Big Technology Podcast - Bonus: The DeepSeek Reckoning in Silicon Valley

Episode Date: January 27, 2025

M.G. Siegler is a writer and investor and the author of Spyglass. Siegler joins Big Technology for a bonus depisode to discuss DeepSeek R1, the Chinese open-source AI model and its impact on the tech industry. Tune in to hear why DeepSeek's ability to match OpenAI's performance at just 3-5% of the cost could upend the AI industry's economic model. We also cover the immediate market fallout, why Silicon Valley's scaling hypothesis might be invalidated, and what this means for companies like Microsoft, Google, and NVIDIA. Hit play for a timely analysis of one of the most significant developments in AI that could reshape the technology landscape. --- Enjoying Big Technology Podcast? Please rate us five stars ⭐⭐⭐⭐⭐ in your podcast app of choice. For weekly updates on the show, sign up for the pod newsletter on LinkedIn: https://www.linkedin.com/newsletters/6901970121829801984/ Want a discount for Big Technology on Substack? Here’s 40% off for the first year: https://tinyurl.com/bigtechnology Questions? Feedback? Write to: [e-mail]

It's time for a bonus episode exclusively about DeepSeek R1 as the Chinese open source AI model Royals markets and threatens to upend the generative AI industry. That's coming up right after this. Welcome to Big Technology Podcast. We're doing a bonus edition today exclusively on DeepSeek, what it means for the AI industry, what it means for markets. We're going to touch on technology. We're going to touch on business. And so thrilled that you're here for a bonus. episode with us. We're joined today by M.G. Siegler. He's a writer and investor. He writes Spyglass. You can find

it at Spyglass.org. It's a great newsletter. It's a must read for me. And he has a great piece out called AI. A Finds Away. Has Deep Seek changed the AI game or just some equations? MG, great to see you. Welcome to the show. Great to see you, Alex. Thanks for having me back. And sorry for my crazy winter beard. It's, uh, it is very cold and rainy right now in London. So I'm not, I'm not ready for spring yet. If hey, it fits the season. I was just out in London to interview Demis from DeepMind. That's right.

I listened to that. That was very good. Yeah, and very timely now. Yes, I can confirm the sun does not shine in that city this time of year. So first of all, I want to talk a lot about, I mean, only about Deep Seek and Deep Seek R1. And what it means for the AI industry right now, it's we are just about, the markets will open on this show. So I'll have a sense as to what it's going to do today. But it's looking pretty bad, especially for Invidia and some others.

as we get going, I just want to thank all the podcast listeners who pointed me to Deepseek because we had some comments that came in over the past few weeks. I was able to ask Demis about it. I was able to get it in as the lead story on Friday show. So thank you. I appreciate all of you for pointing me towards Deep Seek. So let me just talk a little bit because we didn't touch on this Friday and we're going to definitely fill some holes that were left on the Friday show. We talked a little bit about how much it cost to train this model, but not necessarily about the benchmarks it hit and about the cost it costs to use this thing. So first of all, it's an open source model. It's much smaller than any of OpenAI's model. Yet, on the AIME mathematics test, it scored 79.8% compared to

Open AIs 01 scoring 79.2%. So it bests Open AIs best model on that. It scored 97.3% on the math 500, and it beat Open AI, which scored 96.4%. Look, these are lots of different benchmark tests, but you can tell that just by these numbers, it holds its own. And now the most remarkable part about this, it costs 55 cents per million token inputs and $2.19 per million token outputs. Just to give you a sense, Open AI costs $15 per million input tokens and $60 per million output tokens. That's 3.5% of the cost that it costs to run Open AI's 01 models. And you can do it again. It's open source. You could download onto your computer and run it. So basically what Deepseek R1 has done in a nutshell, and then we'll turn it over to MG, is it has created models that are as performant as the state of the art, right?

