1 00:00:00,000 --> 00:00:02,000 [Miles] The S26 batch just started. 2 00:00:02,250 --> 00:00:06,750 [Miles] Decisions went out June 5th, roughly 1% of 10,000 plus applicants. 3 00:00:06,999 --> 00:00:14,599 [Miles] And right now, those founders are sitting in San Francisco staring at a blank calendar trying to figure out what to do before office hours even exist. 4 00:00:14,949 --> 00:00:19,500 [Grant] The gap between getting accepted and understanding what you're actually supposed to be doing, 5 00:00:19,800 --> 00:00:21,600 [Grant] that's the part nobody talks about. 6 00:00:21,750 --> 00:00:22,250 [Miles] Right, 7 00:00:22,300 --> 00:00:25,050 [Miles] so today we're dropping into that exact moment. 8 00:00:25,199 --> 00:00:27,750 [Miles] We've got a founder walking us through the first hours, 9 00:00:27,899 --> 00:00:28,649 [Miles] not the pitch. 10 00:00:28,692 --> 00:00:30,692 [Miles] It's not the narrative they'll tell at Demo Day, 11 00:00:30,841 --> 00:00:32,142 [Miles] the unedited version. 12 00:00:32,341 --> 00:00:39,142 [Grant] And what I want to pull at because I've seen this pattern is how much got decided before any YC structure was there to guide them, 13 00:00:39,292 --> 00:00:40,542 [Grant] who to trust in the room, 14 00:00:40,692 --> 00:00:43,292 [Grant] which signals to act on that all get set early. 15 00:00:43,591 --> 00:00:44,492 [Miles] Yeah, yeah, 16 00:00:44,542 --> 00:00:46,492 [Miles] and we're also going to get into the RFS, 17 00:00:46,542 --> 00:00:48,742 [Miles] the Summer 2026 Request for Startups, 18 00:00:48,891 --> 00:00:53,042 [Miles] which the VC corner covered and is a different animal than what we've seen before. 19 00:00:53,391 --> 00:00:54,591 [Miles] 15 categories. 20 00:00:55,091 --> 00:00:57,192 [Miles] Eight of them require hardware or capital. 21 00:00:57,491 --> 00:00:59,991 [Grant] For what's supposed to be a software accelerator, 22 00:01:00,491 --> 00:01:01,491 [Grant] that's a real statement. 23 00:01:01,792 --> 00:01:03,942 [Miles] I've seen this play out a hundred times. 24 00:01:04,141 --> 00:01:06,042 [Miles] Founders read the RFS as a checklist. 25 00:01:06,491 --> 00:01:11,042 [Miles] The ones who read it as a conviction map are the ones who show up to office hours with something to say. 26 00:01:11,241 --> 00:01:14,991 [Grant] The math on that distinction tends to be brutal and pretty fast. 27 00:01:15,391 --> 00:01:21,391 [Miles] And we close on a founder who's still inside S26 right now, sitting on a call that hasn't happened yet. 28 00:01:21,236 --> 00:01:21,536 [Miles] And yet, 29 00:01:21,786 --> 00:01:23,686 [Miles] build the chip or build the scheduler, 30 00:01:23,986 --> 00:01:25,186 [Miles] no answer exists. 31 00:01:25,386 --> 00:01:26,536 [Miles] Batch just started. 32 00:01:26,936 --> 00:01:27,436 [Grant] So, 33 00:01:27,835 --> 00:01:28,436 [Grant] code open, 34 00:01:28,686 --> 00:01:29,286 [Grant] found a voice, 35 00:01:29,536 --> 00:01:29,886 [Grant] let's go. 36 00:01:30,611 --> 00:01:34,561 [Miles] You know, what nobody told me is that the room is quieter than you'd expect. 37 00:01:34,762 --> 00:01:35,311 [Grant] Yeah. 38 00:01:35,411 --> 00:01:38,961 [Miles] You've got 200 founders roughly all sitting there. 39 00:01:39,012 --> 00:01:40,111 [Miles] And I remember thinking, 40 00:01:40,262 --> 00:01:41,611 [Miles] I've flown here, 41 00:01:41,711 --> 00:01:42,861 [Miles] I've quit my job, 42 00:01:42,961 --> 00:01:47,211 [Miles] I've told everyone I know that this is the thing and the partner just starts talking. 43 00:01:47,461 --> 00:01:48,311 [Miles] No countdown, 44 00:01:48,561 --> 00:01:52,211 [Miles] no fanfare. And I'm looking around at all these people thinking, 45 00:01:52,361 --> 00:01:55,111 [Miles] everyone in here has a version of that story. 