1 00:00:00,040 --> 00:00:03,440 In today's episode, we're going to explore the leaked Open AI 2 00:00:03,440 --> 00:00:06,320 memo that came out just over a week ago. 3 00:00:06,520 --> 00:00:09,520 In that memo, Sam Altman said we're at a critical time for 4 00:00:09,520 --> 00:00:12,120 Chatji BT. Open AI is now playing defensive 5 00:00:12,120 --> 00:00:15,280 against Google Gemini 3. The prediction markets have 6 00:00:15,280 --> 00:00:20,360 shifted dramatically, and Google sits at between 87 and 92% odds 7 00:00:20,640 --> 00:00:24,280 to have the leading model by the end of this year versus Open AI 8 00:00:24,360 --> 00:00:27,040 at just 8%. Three years ago, you could never 9 00:00:27,040 --> 00:00:28,520 imagine something like that happening. 10 00:00:28,680 --> 00:00:32,240 Chatji BT was just released and Google announced Code Red. 11 00:00:32,400 --> 00:00:35,760 Google had their two previous founders come in, Larry Page and 12 00:00:35,760 --> 00:00:38,760 Sergey Brin. They came back and did all 13 00:00:38,760 --> 00:00:41,200 nighters to review code personally. 14 00:00:42,000 --> 00:00:44,440 Those are the two founders of Google that had stepped back 15 00:00:44,440 --> 00:00:47,560 from operations completely. Teams at Google were reassigned 16 00:00:47,600 --> 00:00:50,360 overnight and all resources were reallocated. 17 00:00:50,520 --> 00:00:53,800 It was in that year that it felt like the beginning of the end of 18 00:00:53,800 --> 00:00:56,720 Google's dominance. Fast forward to December 2025 19 00:00:56,720 --> 00:00:58,600 and things have changed dramatically. 20 00:00:58,800 --> 00:01:02,320 In today's episode, we're going to explore the open AI situation 21 00:01:02,520 --> 00:01:05,120 and see if it really is as bad as it seems. 22 00:01:17,850 --> 00:01:22,730 OK, so this week is yeah, another week of being I'll. 23 00:01:23,890 --> 00:01:27,290 The flu has caught me twice in the space of a month. 24 00:01:27,570 --> 00:01:30,290 I'd make it make sense. I have no idea how that even 25 00:01:30,290 --> 00:01:33,010 works. Recovering from a flu, having 26 00:01:33,010 --> 00:01:35,530 like 3 or 4 days recovered and then suddenly catching it again. 27 00:01:36,320 --> 00:01:39,240 So I'm still incredibly bummed up and feel very, very rough, 28 00:01:39,840 --> 00:01:42,120 but the show must go on. So for today's episode, we're 29 00:01:42,120 --> 00:01:45,240 going to start with what Code Red actually means because that 30 00:01:45,240 --> 00:01:48,480 is what Open AI have just kind of announced in that leaked 31 00:01:48,480 --> 00:01:51,080 memo. Well, if we break down what it 32 00:01:51,080 --> 00:01:56,120 actually says, it's really quite interesting because in it he 33 00:01:56,120 --> 00:01:58,880 actually broke down, and by him I mean Sam Altman, he broke down 34 00:01:59,080 --> 00:02:03,680 6 big areas of focus for Open AI, which is ChatGPT 35 00:02:03,680 --> 00:02:05,720 personalisation and customisation. 36 00:02:06,160 --> 00:02:09,479 It's better model behaviour measured by the LM Arena 37 00:02:09,479 --> 00:02:13,200 benchmarks, it's speed and reliability improvements. 38 00:02:13,400 --> 00:02:17,440 It's reducing over refusals which have become just a massive 39 00:02:17,880 --> 00:02:20,280 area of annoyance for most users on ChatGPT. 