It's ranked number three in the chatbot arena at 3 to 5% of the cost. And that has huge implications for the technology, for the business, and we're going to get into those. So M.G., first question for you, if there was an AI Richter scale, right, assessing how big of an earthquake this is, what would you, give this development um so i mean it depends on on what i guess uh level your your you're sort of measuring the magnitudes right because as you noted the markets will open and that's going to be you know right now you know last i looked in in pre market trading nVIDia was down i think 11 per 10 to 11 percent um and that's the biggest hit right now microsoft a bunch of others are like in the three percent range so uh you know from a

pure market perspective, it seems like it's, let's call it an eight. You know, it's not, it's not going to totally destroy the stock market right now, but it's going to be rough, it seems like today. From a bunch of other perspectives, I think, you know, it's probably a little bit less of a shake in these earlier days. And I think that's because everyone's still even now sussing out what exactly this means for all different sorts. of things. You noted how much cheaper it is to run than, say, Open AI's models. And, you know, over the weekend, just reading all of these sort of reports about the model and how many individual startups are even just changing, swapping out, right, already, because it's so much

cheaper to do what they're doing right now by swapping in deepseeks models. And so what does that do immediately? Like, you know, do we have to have price cuts immediately? And, you know, I think you could sort of see open eye doing some stuff. I think Sam Altman tweeted, you know, maybe on Friday, like about how they were like bundling, rejiggering some of the bundles, right, that they have, like what's in the free offering and stuff. And it sort of feels like we're going to see more of that, you know, is a response, obviously, to some of this.

But then, you know, there was a, there was a big report, I think, in the information about meta's response to this in particular, which seemed pretty interesting in that, like, you know, it's all hands on deck, certainly. and there's like all these different teams meeting up. It's war room time. They don't do war in time. You and I remember that from the old school days of Facebook. And so, yeah, it's just like all of these companies are now scrambling.

You have Satya and Adela tweeting out things, you know, which seemed directly aimed at the market to try to, you know, ease that pain a bit. But anyway, going back to the original question on the Richter scale, you know, overall, I think a lot of people are still figuring this out. But right now the market thing is going to be the most acute one because that's obviously going to open. And I think it's going to be pretty hard for, you know, this day at least. And then I think I read some of the early analyst reports on this. And they're all over the place, right? Like there are some folks who are saying like, oh, this is this is awful for NVIDIA. Some folks are saying, you know, this is not a big deal.

This actually could be good in the longer run for NVIDIA in ways. And, you know, and then from big tech on down, what the ramifications. are there. And you mentioned that some startups are already swapping in deep seek R1 for the models they're using right now. How widespread do you think that is? Are they, are any of the startups that you speak with saying, okay, well, to hell with opening eye, to hell with Lama, time to put deep seek in? Or is this just beginning? Because it's again, something that dropped last week. Yeah, I think this is just beginning. I think, you know, people will experiment with it, right, just to see like how much you could, you know, get while swapping them out, given the price differentiation you were talking about. But also,

there's downsides, of course, like people have noted sort of the, you know, the censorship within China and of certain terms. And so, you know, I don't think everyone is quite certain what's in there. You know, it's an open source in that it's open weight, but it's not, you know, it's not clear exactly everything that's going on in there right now. And so I do think that if this proves out, say if Deep Sea can release another iteration of the model and it still is on the same sort of, you know, footing, I think that then you'll start to see more startups potentially taking it really seriously. I think now it's just a wait and see approach for sure and just people trying out to see if it is, in fact, as good as they say. Because I think,

you know, part of this, like my initial gut reaction, you know, Deep Seek, obviously, as you noted, had been around for, you know, basically since December and didn't really get all of the mass of pylon until sort of Friday, right, when R1 came out. And in part, it's like, you know, I've just, I don't know why my mind was drawn to this, but it's sort of like when they were talking about the room temperature conductor, right? Like, and everyone was talking about, oh, my God, like, there's this, there's this huge breakthrough that's happened. And this is going to revolutionize everything. And then it turns out, oh, you know, maybe there was some, some funny business in that claim. And maybe it wasn't, you know, all it was cracked up to be. And

of course, that turned out to be the case. And so I'm not saying, obviously, that's not the case with deep seek. It seems like now this R1 release has legitimized it. And it. As you note, on leaderboards and whatnot, people have been testing this. And again, the startups are part of that pressure test. Right. And so the funny business, just to get this out of the way, the funny business might be on the training side. Like, we think that they trained it for much less money. We think that they trained it with inferior GPUs that have been sort of the only things that can get their hands on due to export controls.

We're not 100% sure if that's the case. Right. But I think the bottom line here is that this is an open source model. it has been replicated. I mean, it has been downloaded to people's computers and used as effective as it is. And I think that the thing is, the methods and the cost savings and the performance, that's all real. So even if, you know, basically all of Silicon Valley without those export controls couldn't do this or didn't do this. And maybe it's because they had a different method and we'll get into that.

