WEBVTT - Bloomberg Tech Live in San Francisco 0:00:01.520 --> 0:00:02.240 From our heart. 0:00:02.360 --> 0:00:06.800 We're Innovation, Money and Power Collie in Silicon Valley, NBN. 0:00:07.120 --> 0:00:11.640 This is Bloomberg Technology with Caroline Hyde and Ed Ludlow. 0:00:27.400 --> 0:00:29.920 Live from sunny San Francisco and Caroline Hyde. 0:00:29.720 --> 0:00:30.520 And our Ed Ludlow. 0:00:30.600 --> 0:00:34.519 This is a special edition of Bloomberg Technology. Coming up 0:00:34.640 --> 0:00:37.760 will bring you live coverage from our technology event as 0:00:37.760 --> 0:00:41.080 we speak with the CEOs and visionaries that are driving 0:00:41.200 --> 0:00:43.360 change in Silicon Valley and beyond. 0:00:43.520 --> 0:00:46.599 Now at this hour, we're speaking with the CEOs of Arm, 0:00:46.880 --> 0:00:50.240 of Hugging, Face, of Writer, AI and more as part 0:00:50.360 --> 0:00:51.680 of our live special. 0:00:51.720 --> 0:00:52.880 And a lot more to come. 0:00:52.880 --> 0:00:55.480 We'll push your head to our keynote speakers later today. 0:00:55.760 --> 0:00:59.880 That includes Adam Newman, Whitney Wolf Heard, Evan Spiegel and more. 0:01:00.080 --> 0:01:00.720 And while it. 0:01:00.640 --> 0:01:02.440 May not be a surprise to anyone too, did it's 0:01:02.440 --> 0:01:05.840 certainly not a surprised to us. Artificial intelligence is probably 0:01:05.880 --> 0:01:10.560 the overriding theme of the event. It's AI everything, even 0:01:10.600 --> 0:01:12.440 in some cases if you're not an AI company. 0:01:12.640 --> 0:01:14.119 Yeah, I mean I think it has to be. Even 0:01:14.160 --> 0:01:16.080 if you're not the AI company at heart, you're thinking 0:01:16.080 --> 0:01:17.560 about how you adapt to it, push it forward and 0:01:17.560 --> 0:01:19.760 bring out the productivity. But Ultimately, how do we live 0:01:19.840 --> 0:01:22.520 up to the hype? The valuations are so extraordinary in 0:01:22.560 --> 0:01:24.880 the private markets, they've been pretty heavy in the public 0:01:24.920 --> 0:01:27.160 markets as well. Are we really seeing the level of 0:01:27.160 --> 0:01:29.640 producted divity and growth and real use cases? 0:01:29.680 --> 0:01:33.440 Do mindication that these events are always very interesting, very engaging. 0:01:33.560 --> 0:01:35.960 Everyone's very positive, But I would say in the background, 0:01:36.319 --> 0:01:39.800 there's a talent war, frankly, and people are running out 0:01:39.800 --> 0:01:40.480 of cash. 0:01:40.280 --> 0:01:42.200 So we've got to ask those difficult questions too. 0:01:42.319 --> 0:01:45.839 And geopolitics, how are you navigating the issue of China 0:01:45.920 --> 0:01:48.520 as well? So so much to talk about in here 0:01:48.560 --> 0:01:52.920 and the now. We're talking about one key infrastructure playing 0:01:53.000 --> 0:01:55.760 When it comes to artificial intelligence, we're of course going 0:01:55.800 --> 0:01:58.560 to be talking about the chip design firm ARM, which 0:01:58.800 --> 0:02:01.480 has really bounced offlows in terms of its share price 0:02:01.520 --> 0:02:03.960 throughout the trading of today. We're currently over the last 0:02:03.960 --> 0:02:05.960 two days down by one point few percent. Coming out 0:02:05.960 --> 0:02:09.040 after the bell with its earnings, tapid forecast head was 0:02:09.080 --> 0:02:11.320 what the market seemed to be focusing in on, even 0:02:11.320 --> 0:02:13.840 though they absolutely smashed it in terms of their fiscal 0:02:13.880 --> 0:02:16.760 fourth quarter numbers and their fourth first quarter. To look ahead, 0:02:17.120 --> 0:02:19.720 let's stick in to some of that caution with Renee 0:02:19.760 --> 0:02:22.600 has is the arm CEO. Renee wonderful to have time 0:02:22.639 --> 0:02:24.720 with you. And look, there does seem to be a 0:02:24.760 --> 0:02:28.880 worry about your full year forecasts? Are you being cautious? 