40 00:02:20,280 --> 00:02:22,760 They're going to try and launch a new reasoning model which they 41 00:02:22,760 --> 00:02:25,680 believe, although that it's just a claim, it's supposed to be 42 00:02:25,840 --> 00:02:29,320 ahead of Gemini 3. And finally, image generation 43 00:02:29,320 --> 00:02:32,120 improvements too. So Mark Chan, Opening Eyes Chief 44 00:02:32,120 --> 00:02:35,640 Research Officer, has become personally accountable to 45 00:02:35,640 --> 00:02:39,400 leading all daily progress calls on ChatGPT because fundamentally 46 00:02:39,400 --> 00:02:42,480 this is a refocusing on ChatGPT as a core product. 47 00:02:43,560 --> 00:02:46,200 So yeah, this is massive internal urgency at scale Open 48 00:02:46,200 --> 00:02:49,840 AI, and it's gone from yellow warnings to a red warning, which 49 00:02:49,840 --> 00:02:52,280 is their highest alert level internally yet. 50 00:02:52,280 --> 00:02:56,960 Most coverage online so far has completely ignored the fact that 51 00:02:56,960 --> 00:03:00,800 this wasn't a sudden change to, you know, code red. 52 00:03:01,400 --> 00:03:05,560 Code red was the escalation of a crisis that began a good few 53 00:03:05,560 --> 00:03:08,560 weeks ago. Now, back in October, Sam Altman 54 00:03:08,560 --> 00:03:12,640 sent an internal memo to different team members warning 55 00:03:12,640 --> 00:03:16,160 staff to expect quite a rough period of time, maybe even much, 56 00:03:16,160 --> 00:03:17,960 much slower growth as well financially. 57 00:03:18,200 --> 00:03:20,320 And he admitted in that memo that Google's been doing 58 00:03:20,320 --> 00:03:23,520 excellent work recently in almost every single aspect, 59 00:03:23,760 --> 00:03:26,760 particularly in pre training, which is an area that Open AI 60 00:03:26,760 --> 00:03:29,560 really haven't been able to innovate in for quite some time. 61 00:03:29,720 --> 00:03:33,000 So when Code Red has just dropped in December, it wasn't 62 00:03:33,000 --> 00:03:35,160 panic, it was the second alarm bell. 63 00:03:35,200 --> 00:03:37,840 And I think what's most telling with this announcement is the 64 00:03:37,840 --> 00:03:40,720 things that they've actively said they're going to drop. 65 00:03:40,840 --> 00:03:44,400 So all advertising plans, for example, are on hold, and they 66 00:03:44,400 --> 00:03:46,760 were quite far along. You've heard me speak about this 67 00:03:46,760 --> 00:03:49,680 quite a few times, saying how disappointed I would be to see 68 00:03:49,680 --> 00:03:51,400 that happen inside a large language model. 69 00:03:51,680 --> 00:03:53,080 But yeah, they were quite far along with that. 70 00:03:53,080 --> 00:03:56,480 I mean, Android beta code actually found in late November, 71 00:03:57,240 --> 00:04:01,440 explicit references to ADS, features, search ad carousels 72 00:04:01,800 --> 00:04:04,480 and bizarre content. So yeah, opening I were really 73 00:04:04,480 --> 00:04:06,960 building serious ad infrastructure and then forecast 74 00:04:06,960 --> 00:04:09,040 they were going to make a billion dollars in free user 75 00:04:09,040 --> 00:04:13,600 monetization by 2026, which is obviously now now been stopped 76 00:04:13,600 --> 00:04:15,840 and halted as a project. And to me, this shows how 77 00:04:15,840 --> 00:04:18,920 confident they must have been feeling at one point to be going 78 00:04:18,920 --> 00:04:22,000 for ads. And that makes me incredibly sad 79 00:04:22,000 --> 00:04:25,440 personally, that the ads were going to be brought into any 80 00:04:25,720 --> 00:04:28,960 ChatGPT or large language model. To me, this shows how confident 81 00:04:28,960 --> 00:04:31,880 they were clearly feeling at one point to even think about 82 00:04:31,880 --> 00:04:35,360 bringing ads. And in fact, the CTO of an AI 83 00:04:35,360 --> 00:04:40,720 startup called Hyperbolic posted a screenshot on X of Peloton 84 00:04:40,720 --> 00:04:43,800 advertisements inside the conversation and he was paying 85 00:04:43,800 --> 00:04:46,960 the $200 a month license. I mean, I've spoke about this 86 00:04:46,960 --> 00:04:49,880 before some Altman a good three years ago said ads were 87 00:04:49,880 --> 00:04:53,160 unsettling and and a less resort for a business. 