WEBVTT 00:00:00.000 --> 00:00:02.839 Welcome to the podcast. We have a stacked lineup 00:00:02.839 --> 00:00:05.419 today. Apple's lawsuit with OpenAI over their 00:00:05.419 --> 00:00:07.700 trade secrets is heating up, especially as there's 00:00:07.700 --> 00:00:09.339 an IPO coming this year. So we're going to talk 00:00:09.339 --> 00:00:11.140 about what some of the negative repercussions 00:00:11.140 --> 00:00:13.720 on OpenAI are. And all of this is happening as 00:00:13.720 --> 00:00:16.780 Apple has just passed NVIDIA as the most valuable 00:00:16.780 --> 00:00:20.620 company in the world, unseating them. Moonshot's 00:00:20.620 --> 00:00:24.460 Kimi K3 is looking to match OpenAI's Opus 4 .8. 00:00:24.480 --> 00:00:29.829 Their valuation has hit $31 .5 billion. cities, 00:00:29.870 --> 00:00:33.090 Hochul, their governor, used AI to review every 00:00:33.090 --> 00:00:36.390 single rule in state law, but still decided to 00:00:36.390 --> 00:00:39.619 ban data centers. Google has renamed Notebook 00:00:39.619 --> 00:00:43.100 LM to Gemini Notebook and is adding code execution. 00:00:43.640 --> 00:00:46.079 And at the same time, an ex -DeepMind researcher 00:00:46.079 --> 00:00:49.460 has just raised $55 million at a $300 million 00:00:49.460 --> 00:00:53.359 valuation for a visual AI startup called Exlorian. 00:00:53.659 --> 00:00:57.100 Databricks also landed funding at a $188 billion 00:00:57.100 --> 00:00:59.840 valuation. We're getting into all of that on 00:00:59.840 --> 00:01:01.859 the podcast today. First, I wanted to tell you 00:01:01.859 --> 00:01:03.899 about the most exciting feature I have added 00:01:03.899 --> 00:01:07.099 to AI Box, which is an MCP, basically a connection. 00:01:07.280 --> 00:01:10.799 that allows Claude or any other AI tool to get 00:01:10.799 --> 00:01:13.040 all the other AI tools inside of it. Particularly 00:01:13.040 --> 00:01:15.579 exciting for me is that I've added image generation 00:01:15.579 --> 00:01:18.739 inside of Claude. You can also do audio and video, 00:01:18.879 --> 00:01:20.819 but image is what I'm really excited about. I've 00:01:20.819 --> 00:01:23.459 spent the entire day making a Facebook ad campaign, 00:01:23.599 --> 00:01:26.219 and every single ad I created with this tool 00:01:26.219 --> 00:01:29.099 was using the AI Box MCP. Essentially, what you 00:01:29.099 --> 00:01:31.060 do is when you're using Claude, all you have 00:01:31.060 --> 00:01:33.620 to do is say, generate an image of XYZ. You can 00:01:33.620 --> 00:01:35.459 put the regular prompt you'd put into ChatGPT 00:01:35.459 --> 00:01:38.079 or something else and just say, use AI Box, or 00:01:38.079 --> 00:01:39.959 it really just understands that that's a tool 00:01:39.959 --> 00:01:41.620 that can generate images. So you could just say, 00:01:41.659 --> 00:01:44.730 generate an image of XYZ. Claude will go. Use 00:01:44.730 --> 00:01:46.909 your AI box account, generate the image, pull 00:01:46.909 --> 00:01:49.170 it in and open it right there in your cloud desktop. 00:01:49.310 --> 00:01:51.290 So you can see the image inside of cloud. It 00:01:51.290 --> 00:01:53.650 feels magical. I'm not going to lie. And the 00:01:53.650 --> 00:01:54.950 thing that I love about it is that it can do 00:01:54.950 --> 00:01:57.709 multiple images simultaneously. So I'm like, 00:01:57.730 --> 00:01:59.870 I work with cloud to come up with like a template 00:01:59.870 --> 00:02:02.390 of an image that I like for the ad, but it needs, 00:02:02.409 --> 00:02:04.189 you know, images and little graphics inside of 00:02:04.189 --> 00:02:06.230 the ad. So I work on that and then I'm like, 00:02:06.310 --> 00:02:08.210 perfect. This looks great. Now make 20 different 00:02:08.210 --> 00:02:10.169 variations of this, you know, change the title 00:02:10.169 --> 00:02:12.289 and