WEBVTT - IBM Pledges $150B US Investment, Apple Shakes Up AI Unit 0:00:02.520 --> 0:00:09.800 Bloomberg Audio Studios, podcasts, radio news from the heart of 0:00:09.840 --> 0:00:15.600 where innovation, money and power collide in Silicon Valley and beyond. 0:00:16.000 --> 0:00:43.000 This is Bloomberg Technology with Caroline Hyde and Ed Ludlow. 0:00:35.320 --> 0:00:37.879 Live from New York. This is Blueberg Technology Coming up. 0:00:38.240 --> 0:00:40.800 Four of the Magnificent Seven are set to report earnings 0:00:40.840 --> 0:00:44.000 this week. The giants make up nearly twenty percent waiting 0:00:44.240 --> 0:00:45.400 in the S and P five hundred. 0:00:45.720 --> 0:00:46.440 So what's priced in? 0:00:46.960 --> 0:00:47.280 Plus? 0:00:47.320 --> 0:00:49.680 IBM plans to invest one hundred and fifty billion dollars 0:00:49.760 --> 0:00:52.159 in the US over the next five years with Quantumer 0:00:52.240 --> 0:00:53.840 mainframes made in America. 0:00:54.400 --> 0:00:56.360 And we will be joined by the CEO of Palo 0:00:56.440 --> 0:00:57.440 Alto Networks. 0:00:57.680 --> 0:01:01.440 He announces a new deal to leverage the AI security opportunity. 0:01:01.480 --> 0:01:04.600 But first we have Meta Microsoft come Wednesday. We then 0:01:04.680 --> 0:01:05.760 have Apple Amazon. 0:01:06.120 --> 0:01:08.120 What is priced in? Where have it? Investors in? 0:01:08.200 --> 0:01:11.560 Analysts already started to think about the pull forward for 0:01:11.640 --> 0:01:14.399 their fiscal quarter they report because of Taris and what 0:01:14.560 --> 0:01:18.160 of the macro economic environment going into the full year 0:01:18.200 --> 0:01:21.119 earnings expectations. Ran Vastella can can dig in because Ryan, 0:01:21.360 --> 0:01:24.440 what's so interesting about these numbers the fact that many 0:01:24.480 --> 0:01:27.280 analysts just in the lack of clarity, haven't been really 0:01:27.319 --> 0:01:29.440 able to downgrade their expectations. 0:01:30.000 --> 0:01:30.160 Yeah. 0:01:30.200 --> 0:01:33.120 Absolutely, we're starting to see this in some names, especially 0:01:33.120 --> 0:01:35.640 for twenty twenty six, But for the moment, people are 0:01:35.680 --> 0:01:38.480 sort of holding firm with their expectations. We really do 0:01:38.600 --> 0:01:41.639 not know what to expect out of results this quarter. 0:01:41.959 --> 0:01:44.080 We really do not know what any company is going 0:01:44.160 --> 0:01:47.400 to say about really anything about the economic environment, about 0:01:47.400 --> 0:01:49.200 their outlooks for the rest of the year. That makes 0:01:49.240 --> 0:01:52.440 it extremely difficult for analysts to know how to sort of, 0:01:52.480 --> 0:01:55.400 you know, peg their expectations their estimates, and in turn 0:01:55.720 --> 0:01:58.360 that means that valuations right now are sort of up 0:01:58.360 --> 0:01:58.880 in the air. 0:01:58.920 --> 0:02:00.960 We really do not know what fair value. 0:02:00.640 --> 0:02:03.280 Looks like because it is hard to evaluate these companies 0:02:03.480 --> 0:02:05.480 on a price to earnings basis if there is so 0:02:05.600 --> 0:02:07.440 much uncertainty about the earnings. 0:02:07.760 --> 0:02:11.239 I mean, look, we're looking currently at analysts being very 0:02:11.280 --> 0:02:14.280 slow to cut profit estimates on the twelve month earnings 0:02:14.280 --> 0:02:17.240 projections for a Meta for example. But Ryan Meta is 0:02:17.240 --> 0:02:21.079 going to feel the effects of Chinese companies like Timu 0:02:21.680 --> 0:02:25.080 like Shean no longer advertising with them. Potentially we're going 0:02:25.120 --> 0:02:27.800 to see a downdraft there. And also they've got hardware 0:02:27.840 --> 0:02:29.840 that they have to import from China. Are we going 0:02:29.840 --> 0:02:31.600 to get any signal of what that does to the 0:02:31.639 --> 0:02:33.080 bottom line? 0:02:33.440 --> 0:02:35.320 Yeah, I mean, I think