46 00:01:55,512 --> 00:01:57,561 [Grant] Did it feel like validation or pressure? 47 00:01:58,004 --> 00:01:59,754 [Miles] Both, like simultaneously. 48 00:02:00,404 --> 00:02:02,154 [Miles] You want it to feel like an arrival, 49 00:02:02,254 --> 00:02:03,404 [Miles] and it does a little, 50 00:02:03,454 --> 00:02:06,653 [Miles] but the second they start talking about what the batch actually demands, 51 00:02:06,804 --> 00:02:09,503 [Miles] you realize the acceptance letter was the easy part. 52 00:02:10,454 --> 00:02:11,304 [Grant] That tracks. 53 00:02:11,653 --> 00:02:15,103 [Miles] I talked to one S26 founder who described it as, 54 00:02:15,153 --> 00:02:22,204 [Miles] she said it felt like the first day of a job where everyone else already seems to know what they're doing and you're the only one who missed the orientation. 55 00:02:23,503 --> 00:02:24,054 [Grant] Yeah. 56 00:02:24,153 --> 00:02:26,504 [Grant] I mean, that psychological thing is real. 57 00:02:26,515 --> 00:02:28,865 [Grant] You spend months just trying to get in. 58 00:02:29,015 --> 00:02:29,615 [Grant] The application, 59 00:02:29,865 --> 00:02:30,415 [Grant] the interview. 60 00:02:30,915 --> 00:02:35,915 [Grant] Capwave published data saying YC accepts somewhere between 1.5 and 2% of applicants, 61 00:02:36,365 --> 00:02:37,965 [Grant] lower acceptance rate than Harvard. 62 00:02:38,215 --> 00:02:38,816 [Miles] Right, 63 00:02:38,965 --> 00:02:40,115 [Miles] so you've cleared that, 64 00:02:40,215 --> 00:02:42,465 [Miles] and then you walk in and the room just continues. 65 00:02:42,665 --> 00:02:43,665 [Miles] The batch starts. 66 00:02:44,015 --> 00:02:46,115 [Miles] YC's site is pretty plain about it. 67 00:02:46,415 --> 00:02:49,015 [Miles] Three-day in-person kickoff, you meet your group partners, 68 00:02:49,165 --> 00:02:49,966 [Miles] you meet your cohort, 69 00:02:50,115 --> 00:02:50,566 [Miles] that's it. 70 00:02:50,615 --> 00:02:51,415 [Miles] That's the structure. 71 00:02:51,615 --> 00:02:55,415 [Grant] And the gap between what you imagine and what that actually looks like. 72 00:02:55,187 --> 00:02:57,287 [Grant] Like, I think that's the thing that catches people. 73 00:02:57,537 --> 00:03:02,837 [Miles] Because the fantasy version has some kind of ceremony to it, like someone hands you the playbook. 74 00:03:02,937 --> 00:03:10,137 [Miles] But the VCcorner site's piece on the Summer 2026 RFS, they make this point that the batch is more technically ambitious this cycle, 75 00:03:10,237 --> 00:03:15,237 [Miles] more specific about where YC thinks the next generation of companies is actually coming from. 76 00:03:15,287 --> 00:03:16,487 [Miles] So you're not just starting, 77 00:03:16,587 --> 00:03:22,037 [Miles] you're starting inside a room full of people who were selected because YC is placing a directional bet. 78 00:03:21,907 --> 00:03:22,807 [Miles] that right now. 79 00:03:22,857 --> 00:03:26,207 [Grant] And most of those founders don't fully know what that bet is yet, 80 00:03:26,307 --> 00:03:26,707 [Grant] day one. 81 00:03:26,807 --> 00:03:29,057 [Miles] Day one. You're in that room, 82 00:03:29,107 --> 00:03:30,257 [Miles] the partner's talking, 83 00:03:30,457 --> 00:03:32,957 [Miles] and you've got maybe three months to Demo Day. 84 00:03:33,607 --> 00:03:34,907 [Miles] Founders I've spoken to, 85 00:03:34,957 --> 00:03:41,057 [Miles] they describe the 11 weeks leading up to Demo Day as the most productive stretch of their lives, and I believe that. 86 00:03:41,207 --> 00:03:42,457 [Miles] But Hour One, 87 00:03:42,957 --> 00:03:45,157 [Miles] Hour One is disorienting. 88 00:03:45,407 --> 00:03:48,457 [Grant] You either figure out how to use that disorientation. 