headline and change these graphics, go use 00:02:12.289 --> 00:02:14.860 AI box for all of that. So. It uses AI box to 00:02:14.860 --> 00:02:16.580 generate all the little images and graphics and 00:02:16.580 --> 00:02:18.819 pulls them into its template it made and it spits 00:02:18.819 --> 00:02:21.180 out 20 of them. And it does it in like a few 00:02:21.180 --> 00:02:23.219 seconds because it can do simultaneous image 00:02:23.219 --> 00:02:25.699 generation on AI box. That is what I'm super 00:02:25.699 --> 00:02:26.979 excited about. If you want to try it out, it 00:02:26.979 --> 00:02:30.960 is AI box dot AI slash MCP. Okay, let's talk 00:02:30.960 --> 00:02:34.430 about the lawsuit between Apple and OpenAI. I 00:02:34.430 --> 00:02:36.189 think a lot of people are talking about this 00:02:36.189 --> 00:02:39.509 right now, particularly because the story broke 00:02:39.509 --> 00:02:43.310 that Apple has passed Nvidia to claim the world's 00:02:43.310 --> 00:02:45.569 most valuable company title. So now we have the 00:02:45.569 --> 00:02:47.650 biggest company in the world coming against opening 00:02:47.650 --> 00:02:50.449 eye, which is looking to IPO soon. Apple hit 00:02:50.449 --> 00:02:54.229 a $4 .88 trillion market value on Friday. And 00:02:54.229 --> 00:02:56.729 the reason for this isn't just Apple growing, 00:02:56.870 --> 00:03:00.349 but also Nvidia falling. Nvidia went down 3%, 00:03:00.349 --> 00:03:04.659 which basically ended their run. the top in the 00:03:04.659 --> 00:03:06.500 entire market which they've been sitting there 00:03:06.500 --> 00:03:09.300 for quite some time apple's valuation while it 00:03:09.300 --> 00:03:12.080 hit that um it surpassed nvidia's valuation of 00:03:12.080 --> 00:03:15.229 4 .8 trillion Nvidia shares fell, like I said, 00:03:15.330 --> 00:03:18.569 Apple's up 22 % this year, and Nvidia has gained 00:03:18.569 --> 00:03:21.810 about 7%. It's interesting because we're seeing, 00:03:21.930 --> 00:03:24.590 you know, Nvidia that had such a good run for 00:03:24.590 --> 00:03:26.930 so long, and we're shifting out of that being 00:03:26.930 --> 00:03:29.169 the hottest segment. And a lot of Wall Street 00:03:29.169 --> 00:03:32.050 has rotated towards memory and infrastructure. 00:03:32.469 --> 00:03:35.389 Companies like Micron Technologies and SanDisk 00:03:35.389 --> 00:03:38.389 are becoming really popular. And so it feels 00:03:38.389 --> 00:03:41.229 like, you know, maybe Nvidia isn't just the hottest, 00:03:41.250 --> 00:03:43.729 sexiest company. Now, while this is happening, 00:03:43.789 --> 00:03:45.610 Apple is doing incredibly well in the stock market 00:03:45.610 --> 00:03:48.930 and their lawsuit of OpenAI is pushing forward. 00:03:49.150 --> 00:03:52.229 They are alleging that OpenAI hired over 400 00:03:52.229 --> 00:03:55.069 former Apple employees and they misused trade 00:03:55.069 --> 00:03:57.650 secrets through their chief hardware officer. 00:03:57.909 --> 00:04:00.430 And of course, with the IPO that's coming up, 00:04:00.449 --> 00:04:02.840 this is something that... OpenAI obviously doesn't 00:04:02.840 --> 00:04:05.080 want to have, you know, tarnishing their name. 00:04:05.400 --> 00:04:08.159 Apple named OpenAI's chief hardware officer in 00:04:08.159 --> 00:04:11.400 this whole complaint. And this is all happening 00:04:11.400 --> 00:04:13.240 at the same time that OpenAI has released their 00:04:13.240 --> 00:04:16.120 first piece of hardware. It is a $230 keyboard 00:04:16.120 --> 00:04:19.660 for Codex, their coding tool, which I have purchased, 00:04:19.699 --> 00:04:21.319 by the way. So I'll give you guys a review when 00:04:21.319 --> 00:04:22.980 that comes in the mail. I'm actually really excited 00:04:22.980 --> 00:04:25.160 about that. Was it a spur of the moment purchase? 