0:02:30.480 --> 0:02:33.160 Well, thanks both for having me Ed and Caroline. We 0:02:33.320 --> 0:02:36.120 just came off a record year in terms of revenue. 0:02:36.120 --> 0:02:38.200 We were up twenty percent a little bit over twenty 0:02:38.280 --> 0:02:43.120 percent our fiscal twenty three from twenty twenty two, and 0:02:43.160 --> 0:02:46.320 we're actually forecasting even higher growth this year, north of 0:02:46.400 --> 0:02:49.919 twenty percent. And we also signaled to the markets yesterday 0:02:49.960 --> 0:02:53.799 that in twenty five, twenty six, twenty seven, we see 0:02:53.800 --> 0:02:57.880 that growth continuing. So we have incredible visibility to our 0:02:57.919 --> 0:03:01.160 business and we're very, very confident of growth rate going forward. 0:03:04.440 --> 0:03:07.240 We're just seeing your shares actually ticking to positive territory 0:03:07.240 --> 0:03:11.240 rene up now six tenths of one percent. The underlying 0:03:11.320 --> 0:03:14.720 story is the build out in AI infrastructure. Right, we're 0:03:14.720 --> 0:03:20.680 talking about data center powered GP by GPUs. Your numbers 0:03:20.680 --> 0:03:24.320 were good. Tell me about the underlying demand then about 0:03:24.320 --> 0:03:26.920 the long term and the addressable market you think is 0:03:26.960 --> 0:03:28.320 either intact or is not. 0:03:30.040 --> 0:03:30.200 Well. 0:03:30.240 --> 0:03:34.000 I think this AI buildout, as you describe, or maybe 0:03:34.040 --> 0:03:38.360 said another way, just expanding capacity to run these foundation models, 0:03:38.400 --> 0:03:39.840 to do more and more training, to do more and 0:03:39.920 --> 0:03:40.480 more inference. 0:03:41.440 --> 0:03:42.680 We really are only. 0:03:42.520 --> 0:03:44.800 At the very beginning because when you start to think 0:03:44.840 --> 0:03:49.000 about the capabilities that this could unleash, whether it's around healthcare, 0:03:49.400 --> 0:03:54.920 farmer research, productivity, gains, call centers, we're still in the 0:03:55.000 --> 0:03:58.080 very very early days. That all starts with having to 0:03:58.120 --> 0:04:00.320 do this level of training and inprints in the cloud, 0:04:00.720 --> 0:04:03.920 but it ultimately will find itself in every single edge device, 0:04:04.080 --> 0:04:09.120 whether that's a PC, your smartphone, your car, and whether 0:04:09.160 --> 0:04:11.880 it's all those devices I've mentioned from the data center 0:04:11.920 --> 0:04:14.440 to the edge devices. They all run on ARM. So 0:04:14.680 --> 0:04:17.760 we have incredible visibility to where this is all going, 0:04:17.760 --> 0:04:19.680 which is why we're very confident in the growth rates. 0:04:20.279 --> 0:04:22.320 They're also one of the big problems you've got with 0:04:22.520 --> 0:04:25.960 all of these AI data centers is around energy and power. 0:04:26.400 --> 0:04:28.800 So power efficiency being so key, it's what ARM is 0:04:28.880 --> 0:04:32.600 really good at. Increasingly we're seeing the most complex applications 0:04:32.680 --> 0:04:35.680 moving to ARM and most sophisticated training ship on the 0:04:35.680 --> 0:04:39.719 planet that was just announced Grace Blackwell, Well that's based 0:04:39.720 --> 0:04:44.160 on ARM. 0:04:44.320 --> 0:04:46.400 Okay, so you're managing to really think that you're going 0:04:46.440 --> 0:04:50.120 to be the server play as well as the PC play, 0:04:50.120 --> 0:04:51.760 the cell phone play, and I want to focus in 0:04:51.800 --> 0:04:53.760 on the cell phone play, Renee, because that's been where 0:04:53.760 --> 0:04:56.480 your bread and butter has been in history. How are 0:04:56.480 --> 0:04:59.320 we looking from a smartphone perspective? Is the market looking 0:04:59.360 --> 0:05:01.960 strong to you? We've had many a mixed message coming 0:05:01.960 --> 0:05:03.320 from China to MOD for example. 0:05:05.200 --> 0:05:08.000 Overall, what we've seen the smartphone market briftly for ARM 0:05:08.080 --> 0:05:10.880 has been quite a good growth rate in terms of royalties. 