longest losing streak since September 2022. Pre-Chat GBT moment. The expectation,$92 billion of sales double a year ago, double adjusted earnings from a year ago as well. Bigger picture is what's happening in the ecosystem.

38:35Ed Ludlow:Coming up, we're going to speak with Noam Schwartz, the CEO of security startup Alice, about the company's latest fundraise and how it's helping Frontier AI Labs safeguard against rogue models. This is Bloomberg Tech.

38:52The 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.

39:33Go 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. This is the Bloomberg Tech Minute brought to you by ChatGPT. Now with ChatGPT Work. I'm Carol Masser. Bill Gates Breakthrough Energy Ventures is betting on a cheaper way to produce clean jet fuel and made its first investment from a fund that is backed by Alaska Air, American, and several other carriers.

40:19The investment is aimed at accelerating the commercialization of clean aviation fuel. The recipient is Boston-based Lydian, which is developing lower-carbon jet fuel made from hydrogen and carbon dioxide and is one of a growing number of companies developing next-generation clean jet fuel technology. technology. Lydian says it can reduce capital expenses by more than 50 percent compared with competing technologies. Lowering those costs is seen as crucial to making sustainable aviation fuel, or SAF, commercially viable, as it remains far more expensive than conventional jet fuel. Today, clean fuels represent a tiny fraction of the overall market.

40:59That's the Bloomberg Tech Minute brought to you by ChachiPT. Put ChachiPT to work on your most ambitious ideas and projects. Get started at ChachiPT.com today by selecting Work Mode, available on Plus and Pro plans. 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.

41:38Bags, notebooks, tools, and tech items that don't just look good, but actually get used. With 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.

42:17Ed Ludlow:Talks between Paramount and California over the company's planned takeover of Warner Brothers are on hold. State Attorney General Rob Bonta has accused Paramount of leaking information about the negotiations to the press, something CEO David Ellison denies. But speaking on Bloomberg's balance of power, A.G. Bonta said talks could resume and laid out the types of changes he wants the company to make. Well, we've always said that we want structural remedies, robust structural remedies. I've said that publicly and Paramount knows exactly what that is. Whether they will propose them or agree to them is up to them.

42:50But they've continuously proposed behavioral remedies, like promising to behave in a way in which they make 30 movies a year, I think for three years. That's a behavioral remedy. It's not structural. So we've identified three markets in our complaint with respect to widespread theatrical release distribution, blockbuster release distribution, and cable channel licensing. And if they want to propose a structural remedy for each of those markets, we'll listen. Last time we had a chance to speak, Attorney General, it was on the eve of the trial against Meta, starting something that you have helped to spearhead as one of four attorneys general.

43:27How is the trial proceeding? And we'll be here from Mark Zuckerberg. We think it's going very well. The evidence has been coming in, as we hoped, very compelling, providing all sorts of evidentiary support for our allegations, which I think are very strong, that Meta designed and deployed a dangerous product that hurt the mental health of kids, and they lied about it. And we do have Mark Zuckerberg on our witness list, and we're prepared to call him.

44:00Ed Ludlow:That was California Attorney General Rob Bonta speaking on Bloomberg's Balance of Power yesterday. As AI systems grow more sophisticated, we're seeing high-profile instances like those from OpenAI and Anthropic where autonomous agents break out of their testing environments and then breach outside organizations. The incidents have driven growing interest in technology that can rein in the misuse of AI tools. Israeli startup Alice is focused on that issue, and it's just closed a$140 million round to expand its efforts. Noam Schwartz, CEO and co-founder of Alice, joins us now. So interesting, the timing, the focus on this area.

44:37Ed Ludlow:Let's start with the round, what you need this capital to do. So, hello. It's incredible to be here. So, we raised$140 million, led by Apex Digital, to make sure that we can keep AI at bay, and we can make sure that models are doing only what we want them to do and not things that we don't want them to do. And we're going to use this capital to continue our research activities to make sure that we can always be at the forefront of everything that can go wrong and make sure that our customers, who are the frontier models and Fortune 500, can use AI and advance unafraid. Noam, recent events have been labeled as sort of watershed moments for AI.

45:22Ed Ludlow:But I assume that your company has some history that you thought about this idea before some of the incidents in the last couple of months. What was the sort of catalyzing concern that you had? So we started a company about eight years ago. And that was before AI was a thing. We were protecting, we're still protecting social media, big tech, against everything that can go wrong on the Internet. anywhere from fraud to abuse to manipulation to cyber attacks on individuals and companies. And during that time, protecting these companies, and if you use the Internet today, you probably were protected by something that we created.

46:01We gathered a massive, it's called like a database of harmful activities or things that can go wrong. and now we're seeing how those things that were used by individuals are now being amplified with AI. So we renewed and got bigger and now we're actually taking that knowledge and understanding how that can be leveraged with AI to cause harm to even more people, to even more enterprises and now we're protecting those kind of companies.

46:33Ed Ludlow:Noam, I won't go over every single detail that's happened in the last three months but basically two OpenAI frontier models escaped a sandboxed environment and breached a third-party platform in Hugging Face. Since that happened, how much has interest in your technology accelerated, literally in response? So the interest was always high, and that incident obviously created even more interest. And to understand what happened, the models effectively escaped a sandbox, which was a human error. It wasn't a model really escaping, and there's no reason to panic. But models are becoming more and more capable.

47:12They're becoming more capable in coding, in cybersecurity, in other topics, and they would be even more capable in creating harms around different topics. And our mission is to make sure that the research is at the frontier, that we're discovering everything that could go wrong, because you can't really protect against things you don't know that can happen. So our mission is to make sure that we're always one step ahead of bad actors and we can help our customers make sure that their models are doing exactly what they're supposed to do and not things they're not supposed to do.

