Transcription
That's the main issue. The main issue isn't so much, oh, is this company going to be profitable in the next year? Or like the the main issue is, is it going to need a bailout from its investors? And if so, how can it go asking for more if it hasn't sort of got a plan?
Hi, I'm Fergus McY and welcome to the tech report. Joining me today for her usual Wednesday slot on the show, Natasha Bernal, tech journalist. Natasha, thanks for coming back on. How are you?
Thank you. Thanks for having me. Yes, fine, thank you.
Um, now I want to start with asking you the question, is open AI falling apart in real time? Now, those aren't my words. I ask because that is the view of George Noble. Um, for people who might not know who he is, he's kind of an experienced investor and sort of hedge fund manager. I think he can be quite good for a quote, I will say. We'll put that at the top. but he uh he sort of over the weekend launched kind of a scathing criticism of OpenAI in a post on X and it was really really long so I'm not going to read the whole thing but it essentially centers around the fact that the company lost 12 billion dollars in a single quarter last year and he then pointed to kind of like the user plateau and drop off from November 2025 onwards more people are switching to Gemini there's also this potential lawsuit from Elon Musk on the horizon which we might come to in a little bit and then he also mentions this analysis from Deutsche bank which said that the company could see over $140 billion of negative cash flow before turning a profit in 2029. Summed it all up in a neat bow by essentially saying that no startup in history has operated with losses of anything approaching that scale ever before. Are the signs there with what he says or is this kind of concern misplaced?
I think he's he's kind of pulled together a few threads that certainly do spell a sort of very troubling picture for open AI and you know a lot of people have um said that you know this is the year that will either make or break open AAI and it does feel like the faith is faltering for a lot of people um who who have been watching from afar. I guess you know he much like anyone else um doesn't have the inside uh track of the company. You know he might say that they've you know wasted or they've lost 12 billion. you know, the company would say it's investing that money um and that it's not pretending that it's going to reach profitability this year. Um I thought, you know, his use uh you know, George Noble, he spent so long in asset management. He he knows what he's talking about. His his like all caps locked um letters saying falling apart in real time, you know. Um was it was it was I think a a grabby thing, you know. He wants people to pay attention. He is he's expressing uh some very valid concerns. You know, he he's saying, you know, this is my this is my experience. I've been watching companies implode for decades and this one has all the warning signs, all the hallmarks of it. Um he he basically he he believes that, you know, the industry's promise of scaling up operations to meet the demand um and that the cost associated to that is is too large. Um, and one of the things he pointed out was that, you know, he said, and I'm quoting him here, here's the big math problem nobody wants to discuss. It's going to cost five times the energy and money to make these models two times better. And now, you know, on this program, we've we've talked before about the the uh the challenge of scaling up and the energy required and the data centers required. And at the moment, everyone's throwing everything but the kitchen sink at it, saying, "Yeah, we'll build all of the capacity. We'll do it all. We'll, you know, put data centers wherever they need to be put." Um and his his point is is very valid in the sense that you know we haven't found a way to um reduce the amount of energy required by these models to make them viable and they still haven't yet proven themselves to be such an incredible game changer that they've changed our lives and the world around us. So you know very very valid points in here. Obviously, what we're missing is the visibility of the actual figures from the company, you know, and we don't, you know, he's citing some November figures of user drop offs, you know, that coincides with Sam Alman's code red in December to people saying we need to focus just on our on our core product chat GPT. Forget about everything else. We don't need to think about anything else. We need to maintain our market, you know, our market competitiveness here. Um, we don't have the figures for January. We don't have figures in real time. Open AAI is not, you know, publicly listed company that has to release all these numbers or answer publicly to like a load of analysts. It does have, you know, certain calls and investor calls and that sort of stuff, but it doesn't have that sort of level of scrutiny that other businesses such as Google or Microsoft who are also playing in that play in that space do. So, I guess that's that's what we're missing here. you know, we've got a lot of people pointing at things that are obvious problems for the business, but it doesn't necessarily mean that OpenAI is going to um collapse tomorrow. And I think that's that's something that some people might interpret reading reading his words.
One thing that does interest me, you said it's kind of like a make or break year. I found the Gemini particularly interesting because I think it's something that like quite a lot of workplaces do. I think I believe my own we use Gemini. Um, and I know ChatGP's global traffic share did fall from around 80% to kind of more like 65 at the tail end of last year. I guess like you said, we don't know the figures, but I guess if you are a company that hasn't yet turned a profit and is operating at a loss, the one thing you do not want is your best competitor then making strides up on you.