But the fact is that no matter, there's no putting the genie back in the bottle right now, which is that this company has created something that can rival open AIs performance at 3% of the cost. That's the big thing. So I also just think like the overall mentality is one of the more interesting sort of earthquakes to, you know, use your phrasing of it that's happening right now. It's like, and I think Steven Sinovsky summarizes well. He wrote, you know, very long tweet thread as he is wont to do. But then he also published it on his newsletter as well.

But he goes into the history and he obviously has a lot of good. historical context from Microsoft days on forward about what, you know, is going on here. But it's also, I think, important to talk through, like, how the constraints that were put in place by the U.S. because of the, you know, everything going on with chip constraints and sort of forcing AI companies not to export to China, you know, led to sort of this very interesting culdron that I think could only happen in a place like China right now because they're so constrained, whereas in the U.S., it's still the period of abundance, right, with AI and everyone's going after the scaling, and it's just not something they were going to focus on trying,

you know, they're making the smaller models, they're making the mini versions of the models, and those are great, and we're seeing that. But China, you know, the folks working in China had to do this this way, and I just think it's something you couldn't have seen in hindsight arise out of the U.S. in our current environment. Right. Okay, so I want to talk quickly about the technology, very quickly about the technology, and then get into some of the more business side applications here. So, MG, could you tell us just at a really high level what DeepSeek has done to be able

to get these results? Because, you know, it's one thing to say, okay, they were able to do it on worse chips with a smaller amount of data. But I think just it's important to very briefly highlight just the technical technological innovation here. Yeah, I mean, so, and I'm not a, I won't be a technical expert on this, but from my understanding, it's basically, you know, obviously, as you know, it started, the deep seek project started out of a hedge fund that was focused on quant trading, you know, in China. And they had acquired a bunch of Nvidia chips. I think they were H-100s, you know, before all the import restrictions came in. And basically they had those servers up and running and, you know,

presumably they were running a bunch of different models, including some of open AIs, but including also a bunch of the Lama stuff that Meta's been working on. And, you know, they've just used the process of distillation to, you know, effectively bring those bigger versions of the sort of state of the art models and distill them down into, you know, smaller models, which eventually led to this R1, you know, the equivalent of O1 on OpenAI side. And again, for a fraction of the cost, fraction of the compute, and a fraction of the size for these to be able to run. And that latter part seems like it's sort of being under discussed right now, but is important because, yeah, all of these models have constraints about how you can run them like on your personal machines, right?

because, you know, they're going to require so much RAM and so much memory to be able to do that. And if you can get them down to really small sizes, which, again, the bigger U.S. companies have been doing with these mini models, but they're sort of taking this bifurcated approach, whereas, you know, now we're getting to the point with this R1 model where it seems like it can run on pretty much a lot of different type of hardware, which, again, they need to do in China because of the restrictions that they have there. Right. And there's also a methodology change here, which is that they've gone from,

effectively self-supervised learning, which is what has been used to train all of the LLMs, all the big LLMs to this point, to pure reinforcement learning where the models tend to figure out what's what the right answer is on their own, which is just fascinating. Yeah, and it seemed like the, you know, sort of the American powers that be maybe felt like we weren't ready for that yet to happen, right? Like that was always the hope that we get to those points and that, you know, we still were in the in the scaling point again where you know you need yeah someone in the loop to be able to check and and make sure all these things are working and china you know this this chinese company because

of the some of the restrictions that we just talked about like just went for it and you know it's proving itself right and and just to harp on one more technical issue before moving on the distillation of models to me is fascinating that they can take any big model and distill it using this form of training and effectively be able to replicate its performance. So I could take they took I could take like a llama model with just 70 billion parameters and distill it and then all of a sudden run it with this reasoning reinforcement learning style approach and it's cheaper more efficient. It's just I mean again like I think the entire world is still trying to wrap their head around this and they'll be more on this feed to talk about exactly how impressive this is. But to me in the early innings of this,

That is astonishing. Yeah. And I mean, it, again, at a high level, it makes sense. It's just, it's incredible how it's happened. Because, like, do you need all of the world's knowledge, you know, in every single model for every single use case? Of course not. Like, that's going to be overkill for almost everything that you're going to do. And so does it point to a world where, yeah, we sort of lead towards more of these specialized models that are distilled?