89 00:03:48,663 --> 00:03:50,013 [Grant] or it uses you. 90 00:03:50,463 --> 00:03:54,413 [Miles] So what does figuring it out actually look like when you're inside it, 91 00:03:54,663 --> 00:03:56,463 [Miles] before office hours even start, 92 00:03:56,614 --> 00:03:59,013 [Miles] before you've had a single conversation with your partner? 93 00:03:59,263 --> 00:04:03,513 [Miles] What gets decided in those first 72 hours, and who's making those calls? 94 00:04:04,627 --> 00:04:06,727 [Miles] So the first real decision, 95 00:04:06,927 --> 00:04:08,227 [Miles] I want to be specific here, 96 00:04:08,327 --> 00:04:10,927 [Miles] not the one founders rehearsed for their YC interview, 97 00:04:11,427 --> 00:04:14,027 [Miles] the one that walked in uninvited on day two. 98 00:04:14,277 --> 00:04:16,327 [Grant] And this is where it gets murky fast, 99 00:04:16,728 --> 00:04:18,827 [Grant] because when you talk to these S26 founders now, 100 00:04:19,127 --> 00:04:20,427 [Grant] everything sounds intentional. 101 00:04:20,527 --> 00:04:21,877 [Grant] It sounds like a plan. 102 00:04:22,228 --> 00:04:24,377 [Miles] Right. Hindsight is a hell of an editor. 103 00:04:24,577 --> 00:04:26,228 [Grant] So where were they actually working with? 104 00:04:26,228 --> 00:04:27,377 [Miles] Almost nothing. 105 00:04:27,527 --> 00:04:30,277 [Miles] I mean, they've been in the room maybe 48 hours. 106 00:04:30,675 --> 00:04:32,026 [Miles] Office hours haven't started, 107 00:04:32,175 --> 00:04:33,875 [Miles] they don't have a partner relationship yet, 108 00:04:34,125 --> 00:04:35,325 [Miles] they're surrounded by what? 109 00:04:35,425 --> 00:04:40,125 [Miles] A couple hundred other companies, all of whom look like they already know exactly what they're doing. 110 00:04:40,225 --> 00:04:41,675 [Grant] That number matters. 111 00:04:42,175 --> 00:04:44,075 [Grant] Young Combinator's S26 batch, 112 00:04:44,475 --> 00:04:47,276 [Grant] 250 to 300 companies per published numbers. 113 00:04:47,475 --> 00:04:49,975 [Grant] You're one of them. And from inside that room, 114 00:04:50,276 --> 00:04:53,225 [Grant] the other 249 all seem further along than you. 115 00:04:53,425 --> 00:04:57,225 [Miles] I've talked to founders across multiple batches who describe that exact 116 00:04:57,172 --> 00:04:57,872 [Miles] exact feeling. 117 00:04:58,422 --> 00:05:00,122 [Miles] Everyone's nodding like they have traction. 118 00:05:00,422 --> 00:05:04,222 [Miles] Nobody's admitting they have no idea what week one actually requires. 119 00:05:04,522 --> 00:05:05,572 [Grant] So what did they decide? 120 00:05:05,822 --> 00:05:07,672 [Miles] The two founders I spoke to, 121 00:05:07,772 --> 00:05:08,622 [Miles] different companies, 122 00:05:08,772 --> 00:05:09,522 [Miles] different categories, 123 00:05:09,872 --> 00:05:11,922 [Miles] both made essentially the same call. 124 00:05:12,122 --> 00:05:16,072 [Miles] They picked who to talk to before they had any framework for who mattered. 125 00:05:16,322 --> 00:05:17,522 [Grant] That sounds small. 126 00:05:17,722 --> 00:05:18,522 [Miles] It's not. 127 00:05:18,822 --> 00:05:21,672 [Miles] Think about what information costs you in that environment. 128 00:05:21,922 --> 00:05:23,122 [Miles] You've got maybe a hundred 129 00:05:22,899 --> 00:05:25,349 [Miles] A hundred conversations you could have in week one, 130 00:05:25,649 --> 00:05:30,299 [Miles] the ones you choose shape what you think is true about your market for the next 12 weeks. 131 00:05:30,649 --> 00:05:32,250 [Grant] And the founders knew that at the time? 132 00:05:32,549 --> 00:05:32,899 [Miles] No, 133 00:05:33,049 --> 00:05:34,899 [Miles] that's the thing I kept pushing on. 134 00:05:35,250 --> 00:05:38,250 [Miles] I'd ask what data did you actually have when you made that call, 135 00:05:38,599 --> 00:05:44,149 [Miles] and the answer was usually some version of a hunch, a conversation from the week before the batch, 136 00:05:44,199 --> 00:05:45,199 [Miles] maybe a Slack message. 