88 00:04:53,160 --> 00:04:56,280 And yeah, earlier this year, he then said he really loves how 89 00:04:56,280 --> 00:04:59,600 Instagram do ads and thought opening I could bring some 90 00:04:59,600 --> 00:05:01,520 really cool product ads into their platform too. 91 00:05:01,680 --> 00:05:04,160 So yeah, they were clearly putting a massive amount of 92 00:05:04,160 --> 00:05:06,040 focus there. Another area they're dropping is 93 00:05:06,040 --> 00:05:09,200 shopping and health agents. So agents that help me buy 94 00:05:09,200 --> 00:05:11,520 healthy things and do healthy things. 95 00:05:11,680 --> 00:05:14,560 They've also paused the Pulse feature, which is that kind of 96 00:05:14,560 --> 00:05:17,560 proactive morning briefing system that Sam Altman said it 97 00:05:17,560 --> 00:05:19,440 was his favorite feature of ChatGPT. 98 00:05:30,600 --> 00:05:32,000 I guess what does this tell us then? 99 00:05:32,040 --> 00:05:35,680 It tells us that Open AI is consolidating its focus and its 100 00:05:35,680 --> 00:05:38,000 effort. They're abandoning revenue 101 00:05:38,000 --> 00:05:42,160 diversification just to double down on their core products, 102 00:05:42,160 --> 00:05:44,680 which is ChatGPT, and the quality they're providing to 103 00:05:44,680 --> 00:05:47,040 users, which has definitely, definitely dropped. 104 00:05:47,160 --> 00:05:50,760 Some may say that it's a very risky bet given the amount of 105 00:05:50,760 --> 00:05:53,240 financial pressures they're under, the fact that the 106 00:05:53,360 --> 00:05:56,400 economy, everybody's talking about AI bubbles and things 107 00:05:56,400 --> 00:05:59,200 going wrong. So there's a lot of pressure 108 00:05:59,200 --> 00:06:02,120 there and they've got incredible financial targets to hit. 109 00:06:02,280 --> 00:06:06,320 However, I do the exact same. I would want to focus on what is 110 00:06:06,320 --> 00:06:09,000 winning most of the market and become the best in the world at 111 00:06:09,000 --> 00:06:11,520 that thing. Because ultimately, if that 112 00:06:11,520 --> 00:06:14,000 foundation dies off, it's game over. 113 00:06:14,440 --> 00:06:17,800 And I mean, Steve Jobs famously said that it's just better to 114 00:06:17,800 --> 00:06:20,720 have less products, but do them incredibly well and do 115 00:06:20,720 --> 00:06:23,360 incredible marketing for those products and having lots of 116 00:06:23,360 --> 00:06:25,680 random products in the market. That confuses everybody. 117 00:06:25,720 --> 00:06:27,480 There's also a financial lens to all of this. 118 00:06:27,960 --> 00:06:30,320 Now, I'm not going to spend too much time here because I feel 119 00:06:30,320 --> 00:06:35,120 like there's so much discourse online about financial pressures 120 00:06:35,120 --> 00:06:37,800 and economic bubbles that, you know, it's so boring. 121 00:06:38,240 --> 00:06:41,960 But but yeah, open eye are of course, under lots of pressure. 122 00:06:42,160 --> 00:06:45,160 So they're burning through cash at a crazy, crazy rate, you 123 00:06:45,160 --> 00:06:49,120 know, $5 billion in losses in 2024, projected to be $8.5 124 00:06:49,120 --> 00:06:54,640 billion this year. And HSBC analysts forecast that 125 00:06:55,040 --> 00:06:59,440 they've got a $207 billion funding shortfall, so they're 126 00:06:59,440 --> 00:07:04,440 short of $207 billion in funding from now to 2030. 