0:05:11.080 --> 0:05:14.320 Our version nine which is now being used in many 0:05:14.360 --> 0:05:17.599 of the premium mobile phones, that drives a higher royalty 0:05:17.680 --> 0:05:21.039 rate for ARM. There's also more complex CPUs that go 0:05:21.120 --> 0:05:24.839 into that that's also better for ARM and going forward carrolling. 0:05:24.920 --> 0:05:26.680 One of the things that we're seeing, and it's not 0:05:26.760 --> 0:05:30.520 just in smartphones, is that as these AI models are 0:05:30.520 --> 0:05:34.600 moving so fast, the hardware can't keep up with the software. 0:05:35.000 --> 0:05:38.919 The software innovation is happening so quickly that by the 0:05:38.960 --> 0:05:41.680 time the hardware is ready to run those models, everyone 0:05:41.720 --> 0:05:44.680 wishes they had more performance, they had more efficiency. 0:05:45.160 --> 0:05:46.280 So what does that mean for ARM? 0:05:46.760 --> 0:05:50.239 It's driving growth in our licensing activity. People are looking 0:05:50.240 --> 0:05:53.440 to do more and more design ships faster and faster, 0:05:53.920 --> 0:05:56.160 and that's all good for us going forward. So I 0:05:56.200 --> 0:05:58.040 think going forward you're going to see more and more 0:05:58.040 --> 0:06:01.080 innovation happening, not only in the smartphones, across all these 0:06:01.120 --> 0:06:01.880 edge devices. 0:06:05.320 --> 0:06:07.520 What's interesting in AY is it's hard to keep up 0:06:07.560 --> 0:06:11.000 with the pace of geopolitical change as well. The latest 0:06:11.000 --> 0:06:13.479 news coming that Huawei, of course is not going to 0:06:13.480 --> 0:06:16.400 have access to poll Kong to Intel chips. You were, 0:06:16.400 --> 0:06:19.800 of course a UK based company, but are affected by 0:06:19.920 --> 0:06:23.720 US policies. Has this impacted your business? The limitations of 0:06:23.760 --> 0:06:28.120 Huawei's access to chip designed to chip technology to licenses. 0:06:29.800 --> 0:06:32.960 Yeah, So that issue they referred to specifically was when 0:06:33.040 --> 0:06:35.960 Huawei was placed on the entity list I think twenty nineteen, 0:06:36.040 --> 0:06:39.839 twenty twenty, companies had to apply for a license to 0:06:40.080 --> 0:06:42.960 exempt them to ship to Huawei. So a number of 0:06:43.040 --> 0:06:47.120 companies asked for those licenses, they got those licenses. Now 0:06:47.120 --> 0:06:49.919 those licenses are being revoked. We don't follow in that 0:06:50.040 --> 0:06:52.599 category in any way, shape or form. We didn't apply 0:06:52.720 --> 0:06:55.440 for any licenses at the time to share. We complied 0:06:55.480 --> 0:06:57.880 with the export controls as they were laid out. So 0:06:58.080 --> 0:07:00.920 there's really a non event for us in terms of 0:07:00.960 --> 0:07:02.719 what you're seeing with Qualcoman or Intel. 0:07:06.160 --> 0:07:09.119 We are speaking live to the ARMS CEO, Renee has 0:07:09.200 --> 0:07:12.520 We're on the ground here at Bloomberg Tech in San Francisco. 0:07:12.880 --> 0:07:15.880 Last week, Renee Christiano mom was on the show telling 0:07:15.920 --> 0:07:19.160 Caroline and I, this is the year of the AIPC. 0:07:20.000 --> 0:07:22.280 You were asked about that on your earning school last 0:07:22.360 --> 0:07:25.560 night and you gave a slightly different answer. And maybe 0:07:25.560 --> 0:07:28.800 it's not the year of the AIPC more the twelve 0:07:28.800 --> 0:07:31.480 to thirty six month window. And you don't want to 0:07:31.520 --> 0:07:34.080 see just one PC supplier, you said you'd like to 0:07:34.120 --> 0:07:38.680 see two or three. What's your beef with Qualcom? 