And obviously, that's been happening. But this one is still, you know, a model that can effectively do. most everything distilled down from from those bigger ones so there's one sort of big question that i think needs to be asked here um which is there's been this all the silicon valley and you point to this in your piece all silicon valley has been operating on effectively the scaling hypothesis which is that you add more compute we talk about all the time on the show add more compute add more data um add more power uh add more training time effectively to these models and you will improve and now what deep seek has shown is that you can actually do all this without that and so

I'm curious if you think that this invalidates the scaling hypothesis because and it might seem kind of like a you know obscure thing but it's very important because this sort of sets up the whole business conversation which is if the scaling hypothesis is invalidated then all that A multi-trillion dollar investment in Nvidia CPUs, my bed, becomes sort of thrown into questions. So what happens to the scaling hypothesis from here? And it's fascinating timing, too, right? Because this is at the same time that everyone has now talked about, sort of the quote-unquote AI

wall being hit, right? And even Demis, you know, when you talk to him, he noted that he doesn't necessarily believe in, you know, a wall. being hit, but he did acknowledge that things are slowing and it'll just take longer to get more, you know, juice out of the squeeze, as it were, right? And so that's sort of a, the natural evolution that's been happening and everyone is now pointing to it or at least acknowledging that that some aspect of that is real. And now at the same time, this comes along and it calls into sort of more question. There's one other element that sort of, I think, is related to this.

which was the big news story last week as well, the Project Stargate, OpenAI and Invidia and Oracle, all coming together. And one of the more interesting elements of that was the fact that Microsoft is effectively pushing off the compute costs to Oracle and some of the other players in that situation. And, you know, there's all sorts of reasons,

you know, potentially why they're doing that, obviously given the interesting relationship between Open AI and Microsoft. But at the very highest level, again, if they're thinking that, you know, our KAPX is going to be, we've already stated, it's going to be $80 billion for the year. We don't want to add another several billion, you know, for this particular project. And why would they do that in part, probably because they're not necessarily sure that it makes sense to pay the billions upon billions to Open AI to keep trying to scale on the frontier

models and this is you know sort of in line with what deep seek just did right yeah it's interesting we're also talking about more uh andreason horowitz who sat out opening eyes last round and we were wondering on the friday show maybe they i heard that saw this coming yeah and it is interesting i mean you you put it pretty pretty uh perfectly in your story uh you say um big tech companies are the now the most largest and sorry you say big tech companies are now the largest and most well capitalized in world, which means that they have effectively all the money that they can put towards scaling and the hammer met the nail. But there's no point hammering the nail after it's already been put into place. And that's the point that can't be predicted, but is obvious once it's done.

The question is if deep seek just pointed to the nail already hammered. Effectively, did they just solve this? It's sort of like going up the scaling question in a similar way. an analog for the same thing, right? And going back to the history of compute, like, right? All these, you know, the powers that be tend to spend, at the time, tend to spend a ton of capital on the buildout of whatever the new technology happens to be. And, you know, there's obviously, we all benefit from it in the long run, but in the short run, you know, this segues into, I guess, you know,

what's potentially going on with Wall Street and what it means for these larger companies with regard to the spend. Yeah, and I just want to ask the question that you put in your newsletter just to you directly. Did they just point to the nail? Like, is it done? I mean, again, I don't want to, you know, caveat this out, but I do feel like it's, it's the exact question that everyone is sort of going to be scrambling to answer over this next week. And I think that it's not going to be as black and white as that for sure. but I do think, if I had to guess at a high level, I do think that there's some element to, yes, the nail is already sort of driven into the board and we're moving on to what the next steps are. It's not to say it's over and, you know, there's no innovation from here, but I think all of these things are in a way related, like that we've just been talking about.

And the fact that they're all coming together at the same time, I don't think it's a coincidence. I think it's because, like, yeah, we're at the point where we now need to move on to sort of the next phase of the AI revolution, as it were. Yeah, and let's get into the business. And I'm smiling here because you're making me think of we have Reid Hoffman on the show on Wednesday. And I interviewed him before R1 came out. And the first half of the conversation is just talking about all the billions of dollars that have been spent and when they're going to get an ROI. And, I mean, I'm still going to run the conversation, but there's going to be some context in there. Yeah, it's interesting knowing after the fact.

But it's also, I think Sinovsky brought this up too. And I was sort of looking into this more last week. You saw it was a smaller news item, but both Microsoft and Google had altered the way that they're basically bundling together AI within either the 365 suite and within the Google suite of apps. Because they're clearly still trying to figure out how exactly you make money off of all this spend. and what the right model is and how you spur on usage of it. And this just comes in and throws a grenade into that equation again. And this gets us to like some of like the real thorny business questions. So just to kick this off, I took a look at what all the big tech companies were doing pre-market.

