1 00:00:00,600 --> 00:00:03,720 Hi, I'm Eric Newcomer. And I am Madeline Renbarger. 2 00:00:04,120 --> 00:00:06,960 And this is the newcomer podcast. 3 00:00:10,280 --> 00:00:14,000 Each week, Eric and I discuss the VC deals and the drama that 4 00:00:14,000 --> 00:00:15,800 went down. Let's do it. 5 00:00:16,079 --> 00:00:20,600 Here we go. Welcome back to the podcast. 6 00:00:20,600 --> 00:00:24,920 Lots to cover in AI this week. Big wins, big raises, some 7 00:00:24,920 --> 00:00:27,640 dramatic, not so wonderful exits. 8 00:00:27,640 --> 00:00:29,360 A lot of. Chest thumping. 9 00:00:29,360 --> 00:00:31,520 You know, I feel like everyone wants to make sure they're 10 00:00:31,520 --> 00:00:35,960 getting headlines this week, but yeah, 1, I don't know, happy 11 00:00:35,960 --> 00:00:39,640 failure or failure that exceeds expectations. 12 00:00:40,000 --> 00:00:42,080 We're we're of course talking about humane. 13 00:00:42,320 --> 00:00:46,160 HP agreed to acquire Humane for 116 million. 14 00:00:46,240 --> 00:00:49,480 It's much less than I think the total they raised was a little 15 00:00:49,480 --> 00:00:52,040 over 230 million over their lifespan. 16 00:00:52,040 --> 00:00:55,520 But it's not nothing for a product that had such a 17 00:00:55,520 --> 00:00:59,080 disastrous rollout as Humane did when it first launched. 18 00:00:59,400 --> 00:01:01,920 Expectations are everything here. 19 00:01:01,920 --> 00:01:04,800 You know, they, they had the advantage of having this sort of 20 00:01:04,800 --> 00:01:08,200 Apple Sheen, you know, their former Apple executives key to 21 00:01:08,200 --> 00:01:12,440 some of Apple's product design. They're launching the first AI 22 00:01:12,440 --> 00:01:15,000 pin. So there's a lot of, Oh my God, 23 00:01:15,000 --> 00:01:19,640 this, this could be huge. And then it, you know, clearly 24 00:01:19,800 --> 00:01:23,800 didn't meet the hype. It's a brutal, brutal launch for 25 00:01:23,800 --> 00:01:25,480 them. I do feel like there's some 26 00:01:25,480 --> 00:01:28,080 points I've heard, some chatter, you know, that while it's 27 00:01:28,240 --> 00:01:30,160 definitely not the worst exit that could have happened for 28 00:01:30,160 --> 00:01:33,800 something like this, there's some kind of dark irony of, you 29 00:01:33,800 --> 00:01:36,640 know, some Apple engineers that went off to start their own 30 00:01:36,640 --> 00:01:39,760 thing that will be now leading a new division with an HP to 31 00:01:39,760 --> 00:01:43,720 integrate AI into PCs rather than, you know, building your 32 00:01:43,720 --> 00:01:46,280 own thing. I've seen a number of, you know, 33 00:01:46,280 --> 00:01:50,360 printer jokes and what I mean. It will be interesting to see, I 34 00:01:50,360 --> 00:01:53,640 think, you know, the actual sort of core. 35 00:01:53,920 --> 00:01:57,720 Do you want to talk to your device now that sort of AI has 36 00:01:57,720 --> 00:01:59,920 better voice recognition? We'll see. 37 00:01:59,920 --> 00:02:03,000 I mean, I, I feel like some of the experiments Humane was 38 00:02:03,000 --> 00:02:05,000 running are still sort of open questions. 39 00:02:05,000 --> 00:02:09,000 Also, do you need this sort of non screen holographic 40 00:02:09,440 --> 00:02:10,960 interface? It'll be it'll be really 41 00:02:10,960 --> 00:02:14,400 interesting to see what, if anything, HP integrates into 42 00:02:14,400 --> 00:02:18,240 their own products versus this is just an opportunity to get 43 00:02:18,240 --> 00:02:21,520 some very talented hardware executives into the company. 44 00:02:21,720 --> 00:02:24,360 And also, you know, AI native hardware executives, I think 45 00:02:24,360 --> 00:02:26,800 there's such a premium on AI talent still. 46 00:02:26,800 --> 00:02:29,800 This could just be viewed as a very savvy Aqua hire. 47 00:02:29,800 --> 00:02:32,320 And we'll see what they end up building with this. 48 00:02:32,320 --> 00:02:34,800 I don't think it was, you know, I think it was just maybe a case 49 00:02:34,800 --> 00:02:37,040 of something too early. There's lots of potential around 50 00:02:37,040 --> 00:02:39,760 this and lots of times the first ones are always the first to 51 00:02:39,760 --> 00:02:42,360 fail. But we'll see if, you know, AI 52 00:02:42,360 --> 00:02:44,720 devices without a screen really take off. 53 00:02:44,840 --> 00:02:48,440 This one clearly did not. But right, I see the vision. 54 00:02:48,560 --> 00:02:52,920 This fits into the sort of, I don't know, the Jet Quibi 55 00:02:54,320 --> 00:02:58,040 humane, I, I'd say the three companies, you know, Jet was the 56 00:02:58,040 --> 00:03:02,640 big e-commerce play Mark Laurie, Quibi was Katzenberg and Meg 57 00:03:02,640 --> 00:03:07,200 Whitman going big on short video and and humane, you know, this 58 00:03:07,640 --> 00:03:10,440 AI device that you can sort of wear and talk to. 59 00:03:10,760 --> 00:03:13,720 I mean, I think all of them represent sort of super hypey, 60 00:03:13,720 --> 00:03:17,640 lots of funding where immediately it was sort of like, 61 00:03:18,440 --> 00:03:22,960 do you really need to go big for this to work? 62 00:03:23,040 --> 00:03:28,840 You know, and jet, you know, the, the argument was where 63 00:03:28,840 --> 00:03:30,680 we're going against the Amazons, the world. 