0:07:38.800 --> 0:07:38.880 Now? 0:07:38.960 --> 0:07:43.040 When I look at the PC ecosystem, one large ecosystem 0:07:43.080 --> 0:07:45.000 has already moved to ARM in a very big way. 0:07:46.000 --> 0:07:49.320 Apple is now one based on ARM. All the Apple 0:07:49.360 --> 0:07:52.680 silicon is based on ARM. And you see amazingly good 0:07:52.760 --> 0:07:57.440 products relative to what they've delivered, fantastic battery life, performance, 0:07:57.520 --> 0:08:00.720 thin and light, no fans. When you think about the 0:08:00.760 --> 0:08:03.760 Windows market, it's a very different market. It's highly fragmented. 0:08:03.800 --> 0:08:06.800 You have lots of different players. The ecosystem matters, the 0:08:06.920 --> 0:08:11.880 channel matters, price points matter, high end gaming machines versus 0:08:12.840 --> 0:08:16.480 low end devices that are like cloud enabled. So what 0:08:16.520 --> 0:08:20.280 does all that mean. It generally has meant that breath 0:08:20.840 --> 0:08:24.640 vendor choice, multiple options to provide a full scope is 0:08:24.640 --> 0:08:28.000 what matters. And what I'm hearing is over the next 0:08:28.000 --> 0:08:30.080 couple of years, the Windows ecosystem is going to be 0:08:30.080 --> 0:08:33.000 able to afford that. And I think over the next 0:08:33.320 --> 0:08:37.280 two three years, I do believe Windows Unarmed will be real. 0:08:37.559 --> 0:08:41.000 I think you'll see multiple players, multiple price points, multiple units, 0:08:41.320 --> 0:08:43.280 and I think you'll see meaningful market share that we 0:08:43.320 --> 0:08:46.120 start to gain the kind of performance you see in 0:08:46.160 --> 0:08:48.080 the other ecosystem. I think we'll find its way into 0:08:48.120 --> 0:08:52.640 the Windows ecosystem. 0:08:52.920 --> 0:08:54.800 Rennie, I wanted to talk about geography really quick. 0:08:55.000 --> 0:08:57.360 We're here in San Francisco, right there's a lot about 0:08:57.679 --> 0:09:01.520 Americas are in d focus on AI related chips. 0:09:01.800 --> 0:09:05.240 Are you seeing this sort of equivalent activity in. 0:09:05.120 --> 0:09:08.480 Europe, for example, any of your customers outside of those markets. 0:09:09.640 --> 0:09:11.400 Yeah, well I'm in San Francisco today too, so I 0:09:11.400 --> 0:09:14.760 will see you a little bit later. But in general, 0:09:15.040 --> 0:09:19.240 I think the geopolitics are something that all tech CEOs 0:09:19.240 --> 0:09:23.400 are now having to figure out and work with AI models, 0:09:23.520 --> 0:09:28.200 foundation models, sovereign clouds, thinking about what level of training 0:09:28.240 --> 0:09:32.720 takes place in a country, versus outside the country where 0:09:32.720 --> 0:09:34.679 the weights sit, et cetera. That these are all the 0:09:34.800 --> 0:09:37.720 kind of things that politicians have never really had to 0:09:37.760 --> 0:09:40.080 think about in the past. So we're involved in a 0:09:40.080 --> 0:09:43.800 lot of those conversations, whether that's in the United States, 0:09:43.840 --> 0:09:46.480 whether that's in Europe, and really just trying to understand 0:09:46.480 --> 0:09:49.960 it because any lawmakers in all these jurisdictions are just 0:09:50.000 --> 0:09:52.440 trying to figure it all out. And as I mentioned before, 0:09:52.880 --> 0:09:56.040 as the software and models are moving so fast, it's 0:09:56.080 --> 0:09:58.720 difficult for everyone to keep up. But we are central 0:09:58.760 --> 0:09:59.880 to all those discussions. 0:10:03.559 --> 0:10:06.160 Renee, what's been keeping up is your valuation? 0:10:06.679 --> 0:10:06.959 Boy? 0:10:07.520 --> 0:10:10.320 I mean, do you think there's too much exuberance around 0:10:10.400 --> 0:10:13.800 AI valuations out there? Are you going to make the 0:10:13.880 --> 0:10:15.800 most of it? By well, we talked to one point 0:10:15.800 --> 0:10:17.080 of listing in the UK too. 0:10:19.000 --> 0:10:21.320 Yeah, you know, I don't think about the valuations as 0:10:21.400 --> 0:10:23.839 much as I just think about the AI opportunity, which 0:10:23.960 --> 0:10:28.600 I frankly believe is undercalled in terms of just what 0:10:28.679 --> 0:10:31.199 it's going to mean relative to society