127 00:07:04,640 --> 00:07:07,440 And there was apparently some leaked financial figures that 128 00:07:07,440 --> 00:07:12,240 they expect to lose up to $74 billion in a year. 129 00:07:12,240 --> 00:07:16,440 The company's also committed to $1.4 trillion in infrastructure 130 00:07:16,440 --> 00:07:17,880 spending over the next eight years. 131 00:07:18,000 --> 00:07:20,520 And this is a really tough one because data centre 132 00:07:20,520 --> 00:07:23,640 infrastructure needs to be built out to hit your growth targets. 133 00:07:23,840 --> 00:07:26,760 So they'll look at the amount of users or products in the market, 134 00:07:26,880 --> 00:07:29,680 therefore how much data centre capacity do they really need? 135 00:07:30,080 --> 00:07:32,440 And based on the products they want to launch, they're going to 136 00:07:32,440 --> 00:07:34,360 need more data centres and GPU's. 137 00:07:34,680 --> 00:07:39,040 So naturally, that creates a fixed amount of spending on data 138 00:07:39,040 --> 00:07:41,920 warehouse infrastructure. And if they don't commit to it 139 00:07:41,920 --> 00:07:44,440 now, they won't have it in the future because it's not 140 00:07:44,440 --> 00:07:46,520 something that you just, you know, it's not like buying a car 141 00:07:46,800 --> 00:07:48,520 and going to the shop and driving away with it. 142 00:07:48,680 --> 00:07:50,200 It needs to be built out for you. 143 00:07:50,680 --> 00:07:53,560 So these are things that they can't escape from as a financial 144 00:07:53,560 --> 00:07:56,720 and a business model. Well, gosh, it makes it very 145 00:07:56,720 --> 00:07:58,760 expensive to run. And they're projected that 146 00:07:58,760 --> 00:08:01,760 they're going to need $200 billion in revenue by 2030 as a 147 00:08:01,760 --> 00:08:04,280 result just to become profitable. 148 00:08:04,760 --> 00:08:07,920 And the current run rate is $13 billion. 149 00:08:08,960 --> 00:08:12,720 And a run rate as a word, it basically means it's a, it's a 150 00:08:12,720 --> 00:08:15,600 very simple and quick financial projection that essentially 151 00:08:15,600 --> 00:08:19,560 takes your current short term revenue and earnings, say like 152 00:08:19,560 --> 00:08:22,760 monthly or quarterly earnings, and then uses them to 153 00:08:22,760 --> 00:08:25,760 essentially guess what your total yearly earnings would be. 154 00:08:27,000 --> 00:08:29,280 So that means the math just doesn't work out. 155 00:08:29,360 --> 00:08:31,920 They're not earning enough money per month and per quarter to 156 00:08:31,960 --> 00:08:35,679 ever hit that amount. So they're going to have to just 157 00:08:35,679 --> 00:08:38,760 keep growing bigger and bigger and faster and raise more and 158 00:08:38,760 --> 00:08:40,880 more money to do it. And the way that that's going to 159 00:08:40,880 --> 00:08:43,320 work out is having market dominance. 160 00:08:43,760 --> 00:08:45,960 If they don't have market dominance, people don't believe 161 00:08:45,960 --> 00:08:47,160 in the product. So they're not going to give 162 00:08:47,160 --> 00:08:49,440 them that much money. You know, their valuation jumped 163 00:08:49,440 --> 00:08:55,520 from about $157 billion in October 2024 to double that, so 164 00:08:55,520 --> 00:09:00,000 double 150 billion to $300 billion by March of this year. 165 00:09:00,160 --> 00:09:03,200 And then by October of this year it was at $500 billion. 