137 00:05:45,349 --> 00:05:48,949 [Grant] So they were filling in missing information with whatever was closest. 138 00:05:49,235 --> 00:05:52,735 [Miles] With whoever was in the room—which is a very specific sample. 139 00:05:53,036 --> 00:05:54,586 [Grant] That's the port I want to sit on, 140 00:05:54,985 --> 00:05:57,586 [Grant] because what they're describing—deciding who to trust, 141 00:05:57,685 --> 00:05:58,336 [Grant] who to ignore, 142 00:05:58,536 --> 00:06:01,685 [Grant] which signal to act on-that's not a day two conversation. 143 00:06:02,135 --> 00:06:02,836 [Grant] That's the batch. 144 00:06:03,135 --> 00:06:06,086 [Grant] That's twelve weeks of compounding off of first instinct. 145 00:06:06,435 --> 00:06:12,235 [Miles] And what's wild is neither of them flagged it as a decision at the time-one of them literally said, 146 00:06:12,336 --> 00:06:13,935 [Miles] I wasn't deciding anything, 147 00:06:14,135 --> 00:06:15,635 [Miles] I was just eating lunch with someone. 148 00:06:16,019 --> 00:06:16,519 [Grant] Yeah, 149 00:06:16,569 --> 00:06:18,219 [Grant] that's how the expensive ones always start. 150 00:06:18,419 --> 00:06:20,419 [Miles] The YC program is structured. 151 00:06:20,870 --> 00:06:22,469 [Miles] Groups of six to ten companies, 152 00:06:22,719 --> 00:06:24,069 [Miles] a dedicated general partner, 153 00:06:24,469 --> 00:06:25,370 [Miles] weekly office hours. 154 00:06:25,669 --> 00:06:28,120 [Miles] But that structure doesn't kick in immediately. 155 00:06:28,620 --> 00:06:29,719 [Miles] There's a gap. 156 00:06:30,120 --> 00:06:31,319 [Grant] And what happens in the gap? 157 00:06:31,569 --> 00:06:40,370 [Miles] That's exactly what the partner sees that the founders don't. One vantage point on a hundred companies making the same call without knowing they're making it. 158 00:06:41,235 --> 00:06:45,286 [Miles] So what a partner sees from week one and what a founder sees, 159 00:06:45,385 --> 00:06:47,635 [Miles] those are two completely different movies. 160 00:06:48,036 --> 00:06:49,036 [Grant] That's the thing, right? 161 00:06:49,286 --> 00:06:51,485 [Grant] Each section is six to ten companies. 162 00:06:51,735 --> 00:06:54,086 [Grant] Your partner is watching all of them simultaneously. 163 00:06:54,536 --> 00:06:56,235 [Grant] You're watching only yourself. 164 00:06:56,536 --> 00:06:59,485 [Miles] According to YC Combinator's own program structure. 165 00:07:00,101 --> 00:07:04,651 [Miles] Each general partner hand selects a small group every batch and meets with them weekly, 166 00:07:05,101 --> 00:07:07,401 [Miles] so the information asymmetry is built in. 167 00:07:07,851 --> 00:07:10,651 [Miles] The partner has the comparison set; you don't. 168 00:07:10,951 --> 00:07:13,751 [Grant] And the pattern they are watching for in week one isn't momentum, 169 00:07:14,201 --> 00:07:18,401 [Grant] is whether founders can tell the difference between being busy and actually moving. 170 00:07:18,851 --> 00:07:21,801 [Miles] I've seen this play out a hundred times in other contexts: 171 00:07:22,301 --> 00:07:23,401 [Miles] someone fills the calendar, 172 00:07:23,651 --> 00:07:24,601 [Miles] fires off emails, 173 00:07:24,701 --> 00:07:25,951 [Miles] takes every intro call. 174 00:07:25,817 --> 00:07:27,067 [Miles] All three weeks in, 175 00:07:27,467 --> 00:07:28,567 [Miles] nothing has changed. 176 00:07:29,017 --> 00:07:30,967 [Grant] Confusing motion for progress. 177 00:07:31,067 --> 00:07:31,717 [Grant] That's the one. 178 00:07:31,817 --> 00:07:32,717 [Grant] And the thing is, 179 00:07:32,767 --> 00:07:34,767 [Grant] it feels identical from the inside. 