Yes, absolutely. And this is a very volatile environment. And I think what we saw with Gemini and you know the latest launch in December was exactly that you know chat GBT has had first move advantage open AI you know came in with a bang you know made this whole thing I suppose happen you know it's it's it's you know it led the charge for LLMs being used in in the way that they're being used today. However, that doesn't mean that they can keep the crown. And that was the proof that we saw last month where Gemini brought out a product that in general got extremely good reviews from people on the market saying, you know, this is a really great great product. We really like it. And there isn't that sort of tackiness that a stickiness that um a user base that is super loyal would have that doesn't exist at the moment within the AI community. And so, you know, just the same as a user would type in um, you know, chat GPT, can I ask you a question? They don't really care. The answers aren't going to be so wildly different that they can't ask Gemini or Llama or any other of of of the LLMs that are out there available to people um to use, you know, that it means that they have to do that they're so affiliated to GPT that they must use OpenAI's products. And that's the big challenge at the moment that they really are on a knife edge. All these businesses where they're going, we need to make sure that our LLM, which is comparable to others in the market, can offer people that little something extra, whether it's accuracy, whether it's better information, more reliable information, or fewer hallucinations to keep them coming or they simply won't, right? And I I think, you know, the big challenge for OpenAI alongside that is making money from its user base. And you know they're convinced that the way to do this is twofold. One is to get people to pay for subscriptions, which that does create that sort of um you know, permanence, right? If you've invested in a you know, annual chat GPT pro, you know, account, you're not going to necessarily ditch it for the next thing that comes along, unless that's good, you know, which case again, you know, um things can be cancelled. But um yeah, you ha you have that and and obviously the corporate accounts side of things and then it's the ads right which is something that you know is it's at the moment testing and we've seen a few examples that have really angered a load of people um with what that looks like.
From the outside looking in that kind of seems I know to me like a tacit emission that we are not making enough money. Well, no. I mean, but I guess I guess that the question is is whether that's a problem, right? And this is what it always comes down to. And it's, you know, you got some you got businesses that are growing so quickly at scale that to ask it to also be profitable at the same time is probably unreasonable. But the question is, how long can that runway last before not only are you not profitable, but you're in a situation where I think it was the New York Times um we had quite recently over the the last week Sebastian Malib who's a senior fellow at the think tank council on foreign relations where he said, "I'm convinced that OpenAI is going to run out of money within the next 18 months if it's not careful." And that's the that's the main issue. The main issue isn't so much, oh, is this company going to be profitable in the next year? Or like the the main issue is, is it going to need a bailout from its investors? And if so, how can it go asking for more if it hasn't sort of got a plan to a kind of make itself buoyant and continue to invest at the scale it needs to invest to stay competitive, but also b have a plan for profitability, which is super hard in this environment. you're in a situation where as George Noble was pointing out, you need infinite amount of resources that they clearly can't afford to make its LLM twice as good, you know, where are they going to get that money, right? And that's that's the big question that I think a lot of people have where it's like you know you you have a lot of contra deals with especially with Nvidia you know where it's like we'll use your chips in exchange for some you know and all those things do help because it's not material money changing hands but they're still spending a lot to try to maintain their comp competi contiveness sorry and so um that's that's a danger for sure for open AI
uh in terms of that money just before we started recording uh you said you'd seen something just a minute ago that they've got another funding round.
Yeah, that's right. So, there's a story in the Wall Street Journal this morning um saying that Soft Bank uh which is a major investor in OpenAI um they've invested 22.5 billion in December. Um their stake has grown to 11% that they they might invest some more. So they're they're in talks according to the Wall Street Journal to invest up to30 billion more pounds in OpenAI which would value the company 830 billion dollars if it succe if it succeeds in in raising the full amount. Um so yeah I mean it's that's that's really good news obviously for OpenAI that you've got a big investor um that is willing to put more cash towards um its goal. Um obviously you know and we've spoken about this before on this program um all of these companies are intertwined in some way or another. So yeah so they've been working on Stargates. So Soft Bank has invested in Nvidia has invested in OpenAI OpenAI has invested in Nvidia. NVIDIA's invested in OpenAI. Again, it's it's a tightening of those bonds that people pointed out time and time again are problematic because, you know, should one of these businesses which are affiliated fail or have a problem or not kind of um you know, keep the faith of the investors, then the rest of them would see some sort of effect. And I guess that that's that's just another example of things that are going on where it's like you're double dipping, right? I mean, you're kind of double dipping where it's like, you know, and I I I know that it, you know, I'm I'm not that old, but I remember the days the good old days of of Wei work and the amount of money that that company got out of SoftBank and it was again and again and again like, "Please Soft Bank, may I have some more?" And SoftBank eventually was like, "Oh my gosh, uh what are you doing?" You know, so even even the patience of SoftBank, which seems infinite in its in its nature, has a history of being very very patient, there is an end to it. And so I guess um yeah, I I I want to know what they're getting in exchange of this, what kind of assurances they might be getting being like, you know, South Bank, this is what we plan to do with your with your 30 billion, like this is this is what we'd like to to do with it. And I yeah, I'm intrigued. I'd love to be a fly on the wall for that.