166 00:09:03,360 --> 00:09:08,320 Now for a what is technically a startup that is a 38 times 167 00:09:08,320 --> 00:09:10,200 multiple on how much they actually earn. 168 00:09:10,280 --> 00:09:15,840 And for context, NVIDIA huge earnings, they are trading at 24 169 00:09:15,840 --> 00:09:18,240 times their revenue. So they're valued at 24 times 170 00:09:18,240 --> 00:09:20,920 their revenue. Open AI is valued at 38 times 171 00:09:20,920 --> 00:09:24,040 their revenue. So yeah, to achieve these crazy 172 00:09:24,040 --> 00:09:27,640 numbers and market narrative and the amount of money that people 173 00:09:27,640 --> 00:09:30,640 want to put into them, they have to be a dominant player in the 174 00:09:30,640 --> 00:09:34,080 market. So Gemini 3 coming along and 175 00:09:34,080 --> 00:09:37,280 potentially beating them is an existential threat, especially 176 00:09:37,280 --> 00:09:40,080 because all of their competitors have access to better credit 177 00:09:40,080 --> 00:09:42,480 than them. Google can get money easier and 178 00:09:42,480 --> 00:09:44,520 quicker than Open AI. Now then if we go further and 179 00:09:44,520 --> 00:09:48,080 look at what users are actually saying, 1 developer on the Open 180 00:09:48,080 --> 00:09:51,240 AI community forum said basically because of all the new 181 00:09:51,240 --> 00:09:55,160 changes, it's basically made ChatGPT as a subscription just 182 00:09:55,160 --> 00:09:57,840 completely useless in most cases because either when they're 183 00:09:57,840 --> 00:10:01,640 trying to do coding tasks it is overly verbose, but people who 184 00:10:01,640 --> 00:10:04,840 are trying to do writing tasks, it's not verbose enough. 185 00:10:04,960 --> 00:10:07,840 It's also got a big problem with over refusals as I've mentioned 186 00:10:07,840 --> 00:10:11,600 before, which is when Open Eye just explicitly refuses to do 187 00:10:11,600 --> 00:10:13,560 things for you. If you ask it what the very best 188 00:10:13,560 --> 00:10:16,480 biscuit type is, it will say that you shouldn't be biased 189 00:10:16,480 --> 00:10:19,520 towards biscuit types. 1 user asked for an explanation of 190 00:10:19,520 --> 00:10:22,960 Russian Roulette and it refused to give them the rules even 191 00:10:22,960 --> 00:10:24,880 though the question was just out of curiosity. 192 00:10:25,440 --> 00:10:27,560 I actually tried to do this over the weekend as a test. 193 00:10:27,800 --> 00:10:30,520 I wanted a series of questions of what would you do in this 194 00:10:30,600 --> 00:10:34,440 situation and I wanted to make them either scary or cross and 195 00:10:34,440 --> 00:10:36,600 rude and you just couldn't get it to work. 196 00:10:36,800 --> 00:10:39,120 And then number one biggest problem is still sick of fancy 197 00:10:39,280 --> 00:10:42,200 the model just agreeing with everything somebody is saying 198 00:10:42,200 --> 00:10:45,440 and refusing to push back or giving false validation. 199 00:10:45,560 --> 00:10:47,080 Now open air, I say they're going to be focusing on 200 00:10:47,080 --> 00:10:48,960 personalization and customization. 201 00:10:48,960 --> 00:10:51,320 So obviously they're trying to fix this, but most people are 202 00:10:51,320 --> 00:10:53,480 wanting a model that is going to disagree with them. 203 00:10:53,560 --> 00:10:56,600 Another major problem open air I've got here is the switching 204 00:10:56,600 --> 00:10:58,800 problem. So one Reddit user perfectly 205 00:10:58,800 --> 00:11:02,400 captured this. Claude smokes GPT for Python And 206 00:11:02,400 --> 00:11:05,480 it isn't even close on my end. Another said that they switched 207 00:11:05,480 --> 00:11:09,400 to Claude yesterday and it enabled them to build an entire 208 00:11:09,400 --> 00:11:13,520 phone app and because of the context window it required you 209 00:11:13,520 --> 00:11:16,400 to press continue 4 separate times so it ran out of context 210 00:11:16,400 --> 00:11:19,640 and then needed to start again, yet perfectly started where it 211 00:11:19,640 --> 00:11:23,160 left off and did a really really good job. 212 00:11:23,360 --> 00:11:27,920 The benchmarks back all of this up on the SWE benchmark, which 213 00:11:28,120 --> 00:11:31,920 is the gold standard for evaluating coding Cloud 4 Sonnet 214 00:11:32,120 --> 00:11:36,120 scored 72% and GPT 4 O scored 33%. 