180 00:07:35,017 --> 00:07:37,317 [Miles] So how does a partner catch it early? 181 00:07:37,667 --> 00:07:40,467 [Grant] The question they're apparently asking at office hours, 182 00:07:40,817 --> 00:07:41,917 [Grant] YC's own framing, 183 00:07:42,267 --> 00:07:42,867 [Grant] is basically, 184 00:07:43,117 --> 00:07:45,867 [Grant] what's actually holding you back from moving faster? 185 00:07:45,967 --> 00:07:47,667 [Grant] Not what did you do this week? 186 00:07:47,767 --> 00:07:48,617 [Grant] What is the obstacle? 187 00:07:48,817 --> 00:07:51,067 [Miles] Which is a very different question. 188 00:07:51,417 --> 00:07:52,417 [Grant] Completely different. 189 00:07:52,517 --> 00:07:54,817 [Grant] And it separates the founders who know the... 190 00:07:54,521 --> 00:07:58,271 [Grant] Know their own bottleneck from the ones who are just narrating their schedule. 191 00:07:58,621 --> 00:08:00,121 [Miles] Narrating their schedule. 192 00:08:00,271 --> 00:08:00,471 [Miles] Yeah. 193 00:08:00,771 --> 00:08:02,721 [Grant] Now flip it to the RFS side. 194 00:08:03,071 --> 00:08:07,571 [Grant] The VC Corner published a breakdown of all 15 S26 categories, 195 00:08:08,121 --> 00:08:09,721 [Grant] AI-Native Service Companies, 196 00:08:10,071 --> 00:08:10,821 [Grant] Company Brain, 197 00:08:11,221 --> 00:08:13,021 [Grant] Inference Chips for Agent Workflows, 198 00:08:13,471 --> 00:08:14,571 [Grant] Counter-Swarm Defense, 199 00:08:14,871 --> 00:08:17,371 [Grant] Electronics and Space. It spans AI, 200 00:08:17,621 --> 00:08:18,121 [Grant] hardware, 201 00:08:18,271 --> 00:08:18,721 [Grant] Defense, 202 00:08:18,871 --> 00:08:19,221 [Grant] agriculture, 203 00:08:19,371 --> 00:08:19,721 [Grant] space. 204 00:08:19,971 --> 00:08:23,421 [Miles] 15 categories and the partners who wrote it are now 205 00:08:23,225 --> 00:08:25,525 [Miles] that are now in the room with S26 founders, 206 00:08:25,825 --> 00:08:31,275 [Miles] they can see in real time who's building towards those gaps and who drifted in with something adjacent. 207 00:08:32,025 --> 00:08:33,725 [Grant] Adjacent is a polite word. 208 00:08:33,825 --> 00:08:37,975 [Miles] It is. And YC Combinator's own RFS page is explicit. 209 00:08:38,175 --> 00:08:39,725 [Miles] This isn't a trend report. 210 00:08:39,825 --> 00:08:41,625 [Miles] These are gaps that are now solvable. 211 00:08:41,875 --> 00:08:43,475 [Miles] The framing is operational, 212 00:08:43,725 --> 00:08:44,825 [Miles] not aspirational. 213 00:08:45,075 --> 00:08:48,525 [Grant] Which means a partner in week one already has a mental model. 214 00:08:48,675 --> 00:08:51,225 [Grant] Is this founder building towards a real gap? 215 00:08:51,449 --> 00:08:55,549 [Grant] Or are they building towards what felt exciting in their own heads six months ago? 216 00:08:55,949 --> 00:08:57,149 [Miles] Narrating their schedule, 217 00:08:57,299 --> 00:08:57,449 [Miles] yeah. 218 00:08:57,749 --> 00:09:01,899 [Miles] That gap between the RFS signal and what's actually being built, 219 00:09:02,049 --> 00:09:07,999 [Miles] that's probably the most useful data the partner has before any office hours conversation even starts. 220 00:09:08,449 --> 00:09:08,699 [Grant] Right, 221 00:09:08,849 --> 00:09:10,699 [Grant] and it raises the obvious question: 222 00:09:11,149 --> 00:09:16,649 [Grant] what does the RFS actually tell you if you read it the way a partner reads it-not as a check list, 223 00:09:16,799 --> 00:09:18,049 [Grant] but as a set of conditions? 224 00:09:18,393 --> 00:09:20,393 [Miles] Which is exactly where we're going, 225 00:09:20,543 --> 00:09:22,843 [Miles] because you've been treating it like term sheet language, 226 00:09:23,043 --> 00:09:24,643 [Miles] and it changes what you see in it. 227 00:09:25,401 --> 00:09:30,601 [Miles] So the VCCorner piece on the S26 RFS, I pulled it apart like a term sheet. 