64 00:03:30,720 --> 00:03:33,680 So to have this scale, we need to like pour a bunch of money 65 00:03:33,680 --> 00:03:35,280 into it. I mean, it worked out for them. 66 00:03:35,360 --> 00:03:38,360 They sold to Walmart. So that was sort of the test 67 00:03:38,360 --> 00:03:39,720 case. I mean, some people think it was 68 00:03:40,360 --> 00:03:44,480 an extremely what a $3 billion acquire for for Mark Laurie. 69 00:03:44,480 --> 00:03:47,480 But and then Quibi was a disaster. 70 00:03:47,520 --> 00:03:49,720 Quibi. It did I I still think like, 71 00:03:50,120 --> 00:03:51,840 man, they could have done it small. 72 00:03:51,840 --> 00:03:55,640 Like I it's not clear for me on that one why it was. 73 00:03:55,640 --> 00:03:57,640 Just such a huge launch all at once, right? 74 00:03:57,680 --> 00:03:59,840 Right. Right, just like do, I mean, 75 00:03:59,840 --> 00:04:01,960 it's just like, I don't know, we're in the content business 76 00:04:01,960 --> 00:04:05,760 and like we, I wrote a we wrote, you know, I quit my job, but you 77 00:04:05,760 --> 00:04:08,400 know, I wrote some stories soft they could work. 78 00:04:08,400 --> 00:04:11,000 And then when people like the stories, we expanded out. 79 00:04:11,000 --> 00:04:13,760 I don't, I don't quite I understand that they wanted high 80 00:04:14,160 --> 00:04:17,519 production value content, but I, I don't know, I think they could 81 00:04:17,519 --> 00:04:19,680 have done it a little leaner than they did. 82 00:04:19,959 --> 00:04:23,040 I think humane, you know, it's like it's a hardware device. 83 00:04:23,040 --> 00:04:27,320 So on the one hand, you needed to spend money to to get and 84 00:04:27,320 --> 00:04:29,440 clearly they didn't even have enough money to do everything 85 00:04:29,440 --> 00:04:31,120 they want. It felt like there was too much 86 00:04:31,120 --> 00:04:35,320 Polish on like the device, like the actual physical, you know, 87 00:04:35,320 --> 00:04:38,360 in this sort of aptly sense that they had this like beautiful, 88 00:04:38,360 --> 00:04:41,480 like little device. And it felt like honestly, a lot 89 00:04:41,480 --> 00:04:44,760 of the challenge is the sort of voice communication. 90 00:04:44,760 --> 00:04:47,680 It also felt like they wanted this whole sort of interface. 91 00:04:47,680 --> 00:04:49,960 There's a reason that people like software companies as 92 00:04:49,960 --> 00:04:51,600 investments more than hardware companies. 93 00:04:51,760 --> 00:04:54,200 Software companies are leaner and you can iterate. 94 00:04:54,400 --> 00:04:56,720 And you'd think AI is making that more possible. 95 00:04:56,720 --> 00:04:59,680 But for some reason, Humane decided we need to build a 96 00:04:59,680 --> 00:05:03,920 custom device just for AI. And I mean, maybe they think 97 00:05:03,920 --> 00:05:07,400 they just didn't get enough money to prove it, but yeah, 98 00:05:07,680 --> 00:05:09,640 they didn't prove it. I mean, it doesn't seem like a 99 00:05:09,640 --> 00:05:13,360 lot of these AI hardware devices have much utility beyond what 100 00:05:13,360 --> 00:05:16,080 you could already do on a smartphone or a different device 101 00:05:16,080 --> 00:05:17,760 that you already have, which then you should. 102 00:05:17,760 --> 00:05:21,320 Just be a magical exactly yes the. 103 00:05:21,400 --> 00:05:24,040 Rabbit R Pin that was, you know, I'm a humane competitor. 104 00:05:24,080 --> 00:05:27,520 And again, it fell into the same problem of what's the use case 105 00:05:27,520 --> 00:05:30,560 for this if I have a phone like it? 106 00:05:30,560 --> 00:05:34,120 It seems to be that this whole category, while exciting and I 107 00:05:34,200 --> 00:05:36,920 appreciate the need to, I appreciate the desire to be 108 00:05:36,920 --> 00:05:39,960 removed to some extent from screens and be able to move more 109 00:05:39,960 --> 00:05:42,360 fully throughout the world and interact with technology. 110 00:05:42,600 --> 00:05:45,560 But like with any wearable that we've seen outside of, you know, 111 00:05:45,560 --> 00:05:49,000 things that enhance your screens like Airpods or Apple Watch, you 112 00:05:49,000 --> 00:05:50,800 know, it doesn't seem like these are. 113 00:05:50,840 --> 00:05:54,280 Those are the two devices exactly like Apple Watch or 114 00:05:54,280 --> 00:05:56,320 Airpods. These are devices we already 115 00:05:56,320 --> 00:06:01,240 have that in some ways are meant to get us off our screens. 116 00:06:01,840 --> 00:06:04,080 And I mean, the problem is they're they're locked down and 117 00:06:04,080 --> 00:06:08,320 sort of controlled by Apple. But you would think somebody 118 00:06:08,320 --> 00:06:12,120 would come up with a strategy that, yeah, on on the watch or 119 00:06:12,120 --> 00:06:15,400 on our Airpods that gets us away from screens and allows us to 120 00:06:15,400 --> 00:06:18,760 interact with AI. Unfortunately, what Apple is 121 00:06:18,760 --> 00:06:22,480 still in the world of Siri, which is what the worst voice 122 00:06:22,600 --> 00:06:25,120 voice platform around. I mean, we haven't even talked. 