and what it 0:10:31.240 --> 0:10:34.240 can do for our planet. I think again we are 0:10:34.280 --> 0:10:37.679 in very very early days in terms of the capabilities 0:10:37.720 --> 0:10:41.800 of what this can unleash for our society incredibly excited 0:10:41.840 --> 0:10:43.760 to be part of it. But I don't think we're 0:10:43.800 --> 0:10:45.680 part of a hype cycle at all. I think there's 0:10:45.720 --> 0:10:48.679 a lot of innovation taking place, and you know, frankly, 0:10:48.920 --> 0:10:52.160 the innovation that's taking place, any inventions that we're seeing, 0:10:52.600 --> 0:10:55.400 it's just breathtaking. So no, I don't personally view it 0:10:55.400 --> 0:10:56.439 as a hype cycle at all. 0:11:00.080 --> 0:11:03.000 They has I'm CEO really grateful you actually be here 0:11:03.000 --> 0:11:06.080 with us later today on site at Bloomberg Tech. Your 0:11:06.120 --> 0:11:08.959 stock open pretty low, and I think it's just a 0:11:09.000 --> 0:11:11.000 little bit higher now during the conversation we've had. 0:11:11.040 --> 0:11:11.760 Thank you so much. 0:11:11.800 --> 0:11:13.719 All Right, coming up on the program, we're going to 0:11:13.760 --> 0:11:17.120 be joined by Clem DeLong, CEO of Hugging Face. That's 0:11:17.120 --> 0:11:19.520 coming up next. Stay tuned, we'll be right back. Is 0:11:19.600 --> 0:11:35.360 Bloomberg Technology. Welcome back to this special edition of Bloomberg 0:11:35.400 --> 0:11:38.559 Technology live in San Francisco at the Bloomberg Tech event 0:11:38.679 --> 0:11:40.840 and artificial intelligence Surprise. 0:11:40.920 --> 0:11:43.080 Surprise is sort of the overarching theme. 0:11:43.400 --> 0:11:45.319 We've got a pretty good guest to talk about that 0:11:45.360 --> 0:11:48.640 with and discuss all things large language model with Clemed Along, 0:11:48.960 --> 0:11:52.679 CEO of Hugging Face. You made this prediction which we're 0:11:52.720 --> 0:11:55.880 going to hold you to account on that by the 0:11:55.960 --> 0:11:58.160 end of this calendar year, and I appreciate we're not 0:11:58.200 --> 0:12:03.880 even halfway. Source models would be equivalent to the best 0:12:04.000 --> 0:12:07.560 closed source models. Give us a status check of that 0:12:07.679 --> 0:12:08.520 prediction place. 0:12:08.880 --> 0:12:10.000 I think it's already happened. 0:12:10.480 --> 0:12:13.560 Open source now is better than closed source for most 0:12:13.840 --> 0:12:19.960 use cases. We've specialized customized models on the companies like 0:12:20.040 --> 0:12:24.160 data sets. I have my meta reband glasses here that 0:12:24.240 --> 0:12:25.720 are powered by Lammatri. 0:12:26.320 --> 0:12:26.480 Right. 0:12:26.559 --> 0:12:30.240 We've seen so many now use cases being powered by 0:12:30.280 --> 0:12:34.079 open source models, and most of the big tech companies 0:12:34.240 --> 0:12:37.439 are now publishing open models. Just last month, we've seen 0:12:38.240 --> 0:12:42.120 Apple releasing open models on hogging Face. We've seen Nvidia, 0:12:42.280 --> 0:12:46.600 we've seen Snowflakes, We've seen Data Bricks, we've seen Microsoft. 0:12:46.760 --> 0:12:48.640 All of them now are publishing open models. 0:12:48.679 --> 0:12:51.040 There maybe some people in attendance who don't agree with him, 0:12:51.040 --> 0:12:52.240 which is why I asked the question. 0:12:52.400 --> 0:12:55.080 Well, also Microsoft published one and then quickly with DURI 0:12:55.480 --> 0:12:58.719 some of the reporting because they hadn't stress tested the 0:12:58.840 --> 0:13:01.880 large language model enough. I hadn't whittled out some of 0:13:01.880 --> 0:13:05.800 the toxicity checks. In particular, how are you feeling about 0:13:06.000 --> 0:13:09.160 the way in which large language models are growing and 0:13:09.280 --> 0:13:11.719 the way in which governance is developing around it. 0:13:12.120 --> 0:13:14.240 Well, we're starting to see that the most important question 0:13:14.440 --> 0:13:17.880 is concentration of power. Right for such an important technology, 0:13:17.880 --> 0:13:20.920 you don't want a world where just a few companies 0:13:21.240 --> 0:13:22.480 are controlling. 