215 00:11:36,360 --> 00:11:39,000 And it also shows in the way that the world has adopted these 216 00:11:39,000 --> 00:11:41,400 different language models. Cursor adopted Claude as their 217 00:11:41,400 --> 00:11:44,800 main programmer language. GitHub did the same Windsurf 218 00:11:45,040 --> 00:11:48,560 source graph, CODI, or bank code, they all did the same. 219 00:11:48,680 --> 00:11:50,960 The financial impact of this is pretty massive. 220 00:11:51,160 --> 00:11:54,280 Claude Code has hit a billion dollars in revenue in six 221 00:11:54,280 --> 00:11:57,480 months, and Claude is out there acquiring a series of different 222 00:11:57,560 --> 00:12:00,400 businesses to make their core code product even better. 223 00:12:00,520 --> 00:12:03,080 So they're not just winning the market or head in the market, 224 00:12:03,320 --> 00:12:07,520 They are investing extremely heavily into becoming the leader 225 00:12:07,520 --> 00:12:10,120 and winner of that market. And you'll notice in the list of 226 00:12:10,120 --> 00:12:14,520 priorities that I said some Altman laid out, coding wasn't 227 00:12:14,520 --> 00:12:16,640 in there. So yeah, my read is that Open AI 228 00:12:16,640 --> 00:12:20,120 is kind of quietly accepting and conceding that the market has 229 00:12:20,120 --> 00:12:22,880 been lost for now for coding to Anthropic. 230 00:12:23,160 --> 00:12:24,760 They've still got their codecs product out there. 231 00:12:24,760 --> 00:12:28,160 They're probably going to still continue it, but clearly it's 232 00:12:28,160 --> 00:12:30,760 not the main priority anymore. And it makes sense because 233 00:12:30,760 --> 00:12:33,360 really they're having huge pressure from the consumer 234 00:12:33,360 --> 00:12:36,680 market when it comes to Gemini and that's their main base of 235 00:12:36,680 --> 00:12:40,400 users and revenue, 800 million ChatGPT users. 236 00:12:40,480 --> 00:12:43,040 So yeah, they are facing huge competition and I think it's 237 00:12:43,040 --> 00:12:45,080 worth spending a little bit of time on that competition. 238 00:12:45,160 --> 00:12:49,640 On benchmarks, Googles Gemini 3 currently sits at #1 on the LM 239 00:12:49,640 --> 00:12:54,480 Arena with a record 1400 score. Anthropic dominates 6 of the top 240 00:12:54,480 --> 00:12:58,640 ten spot and Chat GBTS 5.1 is only in 6th place. 241 00:12:59,760 --> 00:13:01,880 That's not terrible, but it's also not market leadership 242 00:13:02,040 --> 00:13:05,160 because you know, on user metrics, ChatGPT still has 800 243 00:13:05,160 --> 00:13:07,880 million weekly active user, which is which is just enormous. 244 00:13:07,880 --> 00:13:11,800 And apparently according to open AI, they dominate with 70% of 245 00:13:11,800 --> 00:13:13,960 all AI assistant usage. That's not going to last 246 00:13:13,960 --> 00:13:16,680 forever. Gemini just jumped from 450 to 247 00:13:16,680 --> 00:13:20,680 650 million active users and that's monthly active users. 