So this will obviously change across the day. But I imagine they'll stay directionally kind of the same. Invidia down 10%, Microsoft down 4%. Google down 3%. meta down 2.6% SMP down 2%. So this is all based off of this deep seek reckoning or this deep seek realization. And let me just put the sort of question to you, I think about as pointedly as I can, which is that the AI industry up until this point, like all the numbers we're seeing within Wall Street, the trillion dollar market caps, the billions of investment, the billions that have been raised,

by companies like OpenAI and Anthropic, from companies like Microsoft and Amazon, right? So this is basically the whole game here. They have effectively been what's been driving the numbers. And the question is, can we, you know, basically Wall Street has been following that and saying we expect them to get a return based on those numbers. And in fact, a lot of this AI spend

was just a wealth transfer, I would say, from like meta-advertising to Lama, from Google search revenue to Gemini, from Microsoft Azure to OpenAI. So what happens here? Because, you know, basically if they, if a lot of the AI industry has been driven based off of subsidies coming from other businesses and doesn't need that type of spend anymore, like does the party end? So I think it's different for each company, probably Microsoft and Google are closest,

you know, aligned in terms of where they net out. And it's sort of interesting, you know, the numbers you just rattled off with where the stocks are at, that feels, you know, just like a very clear picture from Wall Street, what they think now, right? Like they think, NVIDIA is going to get hit fast because in this, in this doomsday scenario, because obviously they're the beneficiary from everyone, from all of those companies, all those other companies that you mentioned, Big Tech is pouring as much money as possible as they can. They can't get enough chips fast enough into NVIDs. And if they pause that, that obviously is bad news for Nvidia in the short term. Again, I think there's longer term stuff that that's different for Nvidia, which we can talk about,

but to just hit on the rest of this question right now, I think that Microsoft and Google, which are, as we just mentioned, you know, are trying to sort of figure out the right models for how to charge for AI. I think that this puts them in a really tricky situation if the underlying economics just totally changed overnight of what AI's underlying economic model should be.

And so they were, you know, moving around different pieces, trying to get to the right end state so that, yeah, they could ultimately prove to Wall Street. Like, look, we're adding, you know, X amount on top of what we were already doing, revenue-wise, thanks to AI. And a little bit, there's a little bit of weird obfuscation stuff going on there, right? It's like, well, it's bundled in now to 365. And so, you know, we don't necessarily need to tell you exactly what the uplift is, but you can just, you know, assume that it's a part of this because it's all baked in. And AI is like, you know, the new internet and blah, blah, blah. And so, you know, there's ways that they can finesse the messaging around that.

But that, you know, to your exact question, I do think that there's varying degrees of being worried, certainly within Google and Microsoft. Meta's more interesting because their open source. philosophy, open weight philosophy and model is so similar to what DeepSeek has done, right? And so the problem there, in my mind, at least, is again, they're spending whatever Zuckerberg just threw out, 65 million or whatnot, he said, you know, at the end of last week that they're going to spend on CAPEX. And so why are they spending that amount now if, you know, DeepSeek can do it for, you know, pennies on the dollar, if not even less than that? And so what is

that mean for their world. So in my view, high level, I think that meta is probably at a bit better position than the other ones just because at the end of the day, they do want, like, you know, their whole philosophy is to open sources, not for necessarily altruistic reasons, but because they know that it's historically helped them, help their business, you know, to open source these things. The question of if it's not them open sourcing it becomes pretty complicated, if someone else's, you know, you have to use someone else's models. But they can pull back spend. It feels

like a little bit easier than the other folks can. On the other end of the spectrum, OpenAI, like they're, you know, the entire business is sort of built around being at the frontier and they've done a great job with that. They're a little bit different than than Google and Microsoft in my mind just because they've done a good job getting mine share, both in terms of brand and product, right? Like JetTBT is number two in the opposite. story right now behind Deepseek, you know, for a reason. People are interested, it's a brand and they know it. And so what does it look like, though, if they're not the one sort of powering that models? I don't think that they would give up and, you know, go with Deep Seek's model

necessarily. But what does it mean if they're not sort of the only one or the main frontier, you know, model maker providing that? Like, so there's all sorts of interesting offshoots and ramifications of that. So, MG, there's like two views right now in terms of like what could happen with all the spending, right? One is Silicon Valley will continue to spend these billions and they might get, you know, incrementally better performance and stay slightly ahead of the open sources of the world, deepseeks of the world that can just emulate their models. The other side of it is that they continue to spend and then they basically hit AGI or like, you know what I'm saying? Like we've, if the performance increases that we've seen with such little, sorry, if the performance increases

that we've seen with such efficient use of capital from deep seat can be emulated, then imagine what you could do with a hundred times the amount of spend. So the models are about to become much more powerful in all these fantasies that people have about what they can do, many of which Demis and I spoke about last week, all of a sudden become feasible because the capital is there. So which side of this deal? And that's a nice thing to say in like a nice high-level mantra. And many of the leaders of these companies will be saying that today to sort of try to

