Latest / Newcomer Pod / OpenAI Code Red: The Drama Before the Next Big Drop
Transcript
- 0:00They're like declaring a code red.
- 0:01We're we're not winning on the product and now we're going to
- 0:05make the Welcome back to the Newcomer
- 0:17Podcast, Tom here, Eric and Madeline.
- 0:19We're going to talk now about Open AI and the state of affairs
- 0:23in the AI race because it's been a very chaotic seeming couple of
- 0:29days. Alert danger It's.
- 0:31Yeah, yeah, there are flashing lights happening over in the
- 0:34Mission St. Yeah, yeah, they're just running
- 0:37through the corridor screaming and throwing papers in the air
- 0:40at Opening the Eye. Because door, yeah, door locks
- 0:42are coming down. You know, there's smoke going
- 0:45through the opening eye offices, it looks like.
- 0:47Alien when they're locking down for the creature.
- 0:50And the, yeah, I know there are people that's like, you're in or
- 0:52out right now, Like, the water is filling into the rooms and
- 0:55people's hands pressed up against the glass.
- 0:58Yeah. So Sam Ultiman recently declared
- 1:00a code Red did. You order the code red.
- 1:03I did the job. Did you order the code red?
- 1:05God damn right I did. Inside I don't know why it's so
- 1:09funny to me he he sent out a memo to to employees saying that
- 1:13they they needed a code red alert on on Chachi PT that
- 1:16there's a lot of competitive threats in the app is not
- 1:19working as well as they wanted to right now.
- 1:21And this came a couple weeks after he also kind of warned
- 1:26employees that there was about to be a very bad news cycle
- 1:28around open AI, as Gemini and Google in particular, you know,
- 1:32was breathing down their necks. And they released a new model
- 1:35that was really good. And everyone now is down on open
- 1:38AI. Tom's literal first story in
- 1:41Newcomer said yeah, Gemini, Google, they're in a better
- 1:45place than you might think. Just yeah.
- 1:47Yeah, yeah. Read the newsletter.
- 1:50You could have made a bunch of money.
- 1:51I'd love to take more credit for it, but I was just talking to,
- 1:54you know, sources at Google who were like, yeah, we have a
- 1:57really good model here. I don't know why everyone
- 1:58decided that we were fucked. Like, you know, Google literally
- 2:02invented key parts of the transformer technology.
- 2:06Well, on the David Sachs, all the all in guys are like Google.
- 2:08The culture is gonna ruin the whole thing.
- 2:10It didn't make any sense. And all the hedge run guys have
- 2:12been super like standoffish about Google.
- 2:15I I don't understand it. Yeah, Yeah.
- 2:18Keep going. Yeah.
- 2:18So what was the threat? Obviously, Obviously.
- 2:22And they make money either way. That's the beauty beautiful
- 2:24thing with Google. They have Google Cloud, you
- 2:27know, and they have a search so they can drive the cost like
- 2:30they do with, you know, their workspaces stuff.
- 2:34You know they can screw over everybody else by offering
- 2:36things that some people want to build money, right?
- 2:38They're a monopoly, is what you're trying to say.
- 2:39Yeah, you're. Saying they own title.
- 2:43No. Is that improving the core line?
- 2:44I can't say say anything about that, but.
- 2:47Yeah, it's like they have the full stack.
- 2:49And I think what's been like clear in the last couple of
- 2:51weeks is that like they're competitive at every part of it.
- 2:54Like Google Cloud is a major player in cloud, which we all
- 2:57knew. The Tpus, which people kind of
- 2:59all forgot about, are like the only legitimate competitor to
- 3:01Nvidia's chips. And now they have one of.
- 3:04Capture how much Tpus we're going to be a credible
- 3:06competitor, I don't think. Yeah, well, they've been working
- 3:09on it for a long time. Like you compare that to like
- 3:10open AI, which is very Johnny come lately with all of these
- 3:14things. They're basically like building
- 3:16a airplane in the sky or whatever that metaphor is.
- 3:18Whereas like Google is literally the airplane and like, you know,
- 3:22they they certainly were slow when it came to embracing it and
- 3:25like had a lot of concerns ethically, which is interesting
- 3:28even though of. Course, I hope people listening
- 3:30this podcast already know this fact, but attention is all you
- 3:33need. The Seminole AI paper was
- 3:35written by Google employees, like, so they were, they were
- 3:39there, they knew it. They gave it away for free.
- 3:41They created this whole problem themselves.
- 3:43And so, yeah, it was frustrating that they weren't like, leading
- 3:46from the whole time, like you invented this category.
- 3:49But you know, yeah, not surprising that they would
- 3:51figure it out and and catch up. Right.
- 3:54And So what I guess is more interesting to me now is like
- 3:56the state of affairs going forward because the idea that
- 3:59open AI is fucked, it's it's literally just AAI mean open AI
- 4:04and Sam Altman assisted, but it's a creation of media.
- 4:07Nothing has fundamentally changed about the business like
- 4:10their model maybe isn't ranking as highly.