00:04:25.339 --> 00:04:27.300 Yeah, I actually saw the news drop that this 00:04:27.300 --> 00:04:28.879 thing was available and I was covering it on 00:04:28.879 --> 00:04:31.769 the AI Hustle podcast. And like mid -podcast, 00:04:31.949 --> 00:04:33.769 Jamie starts talking and I'm like, okay, I gotta 00:04:33.769 --> 00:04:35.670 buy this keyboard because it said while supplies 00:04:35.670 --> 00:04:38.170 lasted. I knew the supplies were not gonna last. 00:04:38.329 --> 00:04:41.509 So I went and purchased it. Side note, OpenAI 00:04:41.509 --> 00:04:43.750 also released a basketball that they're selling 00:04:43.750 --> 00:04:46.269 and it just looks like a weird gradient color 00:04:46.269 --> 00:04:48.610 basketball with the OpenAI logo on it. That also 00:04:48.610 --> 00:04:50.649 said supplies last. I just, I saw this and that 00:04:50.649 --> 00:04:52.430 came out, I guess, a few hours ago. I went and 00:04:52.430 --> 00:04:54.050 looked and that was already sold out. So the 00:04:54.050 --> 00:04:55.490 basketball and the keyboard are sold out, but 00:04:55.490 --> 00:04:57.410 it looks like OpenAI is making these physical 00:04:57.410 --> 00:04:59.290 products, which is kind of funny. Will this save 00:04:59.290 --> 00:05:01.449 them from... their IPO woes? I don't think so. 00:05:01.990 --> 00:05:03.930 But it's going to be interesting because that 00:05:03.930 --> 00:05:06.069 was their first piece of hardware that they're 00:05:06.069 --> 00:05:08.470 releasing and they're working with the former 00:05:08.470 --> 00:05:12.050 Apple iPhone designer, Johnny Ive, to come up 00:05:12.050 --> 00:05:15.129 with a new undisclosed piece of hardware. Some 00:05:15.129 --> 00:05:16.829 people say it's basically like a smart speaker 00:05:16.829 --> 00:05:19.089 on your computer. In any case, I've looked over 00:05:19.089 --> 00:05:21.490 a lot of the accusations Apple's made. It looks 00:05:21.490 --> 00:05:25.970 pretty bad. Basically, they brought over a new 00:05:25.970 --> 00:05:28.009 chief of hardware and he came from Apple and 00:05:28.009 --> 00:05:30.500 they gutted the entire... design team of Apple. 00:05:30.560 --> 00:05:32.660 They hired like 400 people from the design team 00:05:32.660 --> 00:05:34.519 of Apple. You know, he used to be over there 00:05:34.519 --> 00:05:36.699 at Apple. So he knew everyone. He pulled them 00:05:36.699 --> 00:05:38.720 in. And when he pulled them in, Apple's allegation 00:05:38.720 --> 00:05:41.019 is that he told all of them as they were coming 00:05:41.019 --> 00:05:43.259 to OpenAI, this is how you get around all of 00:05:43.259 --> 00:05:45.699 the, basically when you quit Apple, how you get 00:05:45.699 --> 00:05:48.319 around all of the, you know, retaining all the 00:05:48.319 --> 00:05:50.439 information. So bring little components, give 00:05:50.439 --> 00:05:52.040 us the secret words, tell us what's going on 00:05:52.040 --> 00:05:54.980 at Apple. And in some case, they said that one 00:05:54.980 --> 00:05:57.540 of the employees kept like a laptop and used 00:05:57.540 --> 00:06:00.199 some like VPN. end loophole to keep access to 00:06:00.199 --> 00:06:02.740 some data at Apple. That's kind of Apple's lawsuit 00:06:02.740 --> 00:06:05.060 in any case. I mean, it's an important lesson 00:06:05.060 --> 00:06:07.540 though, when your employees leave, everything 00:06:07.540 --> 00:06:10.399 in their head also leaves with them. So what 00:06:10.399 --> 00:06:12.379 type of confidential information do they move 00:06:12.379 --> 00:06:15.000 over? That's what the lawsuit is going to be 00:06:15.000 --> 00:06:18.959 covering. Moonshot's Kimmy3 is looking to beat 00:06:18.959 --> 00:06:21.579 Anthropic right now. And on one important benchmark, 00:06:21.699 --> 00:06:24.060 which is a front end developer benchmark, it 00:06:24.060 --> 00:06:28.410 