Com Wall Street. But at the end of the day, you know, aside from sort of open AI, which obviously is again tied with Microsoft and now Oracle, but besides them, the rest of these are public companies. And Wall Street, you know, like it or not, they have a say sort of over what they're going to do. Like if they're going to get hammered, and this is something I've sort of been harping on for a while, not because I think that they were doing the wrong thing necessarily with the spend. But it's just obvious that like it always comes back around, right? where it's like, I equated it last year to when all the movie studios during COVID and TV studios were just balking up on streaming, right, and just spending as much money as possible

as they could in order to build up their streaming services. And Wall Street loved it at that time because, you know, Disney and everyone else was just gaining millions and millions of subscribers. And it seems like they had a path to take on Netflix and, you know, this was the future of the industry. It's still, by the way, the future of the industry. but Wall Street then all of a sudden turned on all that spend and decided like you need to cut like spend X amounts. You need to, you know, unfortunately cut the employee base and basically just become way more efficient while doing the same high level thing.

And it was, you know, always obvious that at some point they were going to do that to the tech companies as well with regard to AI spend. And so again, they can all have the right mentality about like this is the future and say the right things that this is the future and this spend is important. And I don't disagree with any of that, but still, they have to answer to Wall Street, it's, you know, to some degree. Maybe Zuckerberg less so because he, you know, controls the, controls the company so strongly. But, like, certainly Microsoft and Google, to a lesser extent, are going to have to answer for a lot of that spend. And this is the first real, real test.

Meta had some of it, right? Like, there was some backlash last year around their spend and certainly back, dating back to the, you know, VR and AR and XR spend. And so they had the answer for some of that. And Zuckerberg did, right? And he got rewarded for it after the fact. And that's like the game they're playing here. They know that if they cut spend because Wall Street doesn't like to see all the AI spend,

they'll get rewarded in the form of the stock going up and then all the ramifications from that. And so it's natural that that is going to play out that way. And so I think the narrative then shifts to other levels of not necessarily obfuscation. but other ways of framing it, it's like, okay, we agree that we shouldn't spend tens of billions of dollars on NVIDIA server farms, but we need to build out, you know, our in-person AI robotics arms, right, in order to keep these models and keep sort of the next phase going as we march towards AGI and yada.

So markets just open, NVIDIA opens up down 11%. So still above $3 trillion, so it's not like the AI revolution. is over, but down 11%. So just a cool, you know, a couple hundred billion dollars shaved off the market cap in a morning. Let me talk to you a little bit about what these companies are saying back to Wall Street or actually talking to Wall Street about to allow them to keep spending. So Saindella is doing his tweets.

He says, he's talking about Jevon's Paradox. He says, Jevon's Paradox strikes again. As AI gets more efficient and accessible, we will see it to skyrocket, turning it into a commodity. We just can't get enough. of. Let's say that. I don't know if you saw this last night, Gary Tan, you know, the president of YC treated the same thing. And so I'm like, is this coordinated or I mean, it is, you know, it's a group text going on. Yeah, there's a group text maybe going on where it's like, this is the answer.

And it's not a, it's not like a totally BS, you know, answer to it. But there's much more nuance and context that's sort of required to get to, you know, that being the excuse for this. so let me just basically talk about the elephant in the room that's been hanging over this full conversation and will be sort of like the spoken or unspoken part of this discussion as it goes forward this week which is that let's say the cost of intelligence goes down to zero right so that's what everybody is basically aiming for it's one of open a i's stated goals to make intelligence you know close to free as possible they don't really make a lot of money selling off their API or they even maybe might lose um we need to see a i application Like we need to see an economy that takes use of this technology that is so impressive, right? Like you look at the chain of thought even in deep seek and you're just like, how is a computer, you know, quote unquote, thinking through this stuff. But the economy needs to take hold of this powerful technology and make use of it and put it into play for really meaningful economic use. Whether or not the deep seek thing existed, right?