0:13:22.000 --> 0:13:23.360 It, but that is the world we live in. 0:13:24.000 --> 0:13:25.800 I don't think so, I think more and more. 0:13:25.840 --> 0:13:29.040 What we're seeing is that with open source you can 0:13:29.200 --> 0:13:32.480 actually distribute power more and you want to reduce the 0:13:32.520 --> 0:13:35.719 gap between the most powerful companies and the rest of 0:13:35.760 --> 0:13:40.719 the world, not only other companies but policy makers, non profits, 0:13:40.920 --> 0:13:42.439 academia and all of that. 0:13:42.600 --> 0:13:44.080 And that's the purpose of open source. 0:13:44.120 --> 0:13:47.680 It reduces the gap between the most powerful companies and 0:13:47.720 --> 0:13:50.280 the rest of the world, and that's what creates kind 0:13:50.280 --> 0:13:54.880 of like a sustainable, balanced future for AI and technology. 0:13:55.160 --> 0:13:57.000 It's a conversation where you're going to have all day 0:13:57.040 --> 0:14:00.520 long and you point out that basically open source allows 0:14:01.360 --> 0:14:02.960 more groups to go. 0:14:02.920 --> 0:14:03.600 To work on it. 0:14:03.679 --> 0:14:07.880 The problem is, as we're learning the tens of billions 0:14:07.880 --> 0:14:12.480 of dollars it takes to train models with yes, tens 0:14:12.520 --> 0:14:15.440 or hundreds of billions of parameters, and then you go 0:14:15.520 --> 0:14:18.680 lower down and now what we're hearing is that actually 0:14:18.679 --> 0:14:20.920 there are the folks doing this running out of cash? 0:14:21.120 --> 0:14:22.640 Are you seeing that as well? 0:14:23.040 --> 0:14:25.560 So it's more less true because for example, now you 0:14:25.600 --> 0:14:30.360 can use Lamastri that really has been closely but that 0:14:30.880 --> 0:14:34.880 meta has released and finds units for a very small 0:14:35.040 --> 0:14:37.840 amount of money. That's why I'm hugging face. There's over 0:14:38.080 --> 0:14:41.680 one million models that have been trained by companies, and 0:14:41.720 --> 0:14:43.840 a lot of these companies are very small and don't 0:14:43.840 --> 0:14:46.520 have like really really big big budget. I feel like 0:14:46.640 --> 0:14:50.880 today every single company has to build their own AI 0:14:51.840 --> 0:14:55.800 otherwise they run the risk of being left behind. And 0:14:55.840 --> 0:14:58.840 that's what we're seeing, and it doesn't require any more 0:14:58.840 --> 0:15:02.440 really really big budget. An interesting point though, that we're 0:15:02.480 --> 0:15:05.400 going to see this year though, is that we'll need 0:15:05.440 --> 0:15:10.760 to find for AI companies better business models. That's what 0:15:11.120 --> 0:15:14.280 kind of like you hinted at, something we really focused 0:15:14.280 --> 0:15:17.240 on at Talking Face. We are looking grateful to be 0:15:17.520 --> 0:15:22.000 close to profitable, which is very unusual for AI companies. 0:15:22.520 --> 0:15:25.760 But we're starting to see that there are some ways 0:15:25.800 --> 0:15:32.680 to generate revenue and not burn insane amounts of compute 0:15:33.520 --> 0:15:34.720 for AI startups today. 0:15:34.800 --> 0:15:38.720 I mean, how on that profitability perspective of yours, how 0:15:38.760 --> 0:15:40.600 many paying customers do you now have Can you give 0:15:40.640 --> 0:15:42.320 us an update. You've got a million models. What about 0:15:42.320 --> 0:15:43.040 paying customers? 