Allegra Stratton (2:26)

Yeah, it was powerfully called a Sputnik moment. So the analogy there being 1957 when the the US watched Russia put a satellite in space and realized that they'd been outclassed and that Sputnik moment has become this kind of shorthand for sort of this huge game changer in technology, but also geopolitics.

Francine Lacqua (2:43)

And remember, I mean, Deep Seek is a Chinese artificial intelligence startup that's just over one year old. It's quite incredible. So there's consternation.

Allegra Stratton (2:51)

It's an amazing David and Goliath story, isn't it?

Canva Advertiser / City of London Announcer (2:56)

Welcome to the City of London.

Parmi Olson (2:58)

The City of the City.

Allegra Stratton (2:59)

The City of London.

Francine Lacqua (3:00)

The next station is Bank. Please mind the gap between the train and the platform.

Canva Advertiser / City of London Announcer (3:07)

The financial heart of the country.

Allegra Stratton (3:13)

The City.

Hiscox Insurance / Amazon Business / GSK / Verizon Business Advertisers (3:14)

The City.

Allegra Stratton (3:15)

Welcome to in the City.

Parmi Olson (3:18)

Stand clear of the doors, please.

Francine Lacqua (3:22)

I'm Francine Lacqua.

Allegra Stratton (3:23)

I'm Allegra Stratton.

Francine Lacqua (3:24)

So this week we're going to unpack the Deep Seq news, what it means for the AI race and of course, how the uk, with its own AI ambitions, fits into all of this. And with us, Parmi Olson, Bloomberg opinion columnist covering technology and author of the great book Supremacy AI chatgpt and the Race that Will Change the World, which also won the Financial Times and Schroeder's Business Book of the year award in 2024. So Parmi, first of all, congratulations.

Parmi Olson (3:47)

Thank you.

Francine Lacqua (3:48)

And you kind of saw this coming. I mean, in this AI race there was, I guess, an understanding that the AI in the US was so far along and then massive curveball this week. China's not that far behind. They can do the same on the cheap.

Parmi Olson (4:01)

Yeah, I think what happened with Deep Seek really just shows that nothing is set in stone with AI and we can still have surprises come out of the woodwork. The timing is really interesting because just a few weeks ago there was a lot in the news about how AI companies from OpenAI, even the head of Google, DeepMind were all saying publicly, actually the formula that we've been using based on so called scaling laws is starting to have diminishing returns. We're not developing as quickly our AI models as we did before. It's slowing down. If you throw a whole lot of very powerful chips and put them together with a whole lot of data and a whole lot of energy, you get a really powerful AI model. And the more you add to that mix, the more powerful the AI model would get. But that just wasn't the case anymore. And what Deepseek has shown is that actually there is a workaround, there is a way to build a model that is, according to the people who have vetted their model, as good as ChatGPT, but for a fraction of the cost. A fraction of the chips and the resources. So in one way that's very distressing for people in Silicon Valley because it makes it look like all the money they've poured into AI has just been a big old waste. On the other hand, it shows well, there's some hope because we were all worried about these scaling laws just a few weeks ago.

Allegra Stratton (5:29)

Parma, you used that word distressing for your contacts in Silicon Valley. What is going on right now? What are they having huge meetings saying everything we thought we understood, we have to revise the numbers, we have to revise.

Parmi Olson (5:40)

Well, I think this week is earnings week and so the likes of Mark Zuckerberg and Satya Nadella, all the heads of the big tech firms are going to be talking about how much they're spending on AI. And the CapEx has been huge every quarter for the last few quarters and investors have generally been okay with that because there's been this vision. Now, I think with Deep Seek showing well, actually maybe you don't need all that much Capex spending. I think that's where a lot of the panic is, how they're going to recommunicate that to their investors.

Francine Lacqua (6:13)

Parmi, how is it possible that the Chinese have such an advantage to do the same technology on the cheap?