is... ahead of Fable 5 even. So it's really doing 00:06:28.410 --> 00:06:30.350 well. It's a two to three trillion parameter 00:06:30.350 --> 00:06:33.449 open weight model. And with all of this, it's 00:06:33.449 --> 00:06:36.310 coming from a Chinese lab. And the Chinese lab's 00:06:36.310 --> 00:06:39.589 valuation, Moonshot, jumped to $31 .5 billion 00:06:39.589 --> 00:06:42.709 in two months as they're coming out with this 00:06:42.709 --> 00:06:45.550 model. And it's beating even Anthropocon, a bunch 00:06:45.550 --> 00:06:47.879 of different... benchmarks. This is the largest 00:06:47.879 --> 00:06:49.759 open weight model that has ever been released 00:06:49.759 --> 00:06:52.720 from China. They raised $2 billion at a $20 billion 00:06:52.720 --> 00:06:54.899 valuation back in May, and now they're valued 00:06:54.899 --> 00:06:57.199 at 31 billion. So their valuation's increasing, 00:06:57.379 --> 00:06:59.220 their model's getting really good. This is all 00:06:59.220 --> 00:07:01.920 coming to a time with Anthropic, they're getting 00:07:01.920 --> 00:07:03.480 kind of held back by the government. It feels 00:07:03.480 --> 00:07:04.980 like we're getting a lot of regulation, these 00:07:04.980 --> 00:07:08.240 30 day kind of hold back periods where the government 00:07:08.240 --> 00:07:10.199 can go and review the models before they release 00:07:10.199 --> 00:07:11.819 them to the public. None of that is happening 00:07:11.819 --> 00:07:13.860 in China and they're releasing all of these models. 00:07:14.240 --> 00:07:16.540 open weight so people can actually get them and 00:07:16.540 --> 00:07:19.040 fine tune their own models on them. A lot of 00:07:19.040 --> 00:07:20.959 enterprise buyers prefer this because if you 00:07:20.959 --> 00:07:23.600 can grab something like Kimi K3, put it on your 00:07:23.600 --> 00:07:26.139 own hardware, it's not tied to China, it's not 00:07:26.139 --> 00:07:28.019 tied to Moonshot. They don't see into anything 00:07:28.019 --> 00:07:30.040 that you're doing. You fine tune your own stuff 00:07:30.040 --> 00:07:34.399 on this system. And then you can basically access 00:07:34.399 --> 00:07:36.879 it for quote unquote free, but you're using it 00:07:36.879 --> 00:07:39.959 with... all on your own hardware. Now, some people 00:07:39.959 --> 00:07:41.920 are like, hey, don't use the Chinese models. 00:07:42.120 --> 00:07:44.040 I think maybe the play for the Chinese models 00:07:44.040 --> 00:07:46.139 are open, basically release a bunch of these 00:07:46.139 --> 00:07:48.680 open weight models, get people fine tuning on 00:07:48.680 --> 00:07:50.920 top of them. So they're really, they're kind 00:07:50.920 --> 00:07:53.060 of embedded in your ecosystem basically. And 00:07:53.060 --> 00:07:55.519 then you could probably release better models 00:07:55.519 --> 00:07:58.600 that maybe are paid, or you could just say, hey, 00:07:58.620 --> 00:08:00.300 look, if you don't want to run it on your own 00:08:00.300 --> 00:08:03.220 hardware, you can also run it on our, on our 00:08:03.220 --> 00:08:05.759 data, on our like servers and stuff, which is, 00:08:05.779 --> 00:08:09.339 which is a pocket. strategy. Right now we have 00:08:09.339 --> 00:08:12.379 New York's governor, Governor Kathy Hochul. She 00:08:12.379 --> 00:08:15.060 said that she used AI to review every state rule, 00:08:15.160 --> 00:08:17.160 regulation, and policy in the last couple months. 00:08:17.360 --> 00:08:19.939 And she said that the work they were able to 00:08:19.939 --> 00:08:22.279 do in reviewing all of this in the last couple 00:08:22.279 --> 00:08:24.019 months, she said it would have taken the staff 00:08:24.019 --> 00:08:28.339 five years if they didn't have AI. This, in my 00:08:28.339 --> 00:08:30.699 opinion, is showing a massive gap right now. 00:08:30.740 --> 00:08:33.120 New York is using AI to streamline, and particularly 00:08:33.120 --> 00:08:36.139 Kathy is using AI to streamline the government. 