0:15:43.280 --> 0:15:47.360 We have more than ten thousand paying customers out of 0:15:47.440 --> 0:15:51.120 the over one hundred thousand organizations, more than four million 0:15:51.200 --> 0:15:54.480 AI builders that are using our platform, and I think 0:15:54.520 --> 0:15:59.760 we found the right balance between monetizing, especially with like 0:16:00.160 --> 0:16:03.240 big companies that are using the platform in. 0:16:03.200 --> 0:16:05.520 Private enterprise companies exactly in. 0:16:05.560 --> 0:16:10.040 Order to fund all the free community, open source work 0:16:10.400 --> 0:16:12.760 that we're doing, and that is always going to stay 0:16:12.840 --> 0:16:13.760 open source and free. 0:16:13.760 --> 0:16:15.040 Of course, I want to go back to what I 0:16:15.160 --> 0:16:17.080 was saying though about people running out of cash. You 0:16:17.080 --> 0:16:22.080 actually put out a really interesting call on x basically saying, look, 0:16:22.120 --> 0:16:23.920 I'm here if you need me. Hugging face is here. 0:16:23.920 --> 0:16:26.520 If there are good people out there building interesting businesses 0:16:26.520 --> 0:16:28.080 but you're running out of money, we could be a 0:16:28.120 --> 0:16:32.120 home for you. Are you making acquisitions? Is it acquihiring 0:16:32.160 --> 0:16:32.960 that goes on them? 0:16:33.400 --> 0:16:36.239 We make some acquisitions. We're going to have interesting announcements 0:16:36.280 --> 0:16:37.000 in the next few weeks. 0:16:37.040 --> 0:16:39.160 I oh, don't taaser, but that's interesting. 0:16:39.400 --> 0:16:41.920 But I think in geneoin AI you're going to see 0:16:41.920 --> 0:16:45.120 more and more MNA because, as you said, I think 0:16:45.160 --> 0:16:49.160 a lot of companies took very risky bets. A lot 0:16:49.200 --> 0:16:51.640 of them are running out of money. And at the 0:16:51.680 --> 0:16:54.200 same time you have other companies like Hugging Face and 0:16:54.280 --> 0:16:59.080 others that are successful enough to be homes. Some of 0:16:59.120 --> 0:17:01.040 these M and A is going to weird, right, We've 0:17:01.040 --> 0:17:04.879 seen that happening in Lidit with Deck with some. 0:17:05.520 --> 0:17:09.879 Usual marriage with necessity rather than choice. 0:17:10.040 --> 0:17:10.480 May's. 0:17:10.760 --> 0:17:10.920 Yes. 0:17:11.680 --> 0:17:14.440 One thing that's good about summits like these Bloomberg Tech 0:17:14.720 --> 0:17:16.080 by the way, we can go around the room and 0:17:16.119 --> 0:17:17.840 ask who you're going to be shopping for. That's going 0:17:17.920 --> 0:17:20.359 to be interesting. But you get all these people in 0:17:20.359 --> 0:17:24.680 one place. You've also used the time that you've been 0:17:24.680 --> 0:17:28.560 in San Francisco because you're up in Seattle, right, Miami 0:17:28.640 --> 0:17:32.639 or Miami Apologies, you've been hiring, you've been interviewing candidates. 0:17:33.440 --> 0:17:35.680 Is that just a function of the best candidates being 0:17:35.760 --> 0:17:39.600 here in this city? How wide are you casting your net? 0:17:39.880 --> 0:17:40.080 Yeah? 0:17:40.119 --> 0:17:42.119 I think I think San Francisco is still the heart 0:17:42.440 --> 0:17:45.240 of technology and the I right, there's so much talent, 0:17:45.320 --> 0:17:49.720 so much so many interesting companies, so many interesting big 0:17:49.760 --> 0:17:53.119 technology companies being here that it's important for us to 0:17:53.240 --> 0:17:56.160 kind of like you have a foot on the ground here. 0:17:56.200 --> 0:17:58.679 We have a team already here, but we're also hiring 0:17:59.359 --> 0:18:01.600 community for hugging face here. 0:18:02.640 --> 0:18:03.720 Applications being taken. 0:18:04.080 --> 0:18:09.200 Yes, there's a really massive fight and struggle for AI 0:18:09.440 --> 0:18:16.679 talents right now with inflation of packages everywhere. But what 0:18:16.720 --> 0:18:18.800 we're seeing is that when when you have a mission 0:18:18.960 --> 0:18:23.160 that's like interesting two candidates like we open source, then 0:18:23.200 --> 0:18:25.399 you can attract really good talents. That's one of the 0:18:25.440 --> 0:18:28.119 reasons why. Also we're seeing big tech doing more and 0:18:28.119 --> 0:18:31.000 more open source. Right if you look at Meta with 0:18:31.119 --> 0:18:32.040 all the great work. 0:18:31.880 --> 0:18:34.960 That favorites, you keep on. I mean you've only really 0:18:34.960 --> 0:18:37.680 mentioned number three. You're trying to cajole Google into coming 0:18:37.720 --> 0:18:38.720 even more open source. 