Parmi Olson (6:19)

Well, it's funny you use the word advantage because they started with a disadvantage which was sanctions from the US they weren't able to get the high end chips and the same level of high end chips from Nvidia known as GPUs that Microsoft and Google and OpenAI could get. So there's this kind of prevailing theory in technology and in innovation that actually when you limit yourself it forces you to be more creative, to think out of the box. And that seems to be just what's happened in China. There wasn't really an advantage.

Francine Lacqua (6:54)

Parmi, how much do we know about Deep SEQ and the fact that it, I guess it's not government backed or is it, I mean could they be getting a lot more funds than we think? And I've always been under the impression that the Chinese have so much more data available because of the number of people living in the country and because the government just takes the data and so it's easier in some way to basically use this to model some of the AI.

Parmi Olson (7:20)

Yes, you could say that. But at the same time, Chinese companies for the last decade have not been very strong with language models. They've been great at recognizing faces because of the surveillance mandate by the government. So you've got lots of face recognition Companies, the best in the world are in China with Deepseek. They've said that the budget they used for this new model they built was $6 million. That's about what a Google executive gets paid every year. You know, this is like a realm.

Allegra Stratton (7:53)

Yesterday, such a killer fact.

Parmi Olson (7:55)

But I mean, it literally is a fraction of what Google and OpenAI are paying for their own equivalent models. It's interesting you mentioned some of these kind of theories that are floating around, like, is it possible? How could they. I think people are just stumped and stunned and we're grasping for answ. Even experts that I've been speaking to in the last 24 hours, they don't really want to talk about it because they're just, they don't really have predictions or. It's hard to parse through all the noise right now because this has been such a shock to the system. But those who have been working with open source AI say actually we've known this is coming. Open source models have been getting better. Meta Mark Zuckerberg's company, his approach to AI has been to open source. It strictly, it's not, some people say, is not really open source, but essentially it's free. And those models have been getting as good as OpenAI's models. So you could sort of see this coming. I think people just didn't expect it to come from China.

Allegra Stratton (8:54)

Where does this all shake down? I mean, yesterday, Nvidia, the crash was bigger than the entire amount. Did you write this? Somebody, Somebody wrote the crap. The amount they lost yesterday was bigger than all of the money that Donald Trump managed to get all the AI companies to put in last week. 500 billion or whatever. Right, right. So, so where, where does this terms of. If you are a Silicon Valley company, you've got your earnings this week, they'll obviously be looking to get through that. But the other side in six months and so on. We hear what you're saying about a lot of the experts you speak to feeling it's way too early to say, but it feels like it's hard to see that those hugely, those massive numbers that Silicon Valley kick around are sustainable medium term.

Parmi Olson (9:36)

And that's actually what the critics of the AI industry have been saying for some time, that we're in a bubble, that there's way too much money being spent on this. And so in some ways, what happened to the NASDAQ yesterday, you could call it a healthy correction to the market. The hype has been way overblown, but at the same time, an optimistic way of Looking at the situation if you're in Silicon Valley is, well, hang on, deepseek is open source. They are literally posting their secret sauce on the Internet. So there's no reason why Google and Facebook can't copy their methods and use them with all that extra infrastructure and chipset.

Chase Sapphire Reserve Business Advertiser (10:18)

Sure.

Allegra Stratton (10:18)

But then there's no reason why, if you're a shareholder, you don't say, given it's on the Internet, why does it cost you this much?

Parmi Olson (10:24)

Yeah, absolutely. So I think the one thing that Meta and Google and OpenAI can say is, well, we're going to make something even better now because we've got this new information, we're going to learn what we can from Deep Seek. I think the next few months for them are just going to be really difficult, especially this week.

Francine Lacqua (10:42)

Parmesan again, it's my understanding. I keep on saying it's my understanding because I'm not sure what we're certain about anymore or not. Given the week that we've had, I think that's fair.

Parmi Olson (10:50)

But it's my.

Francine Lacqua (10:50)

It was my understanding that, you know, one of the reasons why it was so difficult for the Chinese to agree to sell TikTok in the US was that they didn't want to give the algorithm to anyone who would buy it. So why would Deep Sea have an open. I mean, is it really open or are they keeping parts for themselves?

Parmi Olson (11:08)

No, as far as I know, it's open, it's out there. Yeah. You can use this accessories.