00:08:37.289 --> 00:08:39.169 she's doing all of this while at the exact same 00:08:39.169 --> 00:08:43.419 time she put a one year ban on new data center 00:08:43.419 --> 00:08:46.879 build -outs inside of their state. So she's using 00:08:46.879 --> 00:08:48.840 the AI, but she's banning it. It feels like all 00:08:48.840 --> 00:08:52.360 of the people that fly on private jets to environmental 00:08:52.360 --> 00:08:56.379 conferences and burn a billion pounds of fuel 00:08:56.379 --> 00:08:59.519 and then tell everyone not to drive their car 00:08:59.519 --> 00:09:02.740 because it uses too much fuel. So that's kind 00:09:02.740 --> 00:09:06.320 of what it feels like to me. She said with what 00:09:06.320 --> 00:09:08.279 she was able to do with the AI though in reviewing 00:09:08.279 --> 00:09:10.379 all the laws, which by the way, phenomenal. So 00:09:10.379 --> 00:09:12.149 I'm not like... not begging on everything. I 00:09:12.149 --> 00:09:14.289 think that's amazing if she can use AI to streamline 00:09:14.289 --> 00:09:16.929 all of the laws. She said she found a bunch of 00:09:16.929 --> 00:09:20.149 outdated laws. One of them was a $25 dog hunting 00:09:20.149 --> 00:09:23.169 fee. If you wanted to go hunting with your dog, 00:09:23.250 --> 00:09:26.870 you had to pay a $25 fee. Apparently there was 00:09:26.870 --> 00:09:29.870 a fee for pregnant workers. They had to go get 00:09:29.870 --> 00:09:31.990 a permit if they wanted to work after midnight. 00:09:32.190 --> 00:09:34.610 There's just like all these random laws and regulations. 00:09:35.289 --> 00:09:37.269 And you know, this just kind of happens when 00:09:37.269 --> 00:09:40.950 a a state has been around for so long and all 00:09:40.950 --> 00:09:42.750 these regulations get added no one ever goes 00:09:42.750 --> 00:09:46.409 and takes the regulations away so i'm all for 00:09:46.409 --> 00:09:48.230 deregulating and i think they did you know a 00:09:48.230 --> 00:09:50.350 great job they found a bunch of things hoshel 00:09:50.350 --> 00:09:53.090 didn't say what ai model they used for all of 00:09:53.090 --> 00:09:56.570 the reviews and they didn't say what their methodology 00:09:56.570 --> 00:10:00.320 was for for doing all this but Overall, she said 00:10:00.320 --> 00:10:03.500 that this was a huge productivity jump. She said 00:10:03.500 --> 00:10:06.100 it's 30x productivity jump on what she was able 00:10:06.100 --> 00:10:08.519 to do. So she gets some points for looking like 00:10:08.519 --> 00:10:12.600 the cool governor for using AI to streamline. 00:10:12.740 --> 00:10:14.480 And honestly, I hope every state does that. So 00:10:14.480 --> 00:10:16.740 good on her for that. Bad on her for banning 00:10:16.740 --> 00:10:19.120 data centers for a year. Although it doesn't 00:10:19.120 --> 00:10:20.379 really matter because every other state will 00:10:20.379 --> 00:10:23.769 get them instead, I guess. Okay. Google has renamed 00:10:23.769 --> 00:10:26.429 Notebook LM to Gemini Notebook. I mean, it makes 00:10:26.429 --> 00:10:28.230 sense. They're trying to keep with the branding. 00:10:28.289 --> 00:10:30.429 Google loves changing the name and changing the 00:10:30.429 --> 00:10:32.169 products and killing the products of everything. 00:10:32.309 --> 00:10:34.870 I'm just happy that Notebook LM is still alive. 00:10:35.169 --> 00:10:37.309 They've added code execution to it. It's funny 00:10:37.309 --> 00:10:39.210 because they just basically had a runaway product 00:10:39.210 --> 00:10:41.429 that was random. They released tons of products. 00:10:41.570 --> 00:10:43.490 Notebook LM just caught fire for some reason. 