these kinds of questions are 0:02:35.320 --> 0:02:38.120 really what is paramount for investors right now. Obviously, these 0:02:38.120 --> 0:02:42.160 companies have their own sort of individual impacts related to 0:02:42.200 --> 0:02:44.280 tariffs and so forth. A company like Meta, which is 0:02:44.320 --> 0:02:46.920 so heavily reliant on online advertising, it's not going to 0:02:46.960 --> 0:02:48.880 be as direct for them as it will be for 0:02:48.919 --> 0:02:51.960 a company like Apple, which is hardware based. So I mean, 0:02:51.960 --> 0:02:54.640 obviously we're going to see different kinds of impacts, different 0:02:54.680 --> 0:02:56.960 levels of impact. But I think overall, there's still so 0:02:57.080 --> 0:03:00.560 much uncertainty about what the economy is going to look like, 0:03:01.080 --> 0:03:02.920 not even a year from now, but even a few 0:03:03.000 --> 0:03:04.840 quarters from now. We are a couple of months from now, 0:03:04.880 --> 0:03:07.480 So again, any kind of clarity investors will get will 0:03:07.520 --> 0:03:09.560 be certainly welcome, But it's hard to know how any 0:03:09.600 --> 0:03:13.280 company can even be too confident about anything right now. 0:03:13.520 --> 0:03:16.440 Alphabet beat last week, and all of these stocks swored 0:03:17.040 --> 0:03:20.880 the previous week as well. Raan Vas Fleastelica pushes us forward. 0:03:20.880 --> 0:03:23.840 We thank you now President Trump's tarrorf announcements. They've roiled 0:03:23.960 --> 0:03:26.160 Washington to Wall Street for nearly a month now. And 0:03:26.320 --> 0:03:29.040 Chinese retail apps you just mentioned them, tea machine, they've 0:03:29.040 --> 0:03:32.200 already started actually passing on the new import taxes onto 0:03:32.280 --> 0:03:35.040 US consumers, onto you and me by hiking up prices. Now, 0:03:35.120 --> 0:03:38.360 China bus far has denied having any trade talks for 0:03:38.400 --> 0:03:41.640 the United States, though President Trump is still hopeful China 0:03:41.640 --> 0:03:42.840 will make a deal eventually. 0:03:44.040 --> 0:03:46.200 They want to make it deal obviously right now they're 0:03:46.200 --> 0:03:47.000 not doing business. 0:03:47.040 --> 0:03:50.080 Well, you know, people talk about going cold Turkey with 0:03:50.200 --> 0:03:50.560 China and. 0:03:50.560 --> 0:03:52.520 Then they just forget about it. 0:03:53.280 --> 0:03:56.200 And we were losing hundreds of billions of dollars a 0:03:56.280 --> 0:03:57.120 year with China. 0:03:57.960 --> 0:04:00.480 And now they're not doing any business with us, you know, 0:04:00.640 --> 0:04:03.240 because not because of them, because of me. 0:04:04.040 --> 0:04:04.720 Because at one. 0:04:04.640 --> 0:04:06.800 Hundred and forty five percent, you can't do business. 0:04:07.440 --> 0:04:10.160 But something's gonna happen. 0:04:10.320 --> 0:04:11.440 That's gonna be possible. 0:04:12.560 --> 0:04:15.440 It begs Mike Shepherd standing by, and look, we all 0:04:15.440 --> 0:04:17.839 still doing business with China. Many a consumer is still 0:04:17.880 --> 0:04:19.839 buying from Timu and she In, but they're paying the 0:04:19.839 --> 0:04:20.400 price for it. 0:04:21.720 --> 0:04:24.240 Well, you're right, kra and we are seeing now what 0:04:24.400 --> 0:04:27.440 economists have been warning about for some time in relation 0:04:27.640 --> 0:04:31.120 to the president's tariff policy, and that is that consumers 0:04:31.160 --> 0:04:35.000 would bear the increased costs and economic pain of those 0:04:35.080 --> 0:04:37.680 import levies. We are seeing that the prices are being 0:04:37.720 --> 0:04:41.200 passed along now by Timu and by she In. At 0:04:41.240 --> 0:04:44.240 she In, for instance, some goods, the prices have gone 0:04:44.320 --> 0:04:46.960 up by as much as three hundred percent. And keep 0:04:47.000 --> 0:04:49.320 in mind, Carra, it's not just here in the US 0:04:49.440 --> 0:04:53.080 where