47:47Ed Ludlow:We discussed the human failings with the CEO of Irregular and also with Clem DeLong of Hugging Face. You can find those interviews online. You work with Anthropic. So just as a case study, could you tell us about the work you do with them? So I can't refer specifically to the work we're doing with a specific company and everything that is out in the open. And there's a lot of writing about that and publications from their behalf. But what I can say that our work is to stress test models to figure out what could happen and break them and see that they're actually getting broken and provide the solutions for it.

48:22So without getting too much into the details and being a bit of oversimplification, we're helping them train the models in order for them to be safer. We're figuring out what could happen, and then we're providing the solutions to make sure that it will never happen. And we're working with the frontier labs and also with enterprises that are either using their products or open weights models in order for them to be as safe as the best models in the industry. because effectively we're seeing the market demands democratization of usage of AI. And right now there's democratization of harm as well when we want to make sure there's also democratization of protection and safety and security.

49:05And that's all what we're about.

49:09Ed Ludlow:Noam Schwartz, founder, CEO of Alice. It's great to have you on Bloomberg Tech. Thank you very much. Thank you very much indeed. Coming up, key software firms are reporting earnings this week and investors will be looking at whether the latest rally has any staying power. Those are the names that we're watching. That conversation next. This is Bloomberg Tech.

49:38Ed Ludlow:Software investors are betting that the struggling stocks have found their footing again after a strong performance over the past month. Those gains will be tested this week as the Market Digest earnings reports from key software firms, including Intuit, Salesforce and Autodesk. The tech equities reporter Ryan Veselic has been writing about it this morning. There are other earnings this week, but it's so interesting because the performance of software stocks, particularly since the middle of July, have been the opposite of the chip stocks. So now we're waiting to see what happens in earnings. Yeah, this has been a trend that we've been seeing all year, where people seem like they're very bullish on chips and then bearish on software, and then it flips, and it's flipped back and forth a few times.

50:17But at least right now, it does seem like there's growing optimism and improving sentiment towards the software side of things. We've seen some pretty resilient earnings, including from Microsoft and Palantir this quarter, and that's really helping this group recover in a pretty broad-based fashion, following a period where people were really concerned about what kind of impact is AI going to have on the group's growth outlook going forward.

50:43Ed Ludlow:The trade was long semis, short software, and then that changed. What was it that drove the at least psychological change of the market? Well, you had a lot of people who said these things have really just gotten decimated across the board. People are selling indiscriminately and finding a lot of value in the groups. And like I said, we did get some pretty encouraging results out of some high-profile software companies like Microsoft, like Palantir. Or if you go back to last quarter, you had names like Snowflake, Datadog, these kinds of infrastructure and consumption-based names that were really showing some pretty positive results and an indication that they were even benefiting from AI themselves.

51:24So that really had people looking at this group saying they've come down so much. Their valuations have really come off their historical levels. Maybe there's an opportunity here, especially since so far it seems like the earnings are still pretty resilient. We're not really seeing signs that AI is coming in here and eating their lunch.

51:44Ed Ludlow:The big catalyst this week is a chip company, NVIDIA. Recap what to expect. Well, this is going to be a very important sort of a temperature check for just the state of AI overall. People are expecting a pretty strong quarter here, if only because all the hyperscaler results that we got a couple of weeks ago, they really underlined how these companies remain pretty committed to pretty aggressive spending on AI. However, anything the company says about its outlook, what kind of adoption it's seeing, its pricing following the big price hike that was reported over the weekend, And that's all going to be very much in focus and sort of really help determine at least near-term sentiment for AI.

52:22And it wouldn't be surprising if we start to see AI sentiment improving if some of the software starts pulling back.

52:28Ed Ludlow:Bloomberg's Ryan Flostellica with the preview. Thank you very much. That does it for this edition of Bloomberg Tech. Check out the podcast. Find it on the terminal as well as online on Apple, Spotify, and iHeart. NVIDIA after the market Wednesday. This is Bloomberg. When you're running a business, the best days are the ones where priorities stay on track. For midsize and large companies, risk can affect multiple parts of the organization at once, from property and liability to cyber and regulatory challenges. At that level, managing risk becomes an ongoing discipline. At the Hartford, the focus is on helping businesses manage risk before it turns into something more disruptive.

53:07And when losses do happen, that work is paired with insurance coverage shaped by years of underwriting, risk engineering, and claims experience. Learn more at thehartford.com slash risk mitigation. Policies provided by Hartford Fire Insurance Company and its property and casualty affiliates, Hartford, Connecticut. It's time to plan ahead and make sure your brand is showing up in ways that can have an impact. Four Imprints promotional products are designed to work as hard as you do and make a lasting impression. From quality apparel, including exclusive brands, to drinkware, tech, and totes, they've got thousands of options to fit your brand and budget.

53:42Plus, you get free samples, expert help, and their 360-degree guarantee. So you can be 4imprint certain everything shows up just right, right on time. Explore more at 4imprint.com. 4imprint. 4certain. As industries evolve faster than ever, companies need an environment that accelerates strategic growth. And Michigan delivers on that promise. From emerging startups to global enterprises, Michigan offers what executives value most. A resilient, innovative ecosystem. Diverse communities that attract top talent and a quality of life that supports work-life balance. With our unified Team Michigan approach, businesses scale faster and compete at the highest level.

54:22Michigan, pure opportunity. Seize your opportunity at michiganbusiness.org.