00:10:43.509 --> 00:10:45.090 I think a lot of students just would throw their 00:10:45.090 --> 00:10:47.379 textbooks into it and have it. do a podcast talking 00:10:47.379 --> 00:10:49.960 back to them. They liked that. But in any case, 00:10:50.000 --> 00:10:53.299 it can now do code execution. It has a secured 00:10:53.299 --> 00:10:56.080 cloud computer that can write code against your 00:10:56.080 --> 00:10:58.600 own documents for data analysis. The app has 00:10:58.600 --> 00:11:01.840 30 million users. 600 ,000 organizations are 00:11:01.840 --> 00:11:04.519 using Notebook LM. Honestly, it blows my mind. 00:11:04.639 --> 00:11:06.559 I mean, I've played with it. I've used it. I 00:11:06.559 --> 00:11:08.559 know people that religiously use it. It's not 00:11:08.559 --> 00:11:11.820 part of my go -to, but it blows my mind that 00:11:11.820 --> 00:11:13.720 they have 30 million users and now it's called 00:11:13.720 --> 00:11:15.620 Gemini Notebook. So they're just giving Gemini. 00:11:15.789 --> 00:11:18.570 a better name, and a little bit of a good PR 00:11:18.570 --> 00:11:21.230 boost with this popular tool. The code execution 00:11:21.230 --> 00:11:23.730 is live today, so you can go use that for Google 00:11:23.730 --> 00:11:26.429 AI Ultra subscribers and for Workspace customers. 00:11:26.870 --> 00:11:29.110 It's rolling out to all pro users over the coming 00:11:29.110 --> 00:11:31.509 weeks. Notebooks will now sync with the main 00:11:31.509 --> 00:11:33.730 Gemini app, which honestly, I'm all for that. 00:11:33.830 --> 00:11:35.509 I hate it when there's all of these different 00:11:35.509 --> 00:11:37.610 tools and they're not connected, especially when 00:11:37.610 --> 00:11:39.970 it's from the same company. Google really should 00:11:39.970 --> 00:11:41.889 have everything connected, so I think that that's 00:11:41.889 --> 00:11:43.870 great. It's going to appear inside of Google's 00:11:43.870 --> 00:11:46.059 search AI mode. basically let you pull answers 00:11:46.059 --> 00:11:48.179 from your own documents instead of just the web, 00:11:48.320 --> 00:11:50.720 which is really cool, right? If you have a whole 00:11:50.720 --> 00:11:53.179 bunch of documents on your Google Drive and you 00:11:53.179 --> 00:11:55.000 want to, you know, and you go in and search in 00:11:55.000 --> 00:11:56.580 the search bar on Google and all of a sudden 00:11:56.580 --> 00:11:58.340 it gives you answers from that, that could be 00:11:58.340 --> 00:12:01.039 a really useful bit of context. The app started 00:12:01.039 --> 00:12:04.539 as a project at Google I .O. back in 2023 and 00:12:04.539 --> 00:12:06.899 now it has grown. It got really famous and viral 00:12:06.899 --> 00:12:09.830 from TikTok. Okay, all of this is at the same 00:12:09.830 --> 00:12:12.950 time that Andrew Day, a former Google DeepMind 00:12:12.950 --> 00:12:17.669 researcher, raised $55 million for Alorean at 00:12:17.669 --> 00:12:21.230 a $30 million valuation. He did all of this before 00:12:21.230 --> 00:12:23.730 the startup shipped anything. Dai is basically 00:12:23.730 --> 00:12:26.610 betting that visual reasoning teaching AI to 00:12:26.610 --> 00:12:28.970 understand images and video the way the current 00:12:28.970 --> 00:12:31.610 systematic text handles is the next frontier. 00:12:31.830 --> 00:12:33.610 And he basically turned down a whole bunch of 00:12:33.610 --> 00:12:35.450 really high offers to take strategic investors 00:12:35.450 --> 00:12:39.769 like NVIDIA who understand how expensive AI research 00:12:39.769 --> 00:12:42.730 actually is. So NVIDIA is one of his investors. 