the pain is being felt by consumers. The retaliatory 0:04:53.120 --> 0:04:57.279 programs by some other trading partners, including Candidate, including China, 0:04:57.520 --> 0:05:01.200 including the EU have increased costs and goods from the 0:05:01.320 --> 0:05:03.920 US into those countries. And we saw Tesla over the 0:05:03.960 --> 0:05:08.320 weekend urging consumers to try to tap the pre tariff 0:05:08.400 --> 0:05:12.159 inventory before prices went up and before it became too 0:05:12.240 --> 0:05:15.560 late to avoid those increases. So there is a ripple 0:05:15.560 --> 0:05:18.320 effect and corollary effect, not just here in the US, 0:05:18.520 --> 0:05:19.800 but elsewhere around the world. 0:05:19.920 --> 0:05:22.039 Cara, extraordinary reporting done. 0:05:22.120 --> 0:05:25.640 They just highlight you mentioned Tesla having to pass on costs. 0:05:25.640 --> 0:05:26.839 We've also got here. 0:05:26.760 --> 0:05:29.760 Import tariffs wearing on Tiamo's shopping cart. They look at 0:05:29.760 --> 0:05:32.200 some of the most popular things that people buy on Team. 0:05:32.279 --> 0:05:34.240 Apparently it's a fourteen and one power strip with a 0:05:34.279 --> 0:05:35.920 surge protection and one touch control. 0:05:36.040 --> 0:05:37.560 Who knew that's the most popular item. 0:05:37.760 --> 0:05:40.599 But basically it's doubling in price because of tariffs MIC 0:05:41.000 --> 0:05:44.120 And when we think of an Amazon, for example, Bloomberg 0:05:44.120 --> 0:05:46.880 Intelligence thinks Amazon has about seventy percent of its products 0:05:47.800 --> 0:05:50.640 were either offered by Amazon or third party sellers coming 0:05:50.640 --> 0:05:51.159 from China. 0:05:51.200 --> 0:05:53.080 This is going to have a long term tail. 0:05:54.440 --> 0:05:56.120 This is going to have a long tail, and it's 0:05:56.160 --> 0:05:58.720 not really just going to be sticker shot carro It's 0:05:58.760 --> 0:06:01.880 going to be supply live shock potentially. We've heard from 0:06:01.920 --> 0:06:05.880 a number of economists and experts out there, including the 0:06:05.920 --> 0:06:08.440 chief economists that Apollo who one that we could be 0:06:08.480 --> 0:06:11.719 seeing a COVID like shock. Remember this is a crucial 0:06:11.800 --> 0:06:14.200 time of year. You and I were just trying to 0:06:14.200 --> 0:06:16.039 get through the end of the school year, maybe for 0:06:16.120 --> 0:06:19.760 our children, but we are all looking ahead. If you're 0:06:19.760 --> 0:06:23.760 a retailer, you're thinking ahead to Christmas to back to school, 0:06:24.040 --> 0:06:29.240 because now is when that inventory is actually planned, it's purchased, 0:06:29.279 --> 0:06:32.920 and it's already shipped. It's usually in the water by 0:06:33.160 --> 0:06:35.880 April to get here in time for back to school 0:06:35.920 --> 0:06:37.200 and for the holidays. 0:06:36.839 --> 0:06:38.440 So there won't be a ripple effect. 0:06:38.480 --> 0:06:41.600 And we've already seen a drop in cargo shipments from 0:06:41.640 --> 0:06:44.919 the world's second largest economy to the US in the 0:06:44.960 --> 0:06:45.960 wake of these tariffs. 0:06:45.960 --> 0:06:46.840 So there is going to. 0:06:46.800 --> 0:06:50.080 Be a lot of pain felt along the supply chain. 0:06:50.279 --> 0:06:52.560 It's not just going to be consumers at the end, 0:06:52.600 --> 0:06:55.480 like you and me looking at prices in our Temu 0:06:55.520 --> 0:07:00.360 shopping car, or even on the grocery store or other retailers. 0:07:00.400 --> 0:07:02.160 It's going to be in a lot of different places. 0:07:02.480 --> 0:07:05.359 And yet many hold out hopes for a deal. But 0:07:05.640 --> 0:07:08.480 China pushing that once again that no negotiations are happening. 