123 00:06:25,120 --> 00:06:27,840 Sorry, I just well, we haven't even talked about, you know, 124 00:06:27,840 --> 00:06:31,200 Amazon, which dumped tons of money into like any Lexus into 125 00:06:31,200 --> 00:06:33,760 everybody's house. I have someone in one in a box 126 00:06:33,760 --> 00:06:37,440 somewhere. And I mean, I'm, I'm curious 127 00:06:37,440 --> 00:06:39,080 what you think about this. I'm not. 128 00:06:39,680 --> 00:06:42,760 And you know, we're friends of my some of my best friends run a 129 00:06:42,760 --> 00:06:46,520 voice company and they're big voice bulls. 130 00:06:46,520 --> 00:06:49,520 Godspeed. But I don't know, I, I feel like 131 00:06:49,600 --> 00:06:52,880 I'm the class I write professionally, so I, my bias is 132 00:06:52,880 --> 00:06:56,600 clear, you know, I guess I'm on a podcast, but I consume mostly 133 00:06:56,600 --> 00:06:58,760 writing. I think writing is a very 134 00:06:58,760 --> 00:07:02,880 efficient way to get information and I don't really see myself 135 00:07:02,880 --> 00:07:07,400 wanting to go from typing to a computer to talking to one 136 00:07:07,760 --> 00:07:09,440 necessarily. I don't know, do you feel 137 00:07:09,440 --> 00:07:12,680 differently? Well, I do see there being some 138 00:07:12,680 --> 00:07:17,080 utility in voice, maybe more as a listening vehicle to get off 139 00:07:17,080 --> 00:07:20,200 of your phone and interact with more directly in that sense. 140 00:07:20,200 --> 00:07:23,400 For either an assistant, if it, again, the key is always if it 141 00:07:23,400 --> 00:07:26,840 works seamlessly, then yeah, I could find that to some degree 142 00:07:26,840 --> 00:07:29,000 useful. If I'm, you know, trying to 143 00:07:29,000 --> 00:07:31,600 complete some task that normally would require a screen, but I 144 00:07:31,600 --> 00:07:34,080 could just do it with my voice in my home in sort of the 145 00:07:34,080 --> 00:07:37,560 fantasy of what an Alexa or Siri could do as an agent. 146 00:07:37,640 --> 00:07:40,080 That would be awesome. I would absolutely appreciate 147 00:07:40,080 --> 00:07:42,760 that because I could, you know, focus when I'm on my screen, I 148 00:07:42,760 --> 00:07:45,280 could focus on work and when I think of something, I could 149 00:07:45,280 --> 00:07:47,360 instantly complete that task. And that would definitely be 150 00:07:47,360 --> 00:07:48,440 helpful. However. 151 00:07:49,600 --> 00:07:52,040 Give us you need to what? What type of task are you 152 00:07:52,040 --> 00:07:55,480 imagining? Here like ordering groceries if 153 00:07:55,480 --> 00:07:57,080 I've like I'm looking in the fridge and I could. 154 00:07:57,080 --> 00:08:00,240 Just say walking around. No seamless hands free. 155 00:08:00,240 --> 00:08:02,920 Oh, I need to write that down. Hey, put this in my notes app. 156 00:08:03,040 --> 00:08:06,640 But all of this is things that are again, really Apple has the 157 00:08:06,640 --> 00:08:08,040 advantage here. If they were to release 158 00:08:08,040 --> 00:08:09,640 something like this, that would work because it would just 159 00:08:09,720 --> 00:08:12,480 integrate into my existing devices that I already manage. 160 00:08:12,680 --> 00:08:15,600 So I think that's something that's tough for a start up to 161 00:08:15,600 --> 00:08:18,360 break into as well in this. Space I, I guess I agree with 162 00:08:18,400 --> 00:08:22,760 you that voice in the sort of like ambient moments wandering 163 00:08:22,760 --> 00:08:25,720 around if it could just capture the like 5 seconds of 164 00:08:25,720 --> 00:08:28,440 inspiration so you don't have to open up your phone, find the 165 00:08:28,440 --> 00:08:31,120 right app. You know, that's sort of where 166 00:08:31,120 --> 00:08:34,159 the opportunity, and I guess to bring it back to Maine is sort 167 00:08:34,159 --> 00:08:36,440 of the argument for the device that's just sort of on you. 168 00:08:36,840 --> 00:08:39,600 Whereas it feels like your iPhone is still in the world of 169 00:08:40,440 --> 00:08:42,919 you'd have to pull up the specific app to record it. 170 00:08:42,919 --> 00:08:46,240 And it's not an Apple in particular seems to have set up 171 00:08:46,240 --> 00:08:49,720 this privacy situation where they're not just going to say 172 00:08:49,720 --> 00:08:51,280 all of a sudden Apple's the phone. 173 00:08:51,280 --> 00:08:54,960 That iPhone, we ambiently record you all the time so that you can 174 00:08:54,960 --> 00:08:57,560 just shout out into the world, you know, but maybe maybe 175 00:08:57,560 --> 00:09:01,000 Android or somebody will, you know, get in that game. 176 00:09:01,080 --> 00:09:03,480 But even then, it doesn't feel like you need a secondary device 177 00:09:03,480 --> 00:09:04,840 for that. It feels like that's something 178 00:09:04,840 --> 00:09:07,040 that could just be captured with your existing technology. 