Francine Lacqua (11:12)

So this changes everything for.

Parmi Olson (11:14)

For everyone, potentially. Yeah. And if you look at some of the comments made by the founder and the CE of Deep Seat, he's talked about wanting to move into this more collaborative way of working with the west and how in the past they in China would have to wait for innovation to come from Silicon Valley and then ride on the coattails and they want to kind of bring things into a new kind of era. And that might feel kind of awkward in the west, but you have to remember that the culture in China is generally more collaborative. I can't see inside his head and what the true motivation is, but I think there is something genuine about that.

Allegra Stratton (11:53)

Let's look at the politics for a second in America. Obviously Donald Trump stood up a week ago and said, gonna usher in this golden age. What is, what is, what will, what will they, in as much as anyone can think into the Trump White House, but what will they be thinking right now? Will they be saying, right, let's go further, let's go faster, of course they will. You're nodding your head?

Parmi Olson (12:12)

I think so. I think Trump is the kind of person that doubles down and his mandate for a been America first and let's make sure we're ahead of China. And that's, I think, what has excited everyone in Silicon Valley about the potential loosening of regulations, even the potential lighter touch regulation on the part of antitrust. There's been this looming threat of a breakup of Google. But if a breakup of Google makes America America's industrial strategy or just American companies weaker in the face of China, maybe that won't happen after all. So I just, I can't see, honestly, I'm not sure what Donald Trump's policy is going to be here. I'm sure he'll continue to push forward with Stargate, where Elon Musk fits in in all of this as someone who is a massive AI safety advocate, but who still has not spoken up about that, seemingly to Trump, that he's still a wild card and all that.

Francine Lacqua (13:09)

When do people start paying for these services? Does it only become effective when I start paying? I don't know, 599 per week to use.

Parmi Olson (13:17)

I think this is kind of like the big question mark that is hanging over big tech right now is will they truly be able to monetize generative AI? Because, yes, OpenAI made something like 5 billion in revenue, but it also lost 7 billion. So even it is struggling to make a profit and its path to profit is unclear. And yes, people pay for ChatGPT and businesses pay for ChatGPT, but a lot of businesses have experimented with it and just kind of put it aside saying, we don't think it's really going to help us. So it's still an open question. And I think, I mean, in a way, a deep seek maybe will encourage more experimentation by businesses because there's not this big initial outlay. It's like, well, we'll just try it. It's free, no big cost there. But, yeah, I don't think anyone's really answered that yet. And that's why people say we're in a bubble and if they don't answer it, then that bubble will pop.

Francine Lacqua (14:15)

I loved your book. Oh, thank you. It talks about supremacy and this fight between individuals that are very well known and trying to get to the top. So, you know, a lot of these players, you know, up close, who do you think is more stressed this week?

Parmi Olson (14:29)

Oh, 100%. The most stressed person right now is Sam Altman because his entire company's business model is based on selling a foundation model. And he doesn't have anything else. You know, at least if you're Google, you have all the other stuff in your business. If you're Microsoft, you've got all the other enterprise stuff that you're doing and the cloud business. Same with Amazon, same with Apple. But OpenAI also anthropic. But OpenAI, especially if they're being completely undercut by another firm in China. I think the thing that really freaked people out this weekend wasn't just the fact that the deep seat model was so good, but that it was so popular. It's number one on the App Store, on the iPhone App Store, it's number five on Google Play. So I think that's a big problem for OpenAI.

Allegra Stratton (15:18)

Moving on from the agony of these tech bros just talking about AI safety. And I know I always feel slightly parochal when we bring it back to the uk, but we're sitting in the city of London and for, you know, Keir Starmer has made a play, not as recently as Donald Trump, but, you know, a few weeks back, that he still wants the uk, like Rishi Sunak wants the UK to be this kind of beacon of AI safety and that there's some kind of market niche that can be carved out. From my perspective, it looks harder and harder to see how that now works where you've got the two big countries racing ahead, or does it give us reasons to be optimistic?