0:07:09.720 --> 0:07:12.880 Well, that's right, and Scott Besson is indicating that, hey, 0:07:13.000 --> 0:07:15.559 you know, the ball is in China's court. We really 0:07:15.600 --> 0:07:19.280 want talks to get underway. And the President indicated even 0:07:19.320 --> 0:07:21.920 at the end of last week during a time when 0:07:22.000 --> 0:07:25.080 Chinese officials were here in Washington for the World Bank 0:07:25.120 --> 0:07:29.680 and IMF meetings, where maybe there was some contact between 0:07:29.720 --> 0:07:32.720 the two sides, but it was informal and if it 0:07:32.800 --> 0:07:35.720 even took place, And for the Chinese, they really want 0:07:35.760 --> 0:07:39.000 something much higher level, they want something launched formally. They 0:07:39.040 --> 0:07:42.120 would like to be speaking regularly with Scott Besson and 0:07:42.240 --> 0:07:45.400 other officials of his port to be able to get 0:07:45.440 --> 0:07:49.520 all of those key points for them, not just tariffs, remember, 0:07:49.720 --> 0:07:53.280 but all the US export restrictions on advanced technology that 0:07:53.320 --> 0:07:55.080 we've been talking about so much. 0:07:54.920 --> 0:07:57.720 Particularly in the semiconductor space. Mike Sheppard, thank you so 0:07:57.800 --> 0:08:00.480 much for joining me while coming up. IBM to invest 0:08:00.480 --> 0:08:02.720 one hundred and fifty billion dollars in the US over 0:08:02.720 --> 0:08:05.720 the next five years. Why this is blue met technology. 0:08:23.040 --> 0:08:26.000 IBM just raising its earlier gains as the market turns 0:08:26.000 --> 0:08:28.480 a little bit lower. It just announced plans to invest 0:08:28.520 --> 0:08:30.680 one hundred and fifty billion dollars in the United States 0:08:30.720 --> 0:08:33.360 over the next five years. The latest company, of course, 0:08:33.480 --> 0:08:36.120 that follows the likes of Apple or Eli Lilly pledging 0:08:36.160 --> 0:08:39.240 to increase spending in the country following President Trump's election 0:08:39.520 --> 0:08:42.560 and his teriff threats. Let's bring in brody Ford and 0:08:42.600 --> 0:08:45.360 starting on IBM. It seems to be about made in 0:08:45.400 --> 0:08:49.120 America the mainframe quantum compute. How much of an acceleration 0:08:49.240 --> 0:08:51.880 in spending home in the US space is this? 0:08:53.240 --> 0:08:55.360 That is the big question. It's the one that is 0:08:55.400 --> 0:08:58.280 so hard to answer looking at a lot of these announcements, right, 0:08:58.360 --> 0:09:00.880 IBM says we're going to spend their billion in R 0:09:00.960 --> 0:09:04.800 and D in the US. We look that they've reported 0:09:04.800 --> 0:09:07.199 about thirty three billion in R and D over the 0:09:07.280 --> 0:09:09.720 last five years. I believe most of that's in the US. 0:09:09.760 --> 0:09:11.880 So it's hard to say is this an acceleration. Is 0:09:11.920 --> 0:09:14.800 this them you know, putting a pretty bow and spending 0:09:14.800 --> 0:09:17.240 they were going to do anyway? This is the big 0:09:17.320 --> 0:09:20.240 question when we see these announcements from companies like Apple 0:09:20.400 --> 0:09:25.119 or Nvidia. What's clear is that as tariffs hit, companies 0:09:25.160 --> 0:09:27.959 feel a lot of pressure to say, hey, actually we 0:09:28.040 --> 0:09:31.280 have capabilities in the US and upstate New York. We 0:09:31.280 --> 0:09:33.920 can build your main frames too, And so I think 0:09:33.920 --> 0:09:35.680 that's what we're seeing from IBM today. 0:09:36.000 --> 0:09:39.320 Alvin Krishna, the CEO, saying IBM remains the epicenter of 0:09:39.360 --> 0:09:42.439 the world's most advanced computing and AI capabilities. 0:09:42.480 --> 0:09:44.480 Now I want to get to AI capabilities. 