228 00:09:30,851 --> 00:09:31,801 [Miles] 15 categories, 229 00:09:32,151 --> 00:09:33,701 [Miles] 8 of them require capital, 230 00:09:33,951 --> 00:09:34,401 [Miles] hardware, 231 00:09:34,651 --> 00:09:35,501 [Miles] or actual atoms. 232 00:09:35,801 --> 00:09:36,101 [Miles] That's 233 00:09:36,101 --> 00:09:36,301 [Grant] Wow. 234 00:09:36,301 --> 00:09:39,251 [Miles] the biggest physical world emphasis in RFS history. 235 00:09:39,501 --> 00:09:43,201 [Grant] And most founders walking into Week One probably skimmed it. 236 00:09:43,201 --> 00:09:43,601 [Miles] Right, 237 00:09:43,751 --> 00:09:49,601 [Miles] they skimmed it, but the partners didn't write it to be skimmed. Every category has a named partner attached. 238 00:09:49,901 --> 00:09:51,701 [Miles] Garry Tan on agriculture robotics. 239 00:09:51,833 --> 00:09:51,983 [Miles] X, 240 00:09:52,033 --> 00:09:54,083 [Miles] Diana Hu shows up three times, 241 00:09:54,483 --> 00:09:55,183 [Miles] inference chips, 242 00:09:55,433 --> 00:09:56,833 [Miles] semiconductor supply chains, 243 00:09:57,233 --> 00:09:58,683 [Miles] AIOS for companies. 244 00:09:58,833 --> 00:10:00,083 [Miles] That's not an accident, 245 00:10:00,233 --> 00:10:00,833 [Miles] that's a bet. 246 00:10:01,033 --> 00:10:03,583 [Grant] Three entries from one person is a statement. 247 00:10:03,683 --> 00:10:05,083 [Miles] It's a conviction map, 248 00:10:05,233 --> 00:10:07,483 [Miles] and when a partner sits down with you in week one, 249 00:10:07,783 --> 00:10:11,233 [Miles] they already know whose name is on the category closest to what you're building. 250 00:10:11,733 --> 00:10:12,433 [Miles] Okay, but, 251 00:10:12,633 --> 00:10:13,983 [Miles] and I push back on this a little, 252 00:10:14,233 --> 00:10:17,883 [Miles] most founders in S26 didn't get in because they perfectly matched a category. 253 00:10:17,657 --> 00:10:18,257 [Miles] to category. 254 00:10:18,407 --> 00:10:20,457 [Miles] They got in because they had something real. 255 00:10:20,707 --> 00:10:23,607 [Miles] So where does the RFS actually land for them in week one? 256 00:10:23,707 --> 00:10:24,757 [Grant] That's the tension. 257 00:10:25,057 --> 00:10:29,457 [Grant] Reading it as a checklist is the wrong move. Am I AI native services? 258 00:10:29,607 --> 00:10:30,857 [Grant] Am I company brain? 259 00:10:31,007 --> 00:10:32,057 [Grant] That's not the question. 260 00:10:32,207 --> 00:10:33,007 [Miles] What's the question? 261 00:10:33,207 --> 00:10:37,857 [Grant] Well, the question is, what problem does Diana Hu actually lose sleep over? 262 00:10:38,107 --> 00:10:44,907 [Grant] Because company brain as a category is YC saying out loud that every company's most critical knowledge. 263 00:10:44,921 --> 00:10:58,321 [Grant] How to improve a special pricing escalation How to handle a refund escalation It's scattered across Slack histories and the heads of people who might quit tomorrow That's the gap The category name is just the label Always 264 00:10:58,321 --> 00:11:00,321 [Miles] So you're reading underneath the label. 265 00:11:01,121 --> 00:11:09,371 [Grant] I've done this in deal making for 20 years The document tells you what they want to fund The conversation tells you what they're afraid won't get built 266 00:11:09,621 --> 00:11:13,121 [Miles] And if you're a founder who doesn't map cleanly to any of the 15? 267 00:11:13,497 --> 00:11:15,397 [Grant] YC is explicit about this. 268 00:11:15,597 --> 00:11:19,647 [Grant] They fund far more companies outside the RFS than inside it. 269 00:11:19,847 --> 00:11:23,297 [Grant] But the founders who understand the thesis behind the categories, 270 00:11:23,447 --> 00:11:25,047 [Grant] not just the category names, 271 00:11:25,447 --> 00:11:30,897 [Grant] those are the ones who walk into office hours and sound like they've been thinking at the same altitude as the partner. 