248 00:13:20,680 --> 00:13:22,760 So yeah, quite far behind open AI. 249 00:13:22,760 --> 00:13:24,960 But they will and are able to catch up. 250 00:13:25,160 --> 00:13:27,760 And that happened between July and October of this year. 251 00:13:27,840 --> 00:13:32,600 And a metric that will really worry Open AI is session time. 252 00:13:33,360 --> 00:13:37,520 Google Now has a better session time than Open AI. 7 minutes and 253 00:13:37,520 --> 00:13:40,640 8 seconds versus 6 minutes and 25 seconds Open AI. 254 00:13:40,840 --> 00:13:42,920 When users are spending more time with one product over 255 00:13:42,920 --> 00:13:45,000 another, it probably will start to tell you something. 256 00:13:45,520 --> 00:13:49,320 Unless of course the extra couple of minutes and Gemini is 257 00:13:49,320 --> 00:13:53,400 spent being really frustrated, which I don't think is the case. 258 00:13:53,960 --> 00:13:56,920 And chat GPTS daily active users dropped 6% recently. 259 00:13:57,080 --> 00:13:59,400 They're also facing massive competition from an enterprise 260 00:13:59,400 --> 00:14:02,080 market perspective. According to a mid 2025 report, 261 00:14:02,080 --> 00:14:06,000 Anthropic now has 32% of enterprise market share versus 262 00:14:06,040 --> 00:14:09,320 opening eyes 25%. Now just one year prior, Open 263 00:14:09,320 --> 00:14:12,080 air had 50% and Anthropic just had 12%. 264 00:14:12,080 --> 00:14:14,880 That's a massive reversal. Open AI also facing a massive 265 00:14:14,880 --> 00:14:18,480 talent drain right now as well. Dozens of top researchers have 266 00:14:19,040 --> 00:14:23,200 left for competitors, you know, Open AI CTO launch thinking 267 00:14:23,200 --> 00:14:26,200 machines and recruited loads of incredible talent from Open AI. 268 00:14:27,440 --> 00:14:31,000 Meta's new super intelligence lab obviously then poached 269 00:14:31,000 --> 00:14:33,840 people aggressively too. And Mark Chen revealed how 270 00:14:33,840 --> 00:14:36,880 intense this talent wall really is because Meta is spending $10 271 00:14:36,880 --> 00:14:40,240 billion annually on acquiring AI talent. 272 00:14:40,400 --> 00:14:43,560 And apparently Zook has been hand delivering people soup to 273 00:14:43,560 --> 00:14:45,040 try and persuade them to come on board. 274 00:14:45,120 --> 00:14:48,600 So yeah, no wonder the Open AI has has sounded the alarm and 275 00:14:48,600 --> 00:14:50,600 made Code Red such a big priority. 276 00:15:02,520 --> 00:15:05,840 See, over the next period, I think you'd need to be all 277 00:15:05,840 --> 00:15:08,360 watching for a few things and I'll lay them out just now. 278 00:15:08,560 --> 00:15:12,280 So I think first and foremost, the Code Red memo promised a new 279 00:15:12,280 --> 00:15:14,800 reasoning model over the next few that's supposedly better 280 00:15:14,800 --> 00:15:17,560 than Gemini's new model. Apparently Mark Chen, the chief 281 00:15:17,560 --> 00:15:20,960 research officer, said that Open AI had succeeded in infusing a 282 00:15:20,960 --> 00:15:23,920 smaller model with the same amount of knowledge than 283 00:15:24,080 --> 00:15:26,640 previously was required for a big, large system. 284 00:15:26,800 --> 00:15:30,320 This is important because when they throttle intelligence so IE 285 00:15:30,640 --> 00:15:33,600 reduce the number of tokens and make everything very short and 286 00:15:33,600 --> 00:15:37,360 concise, it just makes you feel like it's just stupider. 