0:09:43.960 --> 0:09:46.360 Because I'm an interesting story you've been reporting on for 0:09:46.440 --> 0:09:49.520 over in the UK. Nscale, which seems to be a 0:09:49.600 --> 0:09:53.800 startup that is once again a Bitcoin minor turn compute offerer, 0:09:54.120 --> 0:09:56.280 trying to raise money and they're talking about a deal 0:09:56.320 --> 0:09:57.040 with bike Dugs. 0:09:58.280 --> 0:09:58.480 Yeah. 0:09:58.640 --> 0:10:01.079 Nscale is kind of like a or weave where they 0:10:01.120 --> 0:10:04.520 are focused on building these AI specific data centers and 0:10:04.559 --> 0:10:06.800 so they've been fundraised. They looking for a billion or 0:10:06.840 --> 0:10:10.160 two in debt, telling everybody, hey, we are how we 0:10:10.200 --> 0:10:13.120 have this deal with Byteedance for about assign the paper. 0:10:13.520 --> 0:10:15.840 It's gonna be worth two billion dollars. We're gonna give 0:10:15.840 --> 0:10:19.199 them all these great GPUs and Bye Dance told us 0:10:19.440 --> 0:10:22.480 that's overstating things. We don't have that much business with them, 0:10:22.800 --> 0:10:25.520 and so these situations are hard, right because GPUs are 0:10:25.520 --> 0:10:29.719 a pretty geopolitically sensitive thing. Maybe byte Dance doesn't want 0:10:29.720 --> 0:10:32.440 it out there. Maybe in scale is overstating things. I mean, 0:10:32.480 --> 0:10:36.360 it's an interesting instance that shows us how much the 0:10:36.800 --> 0:10:41.240 financing around AI data centers has become very interesting in 0:10:41.280 --> 0:10:41.960 the last year. 0:10:42.440 --> 0:10:45.000 And of course they're rather home to in video chips. 0:10:45.040 --> 0:10:47.400 Berdie Ford breaks that down. We thank you, and let's 0:10:47.400 --> 0:10:49.600 talk more about in video chips and competition to them, 0:10:49.640 --> 0:10:53.160 because over in China Huawei apparently the company is set 0:10:53.200 --> 0:10:55.800 to test a new AI processor then it hopes can 0:10:55.920 --> 0:10:57.960 rival those of Nvidia. Now it's all according to reports 0:10:58.000 --> 0:11:00.160 on the Wall Street Journal in video trading, the as 0:11:00.200 --> 0:11:02.520 you can see on the day now accelerating losses in 0:11:02.679 --> 0:11:05.480 King talks about this about what's happening over at Intel. 0:11:05.559 --> 0:11:09.079 But first Yan, what is the long term picture of 0:11:09.160 --> 0:11:12.200 Huawei trying to help China become more self sufficient here? 0:11:12.960 --> 0:11:14.680 Yeah, I think what we have to do is because 0:11:14.720 --> 0:11:16.880 it's very difficult to get a clear picture of what 0:11:17.000 --> 0:11:19.920 is going on there and the capabilities of those chips, 0:11:20.679 --> 0:11:22.240 what it is important to do is to take it 0:11:22.320 --> 0:11:24.760 back to what Invidia has said, which is, look, if 0:11:24.760 --> 0:11:26.679 we're not allowed to do business there, if we are 0:11:26.720 --> 0:11:30.440 restricted from exporting chips to China, they will find a 0:11:30.440 --> 0:11:33.040 way to do their own over time. It's just a 0:11:33.080 --> 0:11:35.960 matter of time. And I think this report is another 0:11:36.000 --> 0:11:38.920 example of efforts being made out there to do that. 0:11:39.200 --> 0:11:41.800 Are they competitive with what Invidia can do? Probably not 0:11:41.880 --> 0:11:44.800 at this point. Are they going to be wonderful? Are 0:11:44.840 --> 0:11:45.800 they going to be good enough? 0:11:45.880 --> 0:11:46.040 Though? 0:11:46.120 --> 0:11:49.080 Is really the key question, And the answer is we'll see. 0:11:49.480 --> 0:11:52.400 But they're probably another step on a path towards that. 0:11:52.720 --> 0:11:52.880 Well. 0:11:52.920 --> 0:11:55.920 In there's a more local competitor who hasn't been much 0:11:55.960 --> 0:11:58.920 of a competitor to in video of late, and that's Intel. 0:11:59.040 --> 0:12:02.719 Lit Bhutan written a great Tech Daily report really on 0:12:02.920 --> 0:12:05.880 how is trying to nuance culture over an Intel. 