179 00:09:07,040 --> 00:09:10,400 So hardware, it's a tough game. There's a lot of incumbents. 180 00:09:10,400 --> 00:09:13,480 They do things well. What is on on voice? 181 00:09:13,520 --> 00:09:15,760 I mean, you know, one of the buzziest Silicon Valley 182 00:09:15,760 --> 00:09:18,600 companies right now. Do you know what I'm going to 183 00:09:18,600 --> 00:09:20,240 say? Granola, which has become 184 00:09:20,720 --> 00:09:22,680 somewhat controversial within the newcomer. 185 00:09:23,880 --> 00:09:26,600 Within the newcomer. I'm sort of a skull, a granola 186 00:09:26,600 --> 00:09:30,280 scold or I just, I just hate the idea that it's recording me 187 00:09:30,680 --> 00:09:33,280 without anybody, without anybody having to tell me. 188 00:09:33,320 --> 00:09:35,360 And you know that's true. As reporters, it's very 189 00:09:35,360 --> 00:09:38,720 important on our jobs to like, let our sources know if we're 190 00:09:38,720 --> 00:09:40,920 recording them or not. So I would prefer to not be 191 00:09:40,920 --> 00:09:44,080 automatically assuming that everyone's recording right. 192 00:09:45,200 --> 00:09:46,760 On the other hand, you know, I sort of want to use it. 193 00:09:46,760 --> 00:09:49,760 So it's one of these like norms where you're resistant, but 194 00:09:49,760 --> 00:09:52,000 you're like, I don't know if the floodgates are going to open. 195 00:09:52,000 --> 00:09:54,560 Maybe I'll just sort of I'll get on board. 196 00:09:54,560 --> 00:09:57,840 You know, you can't you can't stop that level of societal 197 00:09:57,840 --> 00:10:00,680 change. I mean, they claim, you know, my 198 00:10:00,680 --> 00:10:03,680 understanding is, you know, it's not record this, it's very 199 00:10:03,680 --> 00:10:06,960 semantic, but you know, they're they they process it, you know, 200 00:10:07,040 --> 00:10:08,960 well, it's like a zoom recording you. 201 00:10:08,960 --> 00:10:12,160 If you're not actually saving it, but it the computer hears 202 00:10:12,160 --> 00:10:15,440 it, you know it's gone. So if zoom is not recording you, 203 00:10:15,760 --> 00:10:20,400 then similarly this is listening and then it turns it into the 204 00:10:20,400 --> 00:10:22,920 text and then it's gone. That's sort of, I think the 205 00:10:23,160 --> 00:10:26,720 defense, but it is like it's I think it's like a word for word 206 00:10:26,720 --> 00:10:30,280 transcript. So if you want some deniability 207 00:10:30,280 --> 00:10:32,360 on what's said in your conversations going all that 208 00:10:32,360 --> 00:10:35,800 would seem to undermine that. I spent like an hour looking 209 00:10:35,800 --> 00:10:39,080 through like you state laws on how they talk about voice 210 00:10:39,080 --> 00:10:41,800 recording and they're not, you know, they're they're just 211 00:10:41,800 --> 00:10:43,640 varied. You know, it's it's like, and 212 00:10:43,640 --> 00:10:46,520 there's all these sort of stray words that I imagine if you were 213 00:10:46,840 --> 00:10:49,640 fighting about granola, you know, you'd be like, well, sure, 214 00:10:49,640 --> 00:10:52,080 maybe you can say it's not recording, but they're, you 215 00:10:52,080 --> 00:10:55,320 know, they'll say like transmit with granola to me. 216 00:10:55,320 --> 00:10:57,560 In some ways it is argument for text. 217 00:10:58,200 --> 00:11:01,520 So you're having these terrible voice meetings and they're like, 218 00:11:01,560 --> 00:11:04,960 oh, let's make sure you have the text transcript. 219 00:11:05,000 --> 00:11:08,960 And so some of this AI stuff we're constantly teasing out, 220 00:11:09,360 --> 00:11:13,080 you know, is it the voice is inefficient and we need to get a 221 00:11:13,080 --> 00:11:16,720 succinct text record or you know, but I don't, I don't know. 222 00:11:16,720 --> 00:11:19,520 What do you think about that? I think, I mean it feels like 223 00:11:19,520 --> 00:11:22,840 there's a utility for both, but it does seem like everything is 224 00:11:22,840 --> 00:11:25,920 still sort of landing on text, which then would imply the need 225 00:11:25,920 --> 00:11:28,640 for a big screen to manage all of this text for you. 226 00:11:28,760 --> 00:11:31,600 And I think maybe with LLMS, there's a good voice to text 227 00:11:31,600 --> 00:11:34,120 interface is pretty slick at this point as we know from 228 00:11:34,120 --> 00:11:37,520 granola that seems to be a medium where this is working 229 00:11:37,880 --> 00:11:41,040 rather well and there's lots of product innovation in this space 230 00:11:41,040 --> 00:11:44,200 and why, why mess with something that's, you know, working well, 231 00:11:44,360 --> 00:11:46,200 I feel. Like outside of Silicon Valley 232 00:11:46,800 --> 00:11:50,400 there's still a lot of people resisting AI like or I find my 233 00:11:50,400 --> 00:11:54,720 own family members are very reluctant to use AI. 234 00:11:54,920 --> 00:11:58,480 I'm curious, are you seeing that when you're out about among the 235 00:11:59,080 --> 00:12:03,560 the non Silicon Valley inclined? What's your read on there AI 236 00:12:04,640 --> 00:12:07,160 embrace? Among the non-technical masses. 