- 4:12On. Traffic.
- 4:13Also I think there are some traffic numbers.
- 4:15Right. Traffic is a little down, yeah,
- 4:16to HPT. But I will say also, Sam Altman
- 4:20is such a savvy media player. He knows that if he sends an
- 4:24e-mail to the entire company that says Code Red in a memo,
- 4:28that that's going to get leaked to a reporter like.
- 4:30And it's very possible Open AI leaked it itself.
- 4:33Like a lot of these company memos, they know they're going
- 4:35to come out and some of them the companies leak.
- 4:38You know, I I don't have proof either way on this one, but just
- 4:41as a reporter, these company memos, especially in the
- 4:44Facebook era, were coming out so often, but I think companies
- 4:48were definitely handing them out to reporters because it's like
- 4:50we're blasting them to all our employees.
- 4:52This is the message we want out there.
- 4:53And it gives it gives the statements some sincerity that
- 4:56you wouldn't get if you put it, if you put out a press release
- 4:58saying, oh, we're we're in trouble.
- 5:00Like people be like, what? But if you're like, it's to
- 5:02employees, it's like, well, that's what they're saying
- 5:04authentically inside. Then we metabolize it in the
- 5:07culture. They wrote this for public
- 5:08consumption, whether they leaked it or not, right?
- 5:10Yeah, it was for the public. I my, my belief is that they
- 5:15didn't leak it on purpose. And I think the reason you can I
- 5:19can say that confidently is that there actually weren't that many
- 5:21outlets that reported it with their own reporting.
- 5:23It's actually only been a small handful of them.
- 5:25So if everybody says we have a source, then the company is
- 5:28like, all right, yeah, yeah, that was true.
- 5:31Here's the here's the memo. Yeah.
- 5:32Yeah, but that said, like, Sam is a smart guy who knows how the
- 5:36media works, and when you do send something out, you know
- 5:38it's going to happen. So the level of intentionality
- 5:40is sort of irrelevant at that point.
- 5:42But I think like Google also had a code red exactly 3 years ago
- 5:48to this time where they were worried about the state of
- 5:52Google search and, and, and ChatGPT was, was taking off like
- 5:55a rocket. And here's where we are now.
- 5:57So, and I think Facebook in the past is, you know, Mark has
- 6:00declared various code Reds over, over different things.
- 6:03And I think actually over Google Plus, when that first came out,
- 6:06there was a lot of concern over there.
- 6:07So it's stupid to take from this like the beginning of the end.
- 6:11And and I saw some people tweeting that out.
- 6:12It's like when you've declared a code red, it's already over.
- 6:16You've already lost, you know, your men are already dead.
- 6:18And and it's like, we're so early in this.
- 6:22And I hate using metaphor, but it literally is true.
- 6:24Like if you can go from like an also ran like Google was a
- 6:27couple of months ago, like when I wrote that piece, like
- 6:30everyone thought Google was done and here we are months later and
- 6:34they have the best model everything.
- 6:35So it's like a lot is going to change.
- 6:37The thing that does sort of start to worry me about open AI
- 6:40is like I do think they have too many plates spinning right now.
- 6:44I think they are literally trying to become Google at a
- 6:47time that Google already exists. And I don't know like where
- 6:50their focus is going to end up being.
- 6:51Like you've got the Chachi PT stuff, you've got the Stargate
- 6:56stuff, you've got the Johnny I've stuff, you've got the
- 6:59browser stuff. Like I'm probably missing 5 or 6
- 7:01other things. They're trying to get an
- 7:02advertising right now. The meme is they're trying to
- 7:04get an ads. They're like declaring a code
- 7:07red. We're we're not winning on the
- 7:09product and now we're going to make the product worse with
- 7:12advertising. Right, right.
- 7:15Which man? The whole advertising thing too.
- 7:17You know, we had Alex Heath on while you were out, Eric.
- 7:19And you know, I, I like Alex a lot.
- 7:21He's an amazing, amazingly well sourced.
- 7:23Supporter. You can go read his newsletter.
- 7:24Yeah, go read sources and his podcast is called Access.
- 7:28But he was, you know, very, I, I thought a little bit too, I
- 7:31think a little, a little too credulous on the fact that ads
- 7:34is going to be a big business for, for open.
- 7:36AII think actually ads is going to be very difficult for them
- 7:39because ads rely on proof of concept and proof of success and
- 7:43also linking out and clicking out.
- 7:44And it's sort of proved that this stuff does not link out
- 7:46very well. So why would you want?
- 7:48I'm shopping for all my Christmas gifts in Shia GBT.
- 7:51I don't agree. Well, anyway, this is a
- 7:53different discussion. I just think it's going to be
- 7:55complicated for them. I don't think it's a done deal
- 7:57and I, I worry for them if I were an investor, which is
- 8:02increasingly everyone basically, you know, VCs on upwards that
- 8:08like they are going to lose focus and what they're trying to
- 8:11win on is like increasingly competitive.
- 8:13And, and that to me is more like the red alert than like Chachi
- 8:17BT is losing customers or something like that.