0:12:06.040 --> 0:12:06.920 How has it gone down? 0:12:08.440 --> 0:12:11.600 Yeah, I mean he really didn't give any of the 0:12:11.640 --> 0:12:14.439 concrete answers that Wall Street wanted in terms of when 0:12:14.440 --> 0:12:16.679 are the new products going to arrive, when you're going 0:12:16.679 --> 0:12:19.600 to have outside customers for your factories. What he did 0:12:19.640 --> 0:12:22.439 focus on, and what he did give a very specifics on, 0:12:22.679 --> 0:12:25.440 was like, Intel's culture is not in a good position, 0:12:25.760 --> 0:12:27.680 and these are some of the shocking things that we're 0:12:27.679 --> 0:12:31.640 seeing going on. Frankly, it was pretty brutal. That clearly 0:12:31.679 --> 0:12:33.760 is something that he believes needs to be fixed. But 0:12:33.880 --> 0:12:37.280 as we know, fixing a company culture and getting everybody 0:12:37.280 --> 0:12:40.720 moving in the right direction is important, takes time, and 0:12:40.800 --> 0:12:43.400 what is arguably more important is that you're getting them 0:12:43.600 --> 0:12:47.640 moving in the right direction towards products, towards customer relationships 0:12:47.679 --> 0:12:49.280 that will turn this company around. 0:12:49.880 --> 0:12:53.320 We have seen the response to the earnings in King 0:12:53.559 --> 0:12:55.280 Tech in Depth, the newsletter out today. 0:12:55.280 --> 0:12:55.839 Go read it. 0:12:55.880 --> 0:13:05.800 Thanks so much. Now, as shakeup is underway for Apple's 0:13:05.840 --> 0:13:08.600 AI and machine learning divisions, the company has begun to 0:13:08.600 --> 0:13:11.200 break up would have been a pretty unified approach for more. 0:13:11.240 --> 0:13:13.160 Blue Meg's Mark German has more. 0:13:13.200 --> 0:13:16.040 You've been writing over the weekend about one particular executive 0:13:16.080 --> 0:13:18.240 they lured away from Google in twenty eighteen, and now 0:13:18.280 --> 0:13:20.960 he's getting some things pulled away from him. 0:13:21.640 --> 0:13:25.280 That's right. So here's the story with Apple. AI serie 0:13:25.360 --> 0:13:29.000 launched in twenty eleven, and by twenty fifteen, twenty sixteen, 0:13:29.280 --> 0:13:31.760 a little bit after that, they really fell behind Google. 0:13:31.800 --> 0:13:36.880 Assistant came online. You had a lot of competitors, including 0:13:36.880 --> 0:13:40.359 Alexa from Amazon, and it felt like Apple had squandered 0:13:40.400 --> 0:13:42.720 a three four year lead. And at the time, Apple 0:13:42.760 --> 0:13:45.520 felt that part of that is because the AI work 0:13:45.720 --> 0:13:49.560 at Apple was scattered across the company, right They worked 0:13:49.679 --> 0:13:54.440 in silos across different divisions, hardware, software services. So they 0:13:54.480 --> 0:13:57.000 hired John gen Andrea, who was running AI in search 0:13:57.040 --> 0:13:59.280 at Google at the time in twenty eighteen. They believe 0:13:59.320 --> 0:14:02.520 putting everything in his group with skyrocket their approach to 0:14:02.559 --> 0:14:05.200 AI and make them a true player in artificial intelligence 0:14:05.200 --> 0:14:05.959 and machine learning. 0:14:06.520 --> 0:14:07.840 So far, it hasn't worked out. 0:14:07.960 --> 0:14:11.559 We saw how far they've fallen behind compared to Chad Shipt, Google. 0:14:11.280 --> 0:14:12.360 Gemini Copilot. 0:14:12.720 --> 0:14:14.920 So now what they're trying is breaking it up, going 0:14:14.920 --> 0:14:17.280 back to their roots, going back to that Apple functional 0:14:17.400 --> 0:14:21.000 organization and taking pieces of the AI group and re 0:14:21.160 --> 0:14:23.760 scattering them around the company, and that started a few 0:14:23.800 --> 0:14:26.200 weeks ago with Siri. They move Siri from their AI 0:14:26.280 --> 0:14:29.760 division to their software engineering division, which makes sense to 0:14:29.800 --> 0:14:33.000 some extent because Siri is well software. And now what 0:14:33.000 --> 0:14:35.800 they're doing is they're taking the robotics division, which was 0:14:35.880 --> 0:14:39.280 also in the AI organization, and moving that to hardware engineering. 