272 00:11:30,997 --> 00:11:39,197 [Miles] There's a version of this where a founder in S26 right now is sitting on something genuinely in the inference chips for agent workflows lane. 273 00:11:39,225 --> 00:11:44,025 [Miles] I mean, current GPUs hitting 30-40% utilization on agentic workloads. 274 00:11:44,325 --> 00:11:46,475 [Miles] That's the gap Diana who is describing, 275 00:11:46,625 --> 00:11:50,075 [Miles] and they don't know it yet because they haven't read closely enough. 276 00:11:50,475 --> 00:11:53,125 [Grant] Or they know it and are not saying it clearly. 277 00:11:53,425 --> 00:11:57,425 [Miles] Which brings us to one founder who's in that exact position right now. 278 00:11:57,825 --> 00:11:58,275 [Grant] Yeah, 279 00:11:58,325 --> 00:11:59,575 [Grant] and they haven't decided yet. 280 00:11:59,875 --> 00:12:01,775 [Miles] That decision is live. 281 00:12:01,875 --> 00:12:03,025 [Miles] We'll get into it. 282 00:12:03,801 --> 00:12:05,051 [Miles] So this founder, 283 00:12:05,151 --> 00:12:06,751 [Miles] the one in the inference chips lane, 284 00:12:06,851 --> 00:12:09,151 [Miles] they've got a real fork in front of them right now. 285 00:12:09,251 --> 00:12:10,751 [Grant] Yeah, walk me through it. 286 00:12:10,801 --> 00:12:21,101 [Miles] Okay, so according to the VC corner's breakdown of the S26 RFS, current GPUs are hitting 30 to 40% of peak utilization on agentic workloads. 287 00:12:21,251 --> 00:12:24,401 [Miles] The work is bursty, context switching is brutal, 288 00:12:24,551 --> 00:12:29,651 [Miles] and Diana Hu is explicitly asking for chips designed around the agent. 289 00:12:29,561 --> 00:12:29,911 [Miles] Agent loop. 290 00:12:30,061 --> 00:12:31,611 [Miles] Fast context switching. 291 00:12:31,761 --> 00:12:35,261 [Miles] KV caches that persist across an entire execution graph. 292 00:12:35,511 --> 00:12:38,261 [Miles] This founder is sitting right in the middle of that gap. 293 00:12:38,261 --> 00:12:39,411 [Grant] So what's the fork? 294 00:12:39,511 --> 00:12:44,211 [Miles] Do you build the chip or do you build the systems that makes existing chips more efficient? 295 00:12:44,461 --> 00:12:46,161 [Miles] Two very different companies. 296 00:12:46,461 --> 00:12:48,261 [Grant] Oh, that's a brutal one. 297 00:12:48,411 --> 00:12:51,361 [Grant] That's like, that's a two-year difference in timeline. 298 00:12:51,711 --> 00:12:54,861 [Miles] Minimum. And the batch is, what, barely a weekend. 299 00:12:55,161 --> 00:12:55,611 [Grant] Yeah, 300 00:12:55,711 --> 00:12:56,361 [Grant] no pressure. 301 00:12:56,729 --> 00:12:57,729 [Miles] And I think, look, 302 00:12:57,929 --> 00:12:59,279 [Miles] if I'm watching this from the outside, 303 00:12:59,529 --> 00:13:05,779 [Miles] the thing I'd be clocking over the next two weeks is whether their first technical hire is a chip architect or a systems engineer. 304 00:13:05,979 --> 00:13:06,729 [Miles] That's the tell. 305 00:13:06,929 --> 00:13:07,529 [Grant] Mmm, 306 00:13:07,779 --> 00:13:10,129 [Grant] because one of those hires closes the door. 307 00:13:10,579 --> 00:13:11,279 [Miles] Completely. 308 00:13:11,429 --> 00:13:12,779 [Miles] You hire the chip architect, 309 00:13:12,979 --> 00:13:15,829 [Miles] you're committing to a three- to five-year capital story. 