237 00:12:08,240 --> 00:12:10,280 The. Normie masses among the normies. 238 00:12:10,280 --> 00:12:14,320 When I return to the normies, I well, I have AI feel like I'm in 239 00:12:14,320 --> 00:12:18,720 a I'm in a bubble a bit around a lot of my network is still very 240 00:12:18,720 --> 00:12:20,480 in the tech world outside of this job. 241 00:12:20,480 --> 00:12:24,000 But I do think that there is some resistance when I was I was 242 00:12:24,000 --> 00:12:28,160 home last weekend with my family for in Austin, TX, which is a 243 00:12:28,160 --> 00:12:31,480 tech happy city to be fair for the presidents day weekend. 244 00:12:31,480 --> 00:12:34,880 And it seems like there's certainly some skepticism, 245 00:12:34,880 --> 00:12:37,520 especially when just talking with, you know, family friends 246 00:12:37,640 --> 00:12:41,880 coming through for the holiday. People were much more, I guess, 247 00:12:42,120 --> 00:12:46,880 cautious, if not outright skeptical or a little suspicious 248 00:12:46,920 --> 00:12:50,600 of a lot of this in a way that I think we forget about in this 249 00:12:50,600 --> 00:12:52,280 bead. It's made me sort of more 250 00:12:52,280 --> 00:12:55,920 bullish on Google, I'd say in that if people aren't willing to 251 00:12:55,920 --> 00:13:00,440 like really dive in with Chachi BT because they're like just 252 00:13:00,440 --> 00:13:04,160 aren't the messaging around AI is not sort of enticing them to 253 00:13:04,160 --> 00:13:06,640 try it out. They are Googling things and so 254 00:13:07,160 --> 00:13:10,000 Google can just sort of gently say, oh, you're, you have this 255 00:13:10,000 --> 00:13:13,320 sort of question and you sort of slowly learn over time that you 256 00:13:13,320 --> 00:13:16,240 can expand the types of questions you can ask. 257 00:13:16,280 --> 00:13:19,720 It does sort of suggest that if there are these holdouts who are 258 00:13:19,720 --> 00:13:24,320 not willing to like experiment with Chatchi BT that Google can 259 00:13:24,320 --> 00:13:26,160 meet people sort of where they are. 260 00:13:26,160 --> 00:13:29,640 We've also talked about this theme that the, the blank text 261 00:13:29,640 --> 00:13:32,680 box is itself intimidating and that a lot of these AI products 262 00:13:32,680 --> 00:13:34,520 are going to have to be more narrow anyway. 263 00:13:34,520 --> 00:13:36,640 Let's talk about the foundation models, though. 264 00:13:37,000 --> 00:13:38,560 Who? Who had the biggest many 265 00:13:38,560 --> 00:13:42,040 billions on the belief that they will get the world to play with 266 00:13:42,040 --> 00:13:44,000 their text box? Who? 267 00:13:44,040 --> 00:13:45,480 Who? Do you have the biggest news 268 00:13:45,480 --> 00:13:47,360 this week? It was a range, obviously. 269 00:13:47,360 --> 00:13:51,880 Mira Murati, former Open AI CTO. Her startup that has been long 270 00:13:51,880 --> 00:13:54,480 rumored is officially launched. Thinking machines. 271 00:13:54,480 --> 00:13:55,600 What changed? Didn't we? 272 00:13:55,880 --> 00:13:58,200 She just came on the record and gave some friendly interviews 273 00:13:58,200 --> 00:14:01,000 and said, yeah, that thing you'd heard, it's it is happening. 274 00:14:02,440 --> 00:14:04,400 She didn't even say how much money they raised or anything, 275 00:14:04,400 --> 00:14:06,200 right? There's been there's been no 276 00:14:06,200 --> 00:14:09,480 public discussion from Thinking Machines about how much they did 277 00:14:09,480 --> 00:14:12,280 raise Amira. Come on the podcast, you have so 278 00:14:12,280 --> 00:14:15,240 many questions to answer. I, I, I just want, I don't think 279 00:14:15,240 --> 00:14:19,720 I've seen this out there, but you know, she seemingly was 280 00:14:19,720 --> 00:14:22,520 somewhat of a Sam Altman critic, right? 281 00:14:22,760 --> 00:14:29,560 The board oust Sam, puts her in charge of open AI, and then she 282 00:14:29,560 --> 00:14:33,440 rallies to Sam. Sam comes back, and then she 283 00:14:33,440 --> 00:14:36,120 leaves the company. I, I don't know, like, what's I 284 00:14:36,120 --> 00:14:37,880 just feel like that story doesn't make sense. 285 00:14:37,880 --> 00:14:41,400 Like, if the board believes in you, many of many employees, 286 00:14:41,400 --> 00:14:43,920 some of whom are now defecting to your company, believe in you. 287 00:14:44,680 --> 00:14:49,720 Why not just, like be the leader of open AI like I, if you have a 288 00:14:49,720 --> 00:14:51,440 desire to run a company. Yeah. 289 00:14:51,920 --> 00:14:55,360 So you could run it all like. And it really does feel like if 290 00:14:55,360 --> 00:15:00,520 she had stuck with the board, she could have won. 291 00:15:00,560 --> 00:15:04,440 The biggest theory would just be like that she read the employee 292 00:15:04,440 --> 00:15:06,720 base and felt like Sam was going to win. 293 00:15:06,760 --> 00:15:09,880 I mean, the dark theory would be that Sam has some sort of like, 294 00:15:09,880 --> 00:15:13,720 leverage point over her or is a very intimidating figure. 