like billions of dollars of economic or trillions of dollars of economic value needed to be created from this generative AI moment. And what do we have now? We have OpenAI who has chat GPT with 300 million users, which is okay, but still losing billions a year to run that thing. Maybe they'll be able to be more efficient and make those couple billion a year from it. We have some enterprises putting this into play, but every enterprise I speak with, There's a couple use cases, cool use cases here or there, but mostly what you see is

proof of concepts. And many of those proof of concepts aren't going out the door. So don't we need to see one way or the other AI applications, whether that's standalone or integrated within business software, that start to prove the real value of this technology that we just haven't seen to date? I mean, the answer is yes, of course. The reality, though, is, you know, maybe this. this confluence of events right now is going to help that because it's sort of just

is forcing a fundamental rethinking of a lot of what we've just been not going through the motions but we've been on this path right to scaling as we were talking about and that you know even right like sam altman has said like they see line of sight now to aGI they just have to, you know, just dot, dot, dot, dot, underpants and then profit from there. Better get that now. But they say they have line of sight, right, to know what they need to do, and it's just a matter of execution and sort of, you know, getting everything aligned in order to do that. And if this moment, with Deep Seek being the, you know, the biggest catalyst thus far of it,

if it doesn't cause the entire industry to sort of rethink that, and at the same time, to your point, like, you know, asking about, does that sort of drive us to move on from, just like this this nonstop scaling of frontier models that is awesome technology but unclear how it works in from a practical standpoint do we start to yeah distill this you know for for lack of better phrase down to actual products and you know when i when i think about that that leads back to like uh whenever it was six months ago seven months ago when apple did their apple intelligence stuff which you and i talked about right and it's like everyone jumps on Apple, and there was another news cycle, I think, this past week because, you know,

Siri can't, can't correctly answer who won previous Super Bowls, which seems utterly ridiculous. Amazing. But Apple's mentality from the get-go with launching Apple intelligence has clearly been, we need the, we, for lack of a better phrase, don't necessarily care so much about, yeah, the frontier of the vanguard of this technology. we care about the day-to-day usage of it, right? And, you know, they have a few things that are sort of front-end facing that haven't really worked that try to use AI like the emoji creator and things like that.

But most of it is just baking it into their products. And that's what we've seen too with obviously what we talked about with Microsoft and Google. They all have, you know, their own like some have video generation, some have some other of their own standalone products. For the most part, they're just going to be baked in. But, you know, to what we were talking about earlier, none of that, is really the promise it felt like right of what this larger movement was going to be and everyone's waiting for you know not not necessarily aGI right now but they just want some other

forward facing user facing um version of AI that that can be good and chatybt has been the closest that we've gotten to that and maybe some of these video products you know end up being the next the next phase of that. But I think that you're right, that this, that ultimately you have to get to something that comes of this, that really sort of moves, moves all sorts of needles. And again, I wonder if this news cycle and just pause now doesn't lead to more of that. I hope that that's the case. Yeah.

And I would say the Apple intelligence is almost the perfect example of the problem that I'm pointing toward, which is that we have this technology that's so promising. and yet even Apple cannot implement it successfully. And that might, I mean, obviously it says something about Apple, but it might say something about the technology as well. Yeah. And, you know, as with everything, like with everything in technology, I think about, you know, dating back to my reporting days and whatnot.

It's just like having seen so much in a few different cycles now, are we too early still, right? Like everyone has been talking about and believing that like this is the moment where this is like really happening and this is great. But I do think that if you took a step back, you might wonder if we're not still doing this too early, you know, and trying, and all of these companies are not raising way too much money when the timing is just not right for exactly what you're, you know, trying to ask the question about. Like, how do you, how do you turn these into products and how do you ultimately turn this into a business that returns the capital that was spent on it? Now, no company would admit that right now. But, you know, hindsight will only prove one way or another, whether that's, you know, the case. And I think everyone still remains super optimistic that now is the right time and you want to keep your foot on the gas. But again, this deep seek stuff sort of causes a pause and a natural reexamination of just how much money to spend and what you should be focused on.