310 00:13:16,029 --> 00:13:17,579 [Miles] You hire the systems engineer, 311 00:13:17,729 --> 00:13:22,779 [Miles] you're betting you can extract enough performance from existing silicon that customers will pay for it now. 312 00:13:23,097 --> 00:13:24,347 [Grant] From a number standpoint, 313 00:13:24,697 --> 00:13:25,547 [Grant] the way I look at it, 314 00:13:25,697 --> 00:13:29,897 [Grant] can you get a design partner to pay for a proof of concept in the next 60 days? 315 00:13:30,047 --> 00:13:31,097 [Grant] If the answer is no, 316 00:13:31,547 --> 00:13:33,447 [Grant] you probably don't have the chip play yet. 317 00:13:33,497 --> 00:13:34,347 [Grant] You have a thesis. 318 00:13:34,947 --> 00:13:36,247 [Miles] That's exactly the line. 319 00:13:36,297 --> 00:13:37,847 [Miles] A thesis isn't a company. 320 00:13:38,047 --> 00:13:40,197 [Grant] YC will pressure test that fast. 321 00:13:40,597 --> 00:13:42,547 [Grant] Week two office hours are not soft. 322 00:13:42,597 --> 00:13:43,597 [Miles] The thing is, 323 00:13:43,797 --> 00:13:48,797 [Miles] NVIDIA paid $20 billion for Groq because they saw this exact gap coming. 324 00:13:48,921 --> 00:13:50,421 [Miles] That's not a small signal, 325 00:13:50,821 --> 00:13:54,471 [Miles] but seeing the gap and being the company that fills it are two different problems. 326 00:13:55,021 --> 00:13:55,271 [Grant] Yeah. 327 00:13:55,771 --> 00:14:00,121 [Miles] This founder knows the architecture; they know the agent execution model; the v 328 00:14:00,202 --> 00:14:07,452 [Miles] The VC Corner called it "the founder who understands both chip architecture and how agents actually execute has a rare window right now. 329 00:14:07,802 --> 00:14:08,902 [Grant] Key word being window. 330 00:14:09,302 --> 00:14:10,802 [Miles] So that's where it sits. 331 00:14:11,002 --> 00:14:12,602 [Miles] The decision hasn't been made. 332 00:14:12,852 --> 00:14:15,952 [Miles] S26 is three months long and it started this week. 333 00:14:16,302 --> 00:14:17,752 [Miles] We don't know which door they walk through. 334 00:14:18,052 --> 00:14:19,302 [Grant] We genuinely don't. 335 00:14:19,452 --> 00:14:20,402 [Miles] And neither do they. 336 00:14:21,452 --> 00:14:22,002 [Miles] All right, 337 00:14:22,102 --> 00:14:22,902 [Miles] that's the episode. 338 00:14:23,518 --> 00:14:26,168 [Grant] Yeah-and I keep coming back to that room: 339 00:14:26,768 --> 00:14:27,418 [Grant] no countdown, 340 00:14:27,768 --> 00:14:31,218 [Grant] no fanfare-the partner just starts talking. 341 00:14:31,918 --> 00:14:33,918 [Miles] That detail stuck with me too. 342 00:14:34,368 --> 00:14:38,118 [Miles] You spend months getting in and then day one it's just... work. 343 00:14:38,818 --> 00:14:40,818 [Grant] And somewhere in that room right now, 344 00:14:41,018 --> 00:14:43,318 [Grant] there's a founder who still hasn't made the call: 345 00:14:43,718 --> 00:14:45,118 [Grant] chip or scheduler. 346 00:14:45,368 --> 00:14:46,968 [Grant] The batch just started. 347 00:14:47,768 --> 00:14:50,418 [Miles] That's a cliffhanger I don't want to let go of. 348 00:14:50,910 --> 00:14:51,610 [Grant] Same. 349 00:14:52,010 --> 00:14:52,310 [Grant] Look, 350 00:14:52,610 --> 00:14:57,410 [Grant] if you know a YC founder in their first year who'll be willing to tell that story honestly, 351 00:14:58,010 --> 00:14:58,760 [Grant] send them our way: 352 00:14:59,210 --> 00:15:01,110 [Grant] yearone at heymeadow dot com. 353 00:15:01,410 --> 00:15:06,060 [Miles] And if this episode helped you think differently about what actually happens inside that room, 354 00:15:06,460 --> 00:15:07,260 [Miles] leave us a review, 355 00:15:07,560 --> 00:15:08,010 [Miles] it helps. 356 00:15:08,710 --> 00:15:09,160 [Grant] Miles: 357 00:15:09,660 --> 00:15:10,160 [Grant] good one today. 358 00:15:11,110 --> 00:15:11,310 [Miles] Yeah, 359 00:15:11,360 --> 00:15:12,160 [Miles] more soon. 360 00:15:12,410 --> 00:15:13,110 [Miles] Thanks for listening, 361 00:15:13,160 --> 00:15:13,410 [Miles] Everybody.