295 00:15:13,720 --> 00:15:18,080 But it doesn't make any sense to me that she rallied all the team 296 00:15:18,720 --> 00:15:21,680 defense, Sam against the board and then leaves and is 297 00:15:21,680 --> 00:15:24,360 undermining our old company by building a competitor. 298 00:15:24,480 --> 00:15:28,200 And she did take around 20 engineers from open AI to come 299 00:15:28,200 --> 00:15:31,520 to thinking machines lap reminiscent of Ilya's company, 300 00:15:31,560 --> 00:15:33,600 you know, safe super intelligence also poached a ton 301 00:15:33,600 --> 00:15:37,200 of engineers. So it seems like the the all the 302 00:15:37,200 --> 00:15:41,240 Co founders have gone off to fund their own thing and, you 303 00:15:41,240 --> 00:15:44,160 know, pull away from what could have just been, you know, 304 00:15:45,000 --> 00:15:48,840 forwarding towards the mission of AGI with the company that has 305 00:15:48,880 --> 00:15:52,280 the most momentum behind it. In a previous episode, I, I 306 00:15:52,280 --> 00:15:54,560 think I went on a rant about this very problem, which is 307 00:15:54,560 --> 00:15:58,080 there's an insane collective action problem where you would 308 00:15:58,080 --> 00:16:01,560 think all the top talent should stay at the same company and 309 00:16:01,560 --> 00:16:05,160 ensure that one of them is super valuable. 310 00:16:05,400 --> 00:16:08,960 But instead every individual actor says, well, maybe I should 311 00:16:09,440 --> 00:16:13,240 have more spiritual control and more equity over the smaller 312 00:16:13,400 --> 00:16:15,480 company that can surely raise a lot of money. 313 00:16:15,760 --> 00:16:19,480 And so then they all sort of like shoot at each other, and 314 00:16:19,480 --> 00:16:22,320 maybe it's going to undermine their collective valuation. 315 00:16:22,520 --> 00:16:25,360 I wonder if it really does come down to just the equity 316 00:16:25,360 --> 00:16:28,000 conversation, not to sound, you know, like, oh, it's about the 317 00:16:28,000 --> 00:16:30,840 funding and not the mission because all of these companies 318 00:16:30,840 --> 00:16:32,600 are very mission driven. A lot of them don't even have 319 00:16:32,600 --> 00:16:34,160 products yet for these new offshoots. 320 00:16:34,520 --> 00:16:37,400 At one point you have to consider that like a good 321 00:16:37,640 --> 00:16:41,800 benefit of leaving is to own your own thing and create your 322 00:16:41,800 --> 00:16:43,120 own thing. We're we're fans of that, you 323 00:16:43,120 --> 00:16:44,640 know, at an independent media company. 324 00:16:44,880 --> 00:16:47,680 So I mean, you are driven to the mission. 325 00:16:47,680 --> 00:16:50,120 To your point, it does seem counterintuitive. 326 00:16:50,640 --> 00:16:53,120 I mean, maybe the mission ultimately would just involve 327 00:16:53,120 --> 00:16:55,520 I'm the one who can do it. You know, at a certain point 328 00:16:55,520 --> 00:16:58,600 when you lose that faith and what's happening, you say I'm 329 00:16:58,600 --> 00:17:00,800 the one that can do it better and that you get the added 330 00:17:00,800 --> 00:17:02,320 benefit of the equity when you do it. 331 00:17:02,360 --> 00:17:06,240 One theme I think we wanted to talk about is AI is making it 332 00:17:06,240 --> 00:17:07,839 easy to build small companies, right? 333 00:17:07,839 --> 00:17:10,240 I think the New York Times, Aaron Griffith, I just had a 334 00:17:10,240 --> 00:17:13,760 piece highlighting that fact. The new Dick measuring in 335 00:17:13,760 --> 00:17:17,000 Silicon Valley is no longer how much money you raise and how 336 00:17:17,000 --> 00:17:20,200 many employees you have, it's how much revenue you can 337 00:17:20,200 --> 00:17:22,880 generate with as few as employees as possible. 338 00:17:22,880 --> 00:17:25,200 I mean I think people mid journeys in the list. 339 00:17:25,400 --> 00:17:27,880 Who else? Oh Cursor for sure is on the top 340 00:17:27,880 --> 00:17:29,680 of the list. And also just the speed in which 341 00:17:29,680 --> 00:17:32,200 they've generated revenue is such an important metric too. 342 00:17:32,400 --> 00:17:34,960 And you can do it with these tiny teams in the age of AI. 343 00:17:35,160 --> 00:17:37,240 Isn't this what the media class has wanted? 344 00:17:37,840 --> 00:17:41,000 Focus on profitability and unit economics? 345 00:17:41,000 --> 00:17:45,000 I mean, now, honestly, I don't know how how ominous was Aaron's 346 00:17:45,000 --> 00:17:47,360 story about this? I mean, there's the implicit 347 00:17:47,360 --> 00:17:52,080 like, oh, AI is really coming for jobs if Silicon Valley is 348 00:17:52,080 --> 00:17:55,360 now like, proudly doing things with as few humans as possible. 349 00:17:55,520 --> 00:17:58,560 To your point there, the first jobs that were seemingly 350 00:17:58,560 --> 00:18:02,680 automated away by AI were software engineers with coding 351 00:18:02,680 --> 00:18:04,480 assistants. This is a use case for this 352 00:18:04,480 --> 00:18:07,320 technology that's becoming very well and anyone can start a 353 00:18:07,320 --> 00:18:09,800 company with a much smaller team because of these tools. 