00:12:42.889 --> 00:12:44.529 A bunch of other people who got turned down. 00:12:44.629 --> 00:12:47.309 Dai spent over a decade at DeepMind on foundational 00:12:47.309 --> 00:12:50.909 work that all later influenced ChatGPT. Lorian's 00:12:50.909 --> 00:12:53.590 valuation to capital ratio is way more aggressive 00:12:53.590 --> 00:12:56.789 than even thinking machines from... a former 00:12:56.789 --> 00:12:59.429 co -founder of OpenAI, which was one of the largest 00:12:59.429 --> 00:13:02.049 seed rounds in US history. Google, OpenAI, and 00:13:02.049 --> 00:13:04.259 Anthropo are all working on multimodal. Understanding, 00:13:04.379 --> 00:13:08.019 which is a recent Stanford benchmark. And that 00:13:08.019 --> 00:13:09.539 benchmark in particular showed that frontier 00:13:09.539 --> 00:13:13.299 vision models are still lagging and they're worse 00:13:13.299 --> 00:13:15.820 than a toddler on basic visual reasoning tasks. 00:13:16.019 --> 00:13:18.820 So he's trying to crack the code and we'll see 00:13:18.820 --> 00:13:22.799 if that is possible. Databricks has landed funding 00:13:22.799 --> 00:13:27.360 at a $188 billion valuation. They just closed 00:13:27.360 --> 00:13:30.080 this funding round. They made it basically in 00:13:30.080 --> 00:13:32.259 the same league as OpenAI and Anthropic on paper 00:13:32.259 --> 00:13:35.600 anyways. The data and AI platform company, they're 00:13:35.600 --> 00:13:37.919 not just building AI models. They're also selling 00:13:37.919 --> 00:13:40.220 a lot of the infrastructure that companies use 00:13:40.220 --> 00:13:42.220 to train, deploy, and run models in production. 00:13:42.779 --> 00:13:45.059 Databricks right now is competing with Snowflake, 00:13:45.059 --> 00:13:48.179 if you've heard of the company, on data warehousing. 00:13:48.669 --> 00:13:51.690 And they're also competing a bunch of other big 00:13:51.690 --> 00:13:55.990 players on modeling serving. They're also strengthening 00:13:55.990 --> 00:13:58.570 their acquisition. They acquired Mosaic ML, which 00:13:58.570 --> 00:14:01.710 I covered on the podcast back in 2023. I don't 00:14:01.710 --> 00:14:02.990 know why, it's kind of funny. That particular 00:14:02.990 --> 00:14:06.850 episode I covered from the San Francisco airport, 00:14:07.210 --> 00:14:10.570 which is funny to me. But anyways, the company 00:14:10.570 --> 00:14:13.690 has basically not done an IPO. So they're not 00:14:13.690 --> 00:14:15.809 publicly listed. They've just done these private 00:14:15.809 --> 00:14:19.500 funding rounds. kind of wild that we have these 00:14:19.500 --> 00:14:21.580 massive ai companies that are doing that right 00:14:21.580 --> 00:14:24.200 now at this new valuation they're now valued 00:14:24.200 --> 00:14:26.580 higher than most established enterprise software 00:14:26.580 --> 00:14:29.059 companies and i I think basically they pushed 00:14:29.059 --> 00:14:30.980 the bar for what it takes to go public because 00:14:30.980 --> 00:14:34.740 at $188 billion, you would have expected this 00:14:34.740 --> 00:14:37.360 type of company to go public in the past. Guys, 00:14:37.419 --> 00:14:38.919 thank you so much for tuning into the podcast. 00:14:39.000 --> 00:14:40.820 If you enjoyed this episode, make sure to leave 00:14:40.820 --> 00:14:42.460 a rating and review wherever you get your podcasts. 00:14:42.700 --> 00:14:45.179 And as always, make sure to go check out AIbox 00:14:45.179 --> 00:14:48.600 .ai if you want to get access to over 80 different 00:14:48.600 --> 00:14:50.740 AI models and the MCP so you can pull all those 00:14:50.740 --> 00:14:53.159 AI models into Cloud. You can get Cloud to generate 00:14:53.159 --> 00:14:56.629 images, audio, video. It will save you so much 00:14:56.629 --> 00:14:59.190 time. It's amazing. It's a buttery smooth experience 00:14:59.190 --> 00:15:01.669 that I've been doing all week long and I love 00:15:01.669 --> 00:15:04.149 it. So if you wanna check it out, AIbox .ai slash 00:15:04.149 --> 00:15:06.389 MCP. All right, I'll catch you guys all in the 00:15:06.389 --> 00:15:06.950 next episode.