287 00:15:37,880 --> 00:15:41,440 Every single person wants the most powerful, cleverest model 288 00:15:41,440 --> 00:15:44,040 possible all of the time. A second thing to look out for 289 00:15:44,040 --> 00:15:47,160 is user attention. So Chachi BT's engagement 290 00:15:47,160 --> 00:15:50,840 metrics are still incredibly strong to 12.74 visits per 291 00:15:50,840 --> 00:15:53,040 unique visitor. But Gemini's user base is 292 00:15:53,040 --> 00:15:57,800 growing really quickly, as I said, from 450 million to 650 293 00:15:57,800 --> 00:16:00,200 million in just a few months. People will be watching very 294 00:16:00,200 --> 00:16:03,520 closely to see if Open AI's improvements in their product 295 00:16:03,840 --> 00:16:08,160 leads to a more retained user base or just a slower attrition 296 00:16:08,160 --> 00:16:10,480 and a slower decline. A third big thing to watch for 297 00:16:10,480 --> 00:16:13,440 will be enterprise wins. The Fortune 500 market is 298 00:16:13,440 --> 00:16:16,160 incredibly important to these products and if you see 299 00:16:16,160 --> 00:16:19,840 Anthropic winning over and over again, well, it will tell you a 300 00:16:19,840 --> 00:16:23,480 lot about where Open AI is heading. 4th is developer 301 00:16:23,480 --> 00:16:26,600 sentiment. The coding market is where AI 302 00:16:26,600 --> 00:16:30,600 accomplishes big long term value through API revenue to 303 00:16:30,600 --> 00:16:33,960 integration, developers building products on their on their 304 00:16:33,960 --> 00:16:37,120 platform. And right now Claude is owning 305 00:16:37,120 --> 00:16:39,720 that conversation. Open AI has to close the coding 306 00:16:39,720 --> 00:16:40,880 gap. See, there's some of the big 307 00:16:40,880 --> 00:16:42,080 things that I've been looking for. 308 00:16:53,440 --> 00:16:59,680 Now to conclude this episode, I think what I find so interesting 309 00:16:59,680 --> 00:17:02,560 about this moment in time right now is that, you know, just 310 00:17:02,560 --> 00:17:05,319 three years ago, smart money would say that Google was going 311 00:17:05,319 --> 00:17:06,640 to struggle. You know, they've been caught 312 00:17:06,640 --> 00:17:10,040 completely flat footed. The founders had to come back. 313 00:17:10,040 --> 00:17:12,800 The panic was real. Market sentiment was very 314 00:17:12,800 --> 00:17:15,119 negative. Yet today Google has the best 315 00:17:15,119 --> 00:17:18,000 rank model in the world, the fastest growing user base in the 316 00:17:18,000 --> 00:17:21,839 world and the infrastructure advantage from building their 317 00:17:21,839 --> 00:17:25,440 own chips, their own products, just vertically integrated 318 00:17:25,440 --> 00:17:28,200 product set, whereas Open AI are now in a position that Google 319 00:17:28,200 --> 00:17:30,680 were in just a few years ago. The difference is Open AI 320 00:17:30,680 --> 00:17:34,720 doesn't have Google's resources, They don't have distribution, 321 00:17:35,520 --> 00:17:38,720 they don't have the hardware vertical integration, and 322 00:17:38,720 --> 00:17:41,320 they're burning through capital are a crazy, crazy rate. 323 00:17:41,520 --> 00:17:44,400 And they're having to fight a consumer war against Google. 324 00:17:45,040 --> 00:17:47,880 Yeah, they're fighting a coding war against Anthropic and an 325 00:17:47,880 --> 00:17:49,960 enterprise market war against Microsoft. 326 00:17:50,160 --> 00:17:52,240 These are all very different markets with very different 327 00:17:52,240 --> 00:17:54,560 dynamics. So, yeah, opening eyes, Code Red 328 00:17:54,560 --> 00:17:58,760 really is a fantastic reminder that perceived market leadership 329 00:17:58,760 --> 00:18:02,760 never lasts forever. However, Google's return also 330 00:18:02,760 --> 00:18:06,840 shows that anything is possible. See, I hope you enjoyed today's 331 00:18:06,840 --> 00:18:08,880 episode and I'll see you next week.