0:14:39.680 --> 0:14:42.320 Robotics that is a hardware play at heart, and so 0:14:42.400 --> 0:14:44.720 it makes sense to put that in the hardware engineering division. 0:14:45.000 --> 0:14:46.920 So we'll see how that works out. But to me 0:14:47.440 --> 0:14:49.440 and to many people at Apple, it feels like they're 0:14:49.440 --> 0:14:52.440 preparing for a world without a true AI chief, a 0:14:52.480 --> 0:14:55.960 world where apples back to being completely functionally organized. 0:14:55.840 --> 0:14:58.360 And mark a world where they're about to report earnings 0:14:58.400 --> 0:15:01.360 this week, they're trying to tackle China supply chain, move 0:15:01.400 --> 0:15:03.800 it more to India. Just brace us for what we 0:15:03.840 --> 0:15:04.880 expect on Thursday. 0:15:06.320 --> 0:15:08.760 I think that Apple this is going to be one 0:15:08.800 --> 0:15:12.840 of the more interesting Apple earnings results presentations in a while, 0:15:13.200 --> 0:15:15.360 not because of the numbers that they're going to share, 0:15:15.760 --> 0:15:17.440 not because of the data they're going to be given 0:15:17.440 --> 0:15:20.240 the prepared remarks, but because they have a lot more 0:15:20.320 --> 0:15:22.920 questions they're going to have to answer than usual. Things 0:15:22.920 --> 0:15:24.960 when it comes to AI, of course is going to 0:15:24.960 --> 0:15:27.160 be big on the minds of analysts on the call, 0:15:27.200 --> 0:15:31.360 but also tariffs, also dealing with the Trump administration, dealing 0:15:31.360 --> 0:15:34.960 with innovation questions. So that's going to be really important 0:15:34.960 --> 0:15:36.840 earnings TOOLL to listen to, not because of what they're 0:15:36.840 --> 0:15:38.120 going to come out and say, but because of what 0:15:38.160 --> 0:15:40.520 they're going to be asked and how they're going to respond. 0:15:40.560 --> 0:15:42.680 So I'm actually very much looking forward to that, and 0:15:42.840 --> 0:15:44.600 everyone should stay tuned for our coverage there. 0:15:45.120 --> 0:15:47.720 Well, we're excited for your reporting on it, Mark German. 0:15:47.960 --> 0:15:50.480 We appreciate all things Apple. And now. 0:15:50.600 --> 0:15:52.120 Meanwhile, in other AI. 0:15:51.920 --> 0:15:56.400 News, enterprise AI platform Writer is launching a new adaptive 0:15:56.480 --> 0:16:00.000 reasoning model that it says is cheaper faster than competing lllms. 0:16:00.320 --> 0:16:03.200 It eaches a one million token context window to create 0:16:03.240 --> 0:16:07.040 more comprehensive responses. Mayhabi write as CEO joins us for 0:16:07.120 --> 0:16:09.880 now and tell me about the new model and what 0:16:09.960 --> 0:16:10.640 makes it different. 0:16:11.680 --> 0:16:14.840 Oh hi Caroline, so good to see you. Well that's 0:16:14.840 --> 0:16:15.400 the headline. 0:16:15.480 --> 0:16:18.680 Palmyra x five is out and it is the first 0:16:18.960 --> 0:16:24.000 enterprise grade, enterprise ready model built for agentic AI. 0:16:24.200 --> 0:16:25.040 So you got it. 0:16:25.120 --> 0:16:28.880 This is an LM with performance on par with other 0:16:29.000 --> 0:16:33.320 frontier models, but it's radically three x faster, and it's 0:16:33.360 --> 0:16:36.480 four times cheaper, and as of today it's available on 0:16:36.600 --> 0:16:37.920 AWS Bedrock. 0:16:38.360 --> 0:16:40.520 Okay, So tell us about how you get that sort 0:16:40.560 --> 0:16:43.800 of cost efficiency. You're offering what sixty cents per one 0:16:43.800 --> 0:16:47.960 million input tokens, you're doing six dollars per million output tokens. 0:16:48.080 --> 0:16:49.400 How do you get to that price point? 