Let me ask you to put your investor hat on for a moment. Are there startups out there that would exist today that don't exist because effectively buying compute from the APIs or running Lama is cost prohibitive, but they would exist if intelligence was zero. And that's effectively what DeepSeek is going to put to the test. Yeah, that's really interesting. I don't, I don't, I don't, I don't, I don't want to just try to come up with something off the top of my head. Not that I know of, but I do think at a high level that your question is a really interesting one. And if, if this is going to be truly transformational, Deep Seek as a whole, it will lead to something like that, right? Like a bunch of companies coming out and not just, yeah, because it's not just

the technical aspect. It's not just driving down costs because that seems like it's sort of going to happen as a result of that, which is great. But does this actually yield new companies that couldn't have existed beforehand? And I don't know, like, I can't think of any off the top of my head, but that's also why I'm not a startup founder. And, you know, hopefully there are startups out there that are that are going to latch on to this. But something tells me that the answer is no. And the reason is is because investors have been dying to throw money at AI companies and I've been willing to lose a lot of money if the idea is promising enough.

And I don't know. We haven't seen a wave of AI startups hit. At least there have been many. But, you know, they're not like, it's not like the beginning of the mobile era where there was like a new consumer startup every day. It just isn't happening that way. In fact, most of the action is enterprise. One other just wrinkle and layer of that, which I feel like has been overshadowed in all of the recent news.

But we talked about it and talked about a lot last year. But as the regulatory regime is changing now, if M&A sort of doesn't pick up with regard to exactly the type of companies you're talking about, right? Like they have great teams. They're working with, you know, this technology and they clearly know how to do. things with it, but they haven't gotten the product right. They haven't gotten the business right. And so they're scooped up by the, you know, the metas, the Googles, Microsoft's the open AIs of the world. And that, you know, in and of itself won't be that interesting other than those companies getting good talent, perhaps. But if it just reignites sort of, you know, a passion within

really early stage startup founders to keep reaccelerate sort of going after new problems, right? Right. Like, I do feel like there was a bit of a chilling effect the past year because M&A had, you know, basically been shut off that sort of kept people staying at Google and staying at meta and staying at Open AI, not forming new startups as they might have in years past because they knew that, you know, there was there was the potential. Obviously, pie in the sky, they want to build a big company. But there was also the potential, frankly, right, to like, you know, sell, build something that's big enough to sell for. multi hundreds, millions of dollars, if not billions of dollars, to some of these other companies. And so, you know, that might come into play with some of this. All right. Let's put a bow in this conversation. You say the real problem is that it won't be so simple to simply pull back spend. Beyond a lot of it already being committing, being committed, there's obviously still a very real risk that deep seek is just a blip on the radar and not the bomb that blows up everything. What are we looking at? What are we looking at over the next couple

months when it comes to the aftermath of this earthquake to go back to our original question. And so that's just a call out to the, you know, the obvious thing that everyone likes to overreact, obviously, to, you know, big news stories and big news cycles. And again, as we've been talking about, like, this is legitimate, but how legitimate is it? Like, right? So we'll even see potentially play out over the course of today in the stock market. Like, do they start to get nerves calmed a bit by yeah this talk of like well actually this isn't so bad for invidia because while it hurts their their immediate um it could potentially hurt their immediate um money coming in the door in the longer run you know it's it's again jvon's paradox stuff where it's like yeah it's it's going

to raise raise all boats um as as this just permeates everything and so they need chips and yada yada and so that could help um but yeah i mean i think that it won't be so easy also for as i noted for all these companies to pull back spend because they've already committed to buying x number of of h 200 chips and and then uh soon enough we'll get the next iteration you know announced uh down the road and so um all these supercomputer uh mega clusters of data centers that are being built right now they're just not going to you know put the brakes on all of that because there's a risk they're all playing in the same game, right? And if one of them pauses, maybe they get a short-term Wall Street, you know, pat on the back.

But if they're wrong, that's like catastrophic. And that's, you know, it's like a fire firing the CEO type offense. You know, if this is just, you know, even a blip on the radar obviously undersells it a bit. But if this is not ultimately like a real fundamental sea change situation and is more just like a step on on the road. they might still want to keep their foot on the gas. Yeah, it's going to be very interesting to watch. The website is spyglass.org, the piece, A.I.F. Finds Away. Joined, of course, by M.G. Great to see you again. Thanks for coming on the show.

Thanks for having me, guys. All right, everybody. Thank you for listening. We're back on Wednesday with my interview with Reed Hoffman. Obviously, a little different now, but maybe, as M.G. puts it out. Maybe we shouldn't be overreacting too much. So looking forward to speaking with you then, and we'll see you next time on Big Technology, Thank you.