- 8:19It's just like what matters and what doesn't matter and should
- 8:21be doing should we be doing everything?
- 8:23And I would be, I know we'll get to predictions in a later
- 8:28episode, but I wouldn't be surprised if in the next year or
- 8:31two you start seeing open AI dropping some of these ambitions
- 8:36that something like the browser just sort of goes away.
- 8:39It's not that necessary or I don't know.
- 8:42I really don't have a lot of bullishness on the Johnny Eye of
- 8:45hardware project. I think that is going to be
- 8:48expensive and and difficult, but that's that's where I would be
- 8:52worried about them. If you're selling Tam, you need
- 8:55to be able to do everything. I mean, there's some valued
- 8:58opening eye and just being ready if a category appears promising
- 9:03that they can lean in. It's like, oh, coding is big,
- 9:05we'll lean into that. So you have projects sort of
- 9:07spinning. Yeah, I don't know.
- 9:10The piece that stuck out to me with all of this too, is that
- 9:13each iteration of the startup boom, whether we should be
- 9:18looking at applications or foundation models and how do we
- 9:20build out products and how do the large labs also build out
- 9:23products? Every time a new model comes
- 9:25out, like this junior Gemini model, the race is totally
- 9:29reset. So all of the people that I
- 9:30think have talked about, you know, will foundation models
- 9:33ever be the Moat? The answer seems to be no again,
- 9:37as it just keeps climbing and chasing back and forth.
- 9:40And every few months we get a new model that tops the
- 9:42leaderboard and then opening eyes back at the top and then
- 9:44Google's back at the top. So it's not that I don't think I
- 9:47would count them out at all. Obviously I think it's still too
- 9:50early, but it just becomes a lot trickier to see, you know, can
- 9:55you be the everything company when the everything company
- 9:57already exists? Like you need to figure out
- 9:59applications beyond ChatGPT that work because you will not get
- 10:02the Moat with foundation models like this.
- 10:04Race seems like it's nowhere near done.
- 10:05Which is why I think ads feels like heart before the horse in
- 10:09that, yeah, opening eye, ChatGPT needs more more clear designed
- 10:15applications before just, I don't know, going into
- 10:19monetization. I want to say 1.
- 10:22I guess my last word on this is I do think it's it speaks to the
- 10:28power of truth that open AI is saying code red.
- 10:31I do think there's value in shaping your company's narrative
- 10:34and just saying, yeah, we're we're in a hard place because it
- 10:37undercuts the media and the commentary.
- 10:40And just like Twitter and regular people would all be
- 10:43like, oh, open AI is in trouble because of Gemini.
- 10:47But when you say, oh, it's a code red, we're in big trouble,
- 10:50then you lower the bar for yourself.
- 10:52So maybe it hurts your reputation, but instead of
- 10:55having to fend off like, oh, Google, are they better than
- 10:59you, you say, yeah, we're we're really needing to work hard.
- 11:01And so then everything you do to sort of be slightly better than
- 11:06a Code Red is now a positive again.
- 11:08And so I think there's a lot of value in just, you know,
- 11:11communication strategy and lowering expectations for
- 11:15yourself so that you can much more easily slip over them.
- 11:18If you think entirely through narratives, then you're setting
- 11:21yourself up for your comeback. You're setting yourself up for
- 11:24like, we, we, we girded down and now everything is great.
- 11:27And we, we, we the same way Google sort of did when they had
- 11:30their code red and and now everyone's, you know,
- 11:32celebrating them. So genius move by Sam Altman,
- 11:35where we we applaud you tip of the cap for your great decision
- 11:39to call yourself a cap. I don't.
- 11:40Think Google's as savvy about the PRI.
- 11:42Think they're just just building and trying.
- 11:45I think Sam is very aware of, you know, how the narrative is,
- 11:49is consumed. Yeah, cool.
- 11:52Yeah. Well, fun to talk about this
- 11:54more next week or whatever when we do our prediction.
- 11:56Yeah. We're gonna do predictions.
- 11:57We're gonna have superlatives. I think it's gonna be a fun end
- 12:00of the year for the newcomer podcast where it's trying to
- 12:03split up segments so that you can listen to the things you're
- 12:07interested in. We're super.
- 12:09Hopefully you're watching this on YouTube.
- 12:10If you're listening on the podcast, we're we're working
- 12:13hard on our YouTube channel. So go subscribe on YouTube,
- 12:16please like comment, we're we're a small channel.
- 12:18I'll come in and reply to your comments.
- 12:20If they're not, they're not too mean.
- 12:22So we appreciate all the support.
- 12:24And of course, you know, we're at newcomer.co.
- 12:27We write a sub stack about stripes and venture capital.
- 12:30And, you know, if you want to be hearing about some of the stuff
- 12:33early, maybe you like to buy stocks, maybe you're tracking
- 12:37start-ups, maybe you're just fascinating.
- 12:39One insiders take. You should subscribe to the
- 12:42newsletter and newcomer.co.