354 00:18:10,000 --> 00:18:14,000 So ironically, a lot of what is the bread and butter of Silicon 355 00:18:14,000 --> 00:18:17,680 Valley is something that is one of the 1st to go in the sort of 356 00:18:17,680 --> 00:18:20,640 automation fears around how much productivity AI can unlock. 357 00:18:20,920 --> 00:18:25,160 Which gives, I don't know, Silicon Valley moral credibility 358 00:18:25,160 --> 00:18:27,480 and high ground. We're willing to kill our own 359 00:18:27,480 --> 00:18:29,160 jobs. We can kill yours too. 360 00:18:29,160 --> 00:18:32,600 You know, I, I mean, I, I am not like a Luddite. 361 00:18:32,600 --> 00:18:35,640 I, I just absolutely do not think we should in any way slow 362 00:18:35,640 --> 00:18:39,440 down technology because of our perceived impact on jobs. 363 00:18:39,440 --> 00:18:42,760 Like we should try to build the best stuff we can as efficiently 364 00:18:42,760 --> 00:18:45,280 as possible. And I think we need to fix that 365 00:18:45,280 --> 00:18:49,320 on the sort of policy side and build a better safety net and 366 00:18:49,320 --> 00:18:52,760 say, oh, maybe someday everybody doesn't have to work. 367 00:18:52,760 --> 00:18:54,760 Obviously that is not the administration we have right 368 00:18:54,760 --> 00:18:58,040 now, which is just like, oh, let's kill every government job 369 00:18:58,040 --> 00:19:00,920 in the world. Like it doesn't make any sense. 370 00:19:00,960 --> 00:19:03,760 But that's another episode. Let's talk about Brock. 371 00:19:03,760 --> 00:19:05,640 I can't believe we've gotten this far in AI Brock. 372 00:19:05,800 --> 00:19:08,400 Brock is honestly there's an argument that everybody else is 373 00:19:08,400 --> 00:19:12,200 trying to make news to over shadow what their new model 374 00:19:12,200 --> 00:19:14,680 release. It's been they're performing 375 00:19:14,680 --> 00:19:15,760 quite well on a lot of benchmarks. 376 00:19:15,800 --> 00:19:18,200 I don't have the latest on the leaderboard. 377 00:19:18,280 --> 00:19:20,920 Well, I saw some debate, you know, that it's so early, but 378 00:19:21,040 --> 00:19:23,320 you know, they have these chatbot arenas, I think. 379 00:19:23,320 --> 00:19:27,200 And but then you don't know if it's apples to apples from from 380 00:19:27,200 --> 00:19:30,440 the Twitter arguments among the AI crew that I saw, it seems 381 00:19:30,440 --> 00:19:34,600 still an open question of whether rock was better on the 382 00:19:34,600 --> 00:19:37,960 same playing field or if it was giving itself some sort of edge. 383 00:19:38,040 --> 00:19:41,480 Part of this is what Grok had some of the most, if not the 384 00:19:41,480 --> 00:19:44,800 most GPU access to train this, which then you don't know if 385 00:19:44,800 --> 00:19:47,720 that's a positive in judging what they've been able to 386 00:19:47,720 --> 00:19:50,600 produce or not, right? It's like it's scary because 387 00:19:50,600 --> 00:19:54,440 they had so many GPUs, but now that it's out it's like it 388 00:19:54,440 --> 00:19:57,920 wasn't so much better. In the era of deep seek, is that 389 00:19:57,920 --> 00:20:00,120 really the flex that it once was? 390 00:20:00,200 --> 00:20:04,080 You can train a model that is. Incrementally better on a much 391 00:20:04,080 --> 00:20:05,520 lower stack. Although of course, you know, 392 00:20:05,520 --> 00:20:08,960 there's some debate around how much of that cost of training 393 00:20:08,960 --> 00:20:11,120 was actually really for the actual training itself. 394 00:20:11,360 --> 00:20:13,720 However, it still seems that it was marginally less. 395 00:20:13,880 --> 00:20:16,400 But still, with Grok, it is impressive. 396 00:20:16,400 --> 00:20:20,160 If they have this massive GPU cluster, maybe less impressive 397 00:20:20,600 --> 00:20:23,120 that the model made a good amount of progress, but more 398 00:20:23,120 --> 00:20:26,160 incremental progress with this amount of power, but it's still 399 00:20:26,160 --> 00:20:28,000 notable that they have this cluster. 400 00:20:28,040 --> 00:20:31,400 It gives them a lot of leverage in this in the space. 401 00:20:31,400 --> 00:20:35,040 And I would not count out Xai and Grok on the talent front, 402 00:20:35,040 --> 00:20:37,680 even if they're not, you know, in the lead with Open AI and 403 00:20:37,680 --> 00:20:40,080 Anthropic at this point in terms of model performance. 404 00:20:40,840 --> 00:20:45,360 And they have access to Twitter and the disinformation miasma 405 00:20:45,360 --> 00:20:48,400 that it is for good and bad. You know, they're going to. 406 00:20:48,720 --> 00:20:52,600 I don't know. I'm Yeah, Elon seems to be going 407 00:20:52,600 --> 00:20:55,240 on the deep end. Just before we got on this, he 408 00:20:55,240 --> 00:20:59,000 was on CPAC. I'm actually in DC right now. 409 00:20:59,080 --> 00:21:01,200 I think they're around here. I'm here for a venture in the 410 00:21:01,200 --> 00:21:05,840 capital. Wonky DC BC conference that 411 00:21:06,040 --> 00:21:07,520 speaking with Kara Fisher anyway. 