0:16:49.440 --> 0:16:49.640 Way? 0:16:50.560 --> 0:16:54.119 So it's really important for folks to understand what's required 0:16:54.160 --> 0:16:57.960 to get enterprise grade agentic AI. When we launched our 0:16:58.120 --> 0:17:01.840 AI HQ a few weeks ago, we had one hundred 0:17:02.000 --> 0:17:05.800 plus pre built agents built for the enterprise, and folks 0:17:05.880 --> 0:17:09.040 were asking me, how are you getting this kind of 0:17:09.240 --> 0:17:15.760 efficiency and accuracy with really complex workflows Agents at a Uber, 0:17:15.840 --> 0:17:19.440 at a Franklin Templeton, things that were capable of taking 0:17:20.040 --> 0:17:24.560 multiple systems, data from third party and internal research, and 0:17:24.680 --> 0:17:29.560 able to generate really complex work We are able to 0:17:29.600 --> 0:17:35.520 take that full million token prompt and process that context 0:17:35.520 --> 0:17:39.119 window in twenty two seconds, firing off multi turn function 0:17:39.240 --> 0:17:43.560 calls in three hundred milliseconds, all while getting this price efficiency. 0:17:44.280 --> 0:17:45.280 It is a. 0:17:45.160 --> 0:17:51.119 Combination of breakthrough both algorithmic and data we train this 0:17:51.240 --> 0:17:54.520 model on about a million dollars of GPUs, so also 0:17:54.600 --> 0:17:57.320 incredibly efficient on the training and we passed. 0:17:57.359 --> 0:17:58.800 How how is this occurring? 0:17:58.840 --> 0:17:59.000 Way? 0:17:59.119 --> 0:18:01.600 Is this the new deep seek trend that we're seeing 0:18:01.640 --> 0:18:03.280 that you just do more and more with this. It's 0:18:03.320 --> 0:18:05.879 more about turning to open source. I mean, how have 0:18:05.920 --> 0:18:07.040 you managed to be so efficient? 0:18:08.080 --> 0:18:08.280 Yeah? 0:18:08.320 --> 0:18:10.280 I mean we deep seeked before deep seek. 0:18:10.760 --> 0:18:13.679 So this is a one hundred percent synthetic data based 0:18:13.720 --> 0:18:20.840 approach which has the IP friendliness that the enterprise really requires. Algorithmically, 0:18:21.480 --> 0:18:26.119 we are doing a very interesting set of techniques around 0:18:26.400 --> 0:18:31.439 a hybrid attention mechanism, blending linear and other approaches, and 0:18:31.480 --> 0:18:34.800 we're doing a mixture of experts set of techniques, so 0:18:34.880 --> 0:18:40.399 we're only activating the most relevant subnetworks and resulting in 0:18:40.600 --> 0:18:42.760 much lower latency, much more efficiency. 0:18:43.160 --> 0:18:46.120 Mate, you talked about some of the very well known 0:18:46.160 --> 0:18:49.280 companies that you currently serve. But to get these sort 0:18:49.280 --> 0:18:52.399 of investors and the vanguards of this world to be 0:18:52.400 --> 0:18:55.480 coming on board, how are you making yourself known? I mean, 0:18:55.640 --> 0:18:57.800 are you having to offer this sort of price efficiency? 0:18:57.840 --> 0:18:59.240 Is that how you arrives above the noise? 0:18:59.240 --> 0:19:01.400 Because it just feels like there are so many offerings, 0:19:01.600 --> 0:19:03.200 enterprise suited offerings out. 0:19:03.040 --> 0:19:03.640 There right now. 0:19:04.320 --> 0:19:06.400 It's absolutely so noisy. 0:19:06.440 --> 0:19:09.720 But what we're seeing is folks who don't have results 0:19:09.760 --> 0:19:13.159 are understanding that they can't build everything themselves. So in 0:19:13.200 --> 0:19:17.760 addition to enterprise grade frontier lms, we've got an end 0:19:17.760 --> 0:19:22.679 to end platform for building, activating, supervising these agents. And 0:19:22.680 --> 0:19:26.359 that's how we stand out, Caroline. That entire package is 0:19:26.400 --> 0:19:31.199 what results in the incredible roy Forster just reported three 0:19:31.320 --> 0:19:35.080 hundred and thirty three percent ROY on implementing writer for 0:19:35.359 --> 0:19:38.640 building and using agents and it's one of the highest 0:19:38.640 --> 0:19:42.199 that they've ever seen. So you're right, right, it's incredibly 0:19:42.240 --> 0:19:46.240 noisy for these enterprises, but CEOs are seeing that their 0:19:46.280 --> 0:19:50.879 CEOs are demanding results. And I think this is AI's 0:19:51.000 --> 0:19:55.040 COVID moment, and COVID, you know, you had this explosion 0:19:55.200 --> 0:19:58.800