0:18:39.200 --> 0:18:42.640 I think as long as companies contribute to the world 0:18:42.680 --> 0:18:45.160 and to the field, if we've open source, we open research, 0:18:45.920 --> 0:18:48.680 I think it benefits everyone. I think we've we've lost 0:18:48.680 --> 0:18:50.920 a little bit this way in the US for the 0:18:50.960 --> 0:18:54.200 past few years. If you look at AI five years ago, 0:18:54.359 --> 0:18:57.080 most of it was open source and open science. It 0:18:57.200 --> 0:18:59.360 changed a little bit when some companies started to make 0:18:59.440 --> 0:19:03.119 money and and changing their approach to things. But I 0:19:03.119 --> 0:19:05.040 think it would be positive for the world to get 0:19:05.080 --> 0:19:08.440 back to an AI domain that is more open, more transparent, 0:19:08.560 --> 0:19:09.280 more inclusive. 0:19:09.560 --> 0:19:11.560 And you asked a calendar date now, by the way, 0:19:11.680 --> 0:19:13.600 because you told us you're going to announce. 0:19:13.280 --> 0:19:16.440 The news when you're ready four weeks. Yeah, we're holding you. 0:19:16.880 --> 0:19:19.760 Thanks clems on, our joy to have you with us 0:19:19.800 --> 0:19:22.280 and let you get to your breakfast where he's holding 0:19:22.320 --> 0:19:36.080 court here seeo hugging face. What a great conversation. Welcome 0:19:36.080 --> 0:19:38.600 back to this very special edition of Blue Meg Technology, 0:19:38.680 --> 0:19:41.720 Live in the Heart in San Francisco. All the grain 0:19:41.800 --> 0:19:44.160 and the good of industry movers shakers when it comes 0:19:44.160 --> 0:19:47.520 to artificial intelligence, in particular the academics, but the companies 0:19:47.560 --> 0:19:51.520 behind it, the CEOs, and notably also the investors. And 0:19:51.560 --> 0:19:53.399 this is an interesting one for the investor base. Right, 0:19:53.480 --> 0:19:56.159 we potentially have a new large language model getting at 0:19:56.160 --> 0:19:56.960 a decent evaluation. 0:19:57.080 --> 0:19:57.280 Yeah. 0:19:57.320 --> 0:19:59.280 So, I think what we reported last night is that 0:19:59.640 --> 0:20:03.639 xa I, the AI company started by Elon Musk and 0:20:04.359 --> 0:20:07.400 which he built out pretty quickly, is closing this kind 0:20:07.400 --> 0:20:12.400 of mega funding round eighteen billion dollar valuation. The thing 0:20:12.480 --> 0:20:15.000 is that we've learned right over the last year or 0:20:15.080 --> 0:20:18.480 more that is not actually that eyewatering. A number the 0:20:18.560 --> 0:20:20.679 numbers involved are not that iwak it wasn't he. 0:20:21.000 --> 0:20:24.280 Actually raising an awful lot considering an eighteen billion valuation. 0:20:24.440 --> 0:20:27.400 Yes, I think we've reported sort of up to six 0:20:27.440 --> 0:20:30.399 billion dollars. The thing is that the compute costs a 0:20:30.480 --> 0:20:33.399 mega and I took a phone call this morning saying, 0:20:33.920 --> 0:20:36.520 look past the cash and start asking whether the XAI 0:20:36.520 --> 0:20:39.000 has got access to the GPUs. Now Elon Musk has 0:20:39.040 --> 0:20:43.160 an existing relationship with Nvidio Jensen hung in the Tesla context, 0:20:43.160 --> 0:20:45.800 but it's a good gossip for the Bloomberg Tec event. 0:20:45.840 --> 0:20:47.480 It's a good thing to discuss well. 0:20:47.520 --> 0:20:50.119 And ultimately who are the investors? What we've seen I 0:20:50.160 --> 0:20:52.160 think really the rise in twenty twenty three and twenty 0:20:52.160 --> 0:20:53.879 twenty four, it's been corporate VC. 0:20:54.600 --> 0:20:56.560 Yes, of course, a strategic investor. 0:20:56.720 --> 0:20:59.679