412 00:21:07,520 --> 00:21:09,440 But Elon was on stage. You seem like under the 413 00:21:09,440 --> 00:21:12,320 influence and like was losing a stream of consciousness. 414 00:21:12,480 --> 00:21:17,600 And at the same time, Grimes is, was she his wife? 415 00:21:17,600 --> 00:21:21,560 No, just his the mother of the mother of some of his children 416 00:21:21,880 --> 00:21:25,120 was tweeting at him that they one of their children is having 417 00:21:25,120 --> 00:21:28,160 a medical emergency and she's been able to contact him. 418 00:21:28,160 --> 00:21:31,040 Like, I don't know. And obviously this week, the 419 00:21:31,040 --> 00:21:35,760 Ukrainian, the abandonment of Ukraine tonally has been 420 00:21:35,760 --> 00:21:38,240 despicable. So I don't know this. 421 00:21:38,520 --> 00:21:44,080 Obviously I'm no Trump and no Elon Trump fan, but just feel 422 00:21:44,080 --> 00:21:47,920 like this could have been the week things jump the shark and 423 00:21:47,920 --> 00:21:50,160 that they will hopefully lose the country. 424 00:21:50,160 --> 00:21:54,480 But yeah, we'll, we'll see. With that wonderful, dark, 425 00:21:55,200 --> 00:21:58,760 calamitous note about the state of our country, let's jump into 426 00:21:59,040 --> 00:22:02,160 a really major funding round that feels more relevant than 427 00:22:02,160 --> 00:22:04,240 ever. Your carrots, No politics. 428 00:22:04,240 --> 00:22:05,560 We're talking about business again. 429 00:22:05,560 --> 00:22:08,200 What's the funding round? Oh, the funding round is 430 00:22:08,200 --> 00:22:10,800 Saronic, which is based out of Austin, TX. 431 00:22:10,800 --> 00:22:15,720 And yes, it is kind of on theme, autonomous boats, so drone boats 432 00:22:15,720 --> 00:22:17,920 for the military. For warfare. 433 00:22:17,920 --> 00:22:20,880 OK, good. So Elon and Trump will be well. 434 00:22:20,880 --> 00:22:25,160 Armed, uh, raked in a good 600 million in Series C funding in a 435 00:22:25,160 --> 00:22:28,560 round led by Elad Gill. Some other participants were 436 00:22:28,560 --> 00:22:31,680 general catalysts had not backed this company yet, but they did 437 00:22:31,680 --> 00:22:33,960 back it for the first time in this round. 438 00:22:33,960 --> 00:22:36,760 And then some existing investors, including a 16Z 439 00:22:36,920 --> 00:22:40,280 caffeinated capital and 8 VC participated in this round, 440 00:22:40,400 --> 00:22:43,160 which really feels like at this point kind of a who's who of who 441 00:22:43,160 --> 00:22:46,600 backs, you know, American dynamism style defense tech from 442 00:22:46,600 --> 00:22:48,520 the mega funds. We have Elad who is an early 443 00:22:48,520 --> 00:22:51,040 investor in Anderl. We have a 16Z investor in 444 00:22:51,040 --> 00:22:55,160 Anderl, 8 VC investor across all of these things, Joe Lonsdale 445 00:22:55,160 --> 00:22:58,120 and then General Catalyst with their global resilience practice 446 00:22:58,120 --> 00:23:02,240 and focus led by Paul Kwan. It's an A list ring of investors 447 00:23:02,240 --> 00:23:05,480 for a massive funding round for autonomous boats. 448 00:23:05,520 --> 00:23:11,200 We support defense tech investing require some sorry I'm 449 00:23:11,200 --> 00:23:13,000 in a dark mood about the American government at the 450 00:23:13,000 --> 00:23:16,440 moment, but previous before today I think I've supported 451 00:23:16,440 --> 00:23:19,480 defense tech investing. But if we're, I don't know if 452 00:23:19,480 --> 00:23:22,880 they're the boats are going to go to war on behalf of Russia, 453 00:23:22,880 --> 00:23:26,840 that will be deeply troubling. Another honorable mention, 454 00:23:26,840 --> 00:23:32,160 funding round and General Catalyst deal together AI which 455 00:23:32,160 --> 00:23:35,280 helps startups fine tune and train models through rented GPU 456 00:23:35,280 --> 00:23:40,120 space raised 305,000,000 at a $3.3 billion valuation. 457 00:23:40,280 --> 00:23:43,520 It was Co led by General Catalyst and Prosperity 7 458 00:23:43,520 --> 00:23:46,160 Ventures, which I believe is out of Saudi Arabia, but then 459 00:23:46,440 --> 00:23:50,040 Salesforce Ventures, NVIDIA, Kleiner Perkins Emergence and 460 00:23:50,040 --> 00:23:52,240 Lux also participated in the round. 461 00:23:52,240 --> 00:23:56,000 So big up round in short. And coding starting up startup 462 00:23:56,000 --> 00:23:58,440 Codium is apparently raising. Anyway, it's boom time. 463 00:23:58,680 --> 00:24:02,240 It's still, I mean AI not we're not getting the same, you know, 464 00:24:03,240 --> 00:24:05,440 foundation model investments that we once did. 465 00:24:05,440 --> 00:24:08,880 But applications and infrastructure clearly are gonna 466 00:24:08,880 --> 00:24:12,640 go strong this year. So and a lot of the tier one BCS 467 00:24:12,640 --> 00:24:14,520 have a lot of appetite for it, have not slowed down. 468 00:24:15,000 --> 00:24:17,000 Cool. Well, I think that's our show. 469 00:24:17,280 --> 00:24:19,520 We'll see you next week. See you next week.