Parmi Olson (15:57)

Okay, so I think first of all, I thought Keir Starmer's AI Opportunities Action plan from the other week was way too broad and he should have narrowed it down to just a few things. It was very ambitious. I think he should really play up the UK's role as a safety regulator, because if you, you look at the number of people who work for the government here on AI safety, There's like about 150 people working at the AI Safety Institute in London. That's more than the number of people working in Brussels on the same thing where they're building this AI act. So there's a lot of expertise, partly because we have DeepMind here, there's just a lot of talent and some of those people are crossing over and helping the civil service. And I think the uk, when it comes to how Britain has regulated tech in the past, it's been very smart, it's been very practical, pragmatic, and it's put out laws that America, American states have copied. You know, it's quick, it's nimble. So I have a lot of faith in what the UK can do there when it comes to innovation. There's just a ton of startups here. Again, we've got DeepMind. So I think when you have clear regulations in place, and by the way, I think the British government, it is planning to write up its own legislation on AI safety. It just, they haven't done it yet.

Allegra Stratton (17:11)

But just to be clear, how does that work when you've got two countries going for it, going as hard as they can with, as you said earlier, America making a play of less regulation.

Parmi Olson (17:21)

But I like to push back on that because America innovates in Britain and Europe regulate America. Do they really innovate? I mean, look at where they are today. Look who just out innovated them. It's China. If you look at Facebook, they're not innovative. They copy companies, they buy companies. That's what they've been doing to AI startups in Silicon Valley. All the most promising ones have been aqua hired by the big guns. And so I think actually Deepseek in a way points to some real optimism. You said that America's really gunning for it because they have so much money and yeah, maybe we don't have those resources here, but Deepseek has shown you don't need that. You just need innovation. You need smart ideas. You need to think out of the box, think creatively and open source might be.

Allegra Stratton (18:09)

That's a really powerful point that there's a kind of London model or a UK model that Deepseak is charting out. And also the other important point to contradict myself is that the UK point or position has always been that we will regulate less than the eu. And so being that kind of more, more palatable version and the big focus.

Parmi Olson (18:27)

On growth, you've heard that from the cma, you've heard that from regulators here. It's all about growth and let's support the industry here. Pardon me.

Allegra Stratton (18:34)

Thank you so much.

Parmi Olson (18:34)

Oh, my pleasure. Thank you for having me.

Allegra Stratton (18:39)

Thanks for listening to this episode of in the City from Bloomberg. It was hosted by me, Allegra Stratton and Francine Lacqua. It was produced by Sama Saadi and Moses Andam. Sound design by Blake Maples. Special thanks to Pami Olson. Please subscribe, rate and review Wherever you listen to podcasts.

Hiscox Insurance / Amazon Business / GSK / Verizon Business Advertisers (19:04)

How can you free your team from time consuming office tasks? Amazon Business empowers leaders to not only streamline purchasing, but better support their teams. Smart business buying tools enable buyers to find and purchase items fast so they can focus on strategy and growth. It's time to free up your teams and focus on your future. Learn more about the technology, insights and Support available at amazonbusiness.com@GSK, our focus is.

GSK Advertiser (19:32)

On doing the right thing for patients. We believe they should be free to focus on doing what they love, especially when they're living with a disease like cancer. That's why we focus where we can make the biggest difference matching the right treatment with the right patient. At gsk, we're pioneering advanced technologies like antibody drug conjugates that precisely target and attack cancer cells. By uniting science, technology and talent, we work tirelessly to stay ahead of cancer together. Visit gsk.com to discover more. As a contractor, I don't pay for materials I don't use, so why would I pay for stuff I don't need in my mobile plan? That's why New My biz plan from Verizon Business is so perfect. Now I can choose exactly what I want and I only pay for what.

Verizon Business Advertiser (20:18)

I need right now with my biz plan. Get our best price as low as 25 a line. Visit verizon.combusiness to get started today. New lines only. Price per month with five plus lines includes auto pay and paper free billing and promotional discount, taxes fees, economic adjustment charge applicable. Add ons prices and terms apply. Guarantee applies to base monthly rate and stated discounts only. Add on prices. Additional offers in November 17, 2025.