I I want you to humor me slightly for a second because I saw a uh I saw another post on X which kind of encapsulated sort of my feelings about AI and the so-called bubble. I'm going to read the whole thing to you. It's from uh the New York Times tech columnist Kevin Roose. I'm not sure if you come across Kevin's work. He I think he's based in San Francisco and he was saying that he follows AI adoption pretty closely and I've never seen such a yawning inside outside gap. People in front San Francisco they're putting multi- aent cloud swarms in charge of their life. Consulting chat bots before every decision. Why heading to a degree only sci-fi writers dare to imagine? Meanwhile, people elsewhere are still trying to get approval to use copilot in teams if they're using it at all. It's impossible. The early adopter bubble uh I'm in has always been this intense, but there seems to be a cultural takeoff happening in addition to the technical one. Not ideal. I mean that I was I was like, wow. Yeah, that's kind of exactly how I feel about it. How do you I mean bridge that gap and that disconnect between what's happening in Silicon Valley and the rest of the world.
It's really it's difficult and I've called this sort of effect, right? You know, I remember when, you know, Google was not a thing and then it became a thing and it became part of our vocabulary and, you know, people use it across the world in different circumstances. And I think this this is such an astute observation, but it kind of hits at the heart of I think what is what is missing in in the world of AI. It's not become so essential that people can't ignore it. And you know, there's so many people in this world who can carry on in their lives and they don't have to worry about it at all. And you know, I almost wonder what their lives are like. I I love I'd love it's probably simpler, you know, but it it is not the kind of technology that is so essential in our everyday life and our work that if we're not using it, it creates such a unfair disadvantage to to those who are not doing so that it could imperil their jobs, for example. And you're right, you know, a lot of the adoption at the moment in Europe and Asia, elsewhere, um, you know, I'm not sure actually that China we can include that in there, but certainly um elsewhere is is driven by your workplace. you know, you're you're told to go off and like use Copilot or you're told to go off and use Gemini and no one really tells you how to do it and so people are just mostly using it as a search engine if anything or maybe to get them to do a little bit of primary research for some of their projects. Um, and that's, you know, a very narrow segment of the population that work at desk jobs that are, you know, able to do that. You know, I sincerely doubt that you have engineers in the field that are trying to do infrastructure projects that are asking Gemini and and Chat GBC how to build a bridge or how to connect, you know, infrastructure projects or anything like that. Like I sincerely doubt that we've reached that point of of going, "Oh, you know what? I'll just do a quick a quick little search." I I don't think that's necessarily happening. And and I guess that's that's the difference in Silicon Valley and and San Francisco and more widely. I mean, they've always been early adopters of of the technology that's coming out of there. this is primarily an American driven wave and it makes sense for them to be early adopters and evangelists of this technology whereas you know other people across the world are kind of like eh you know but but that's but that's that's the challenge
is there then I don't know if I'm going two step one step ahead here is there then a direct link between meeting those kind of lofty valuations we talked about earlier and getting people to use it more and be better at it
100% but I think that what is clear even from that example example is that the technology that is being propagated in its current form is not doing that function. And so you know some people believe that that moment will happen when we reach you know AGI or singularity where like you know that the development of um AI is so great that it's able to think I in an equivalent to human being and it's able to operate at a function that we've not seen before. Some people argue that we've reached that point. If that's the case it's I feel quite disappointing but okay fine. Um yeah um but yeah it it does feel that there haven't been uh case studies or examples that are so applicable to everyone's lives that it becomes something that you know someone's nan might be using and also someone in Silicon Valley and it would make sense for it to be applicable to both and that is the challenge like it doesn't this feels like it's very much the beginning of the journey rather than the end of the journey right and so you have a technology that's still very young that is still being developed that's still being refined. Um, and the case studies will come with time should there be interest in doing so and should the money not run out. Um, because I think we all remember what happened with other examples of these things in the past like the metaverse. Everyone was talking about the metaverse. Everyone was buying real estate in the metaverse. I mean, who's who's next to Snoop Dog's mansion in the metaverse now? I don't know. Does he still have it? I don't know. You know who who's got board apes still? Like, message me. Honestly, I I just be really really intrigued to know like what what what happened to these things and and they were huge as well. And you know, you wonder like what is the real world application to these things and it feels like the competitive advantage is maybe for businesses, but even then it's not so good that we have like a widespread number of examples of businesses that have been like we've increased our revenue by x% because we've been using LLM. We haven't seen those yet.
Just finally, sorry to uh sorry to quote back to you, but I believe the last time you were on the show, you said if we don't have some good positive stories here, there's going to be repercussions across the market. Uh obviously a lot we just talked there are negative stories. What cause for celebration is there for these uh some of these companies we just talked about?
I think that you know
Besides the fact that they're incredibly valuable and probably the most valuable companies the world has ever seen.
Exactly. I mean, yeah, it's great. Where's a lot of money? No, I I think that there are more more examples and and I think, you know, with with the bad always comes the good. You know, we we we had davos recently and and lots of people were talking about the transformation of the job market and there was again a lot of negativity there going, you know, we believe that most jobs will transform or be eliminated or be affected by AI in a way that we we cannot conceive at the moment, but it will happen. It will happen soon. But what also came out of that which gave me a lot of hope was this idea that like loads more jobs that we also could not imagine will come out of the woodworks and I think you know a lot of the jobs that we're doing at the moment um and even like us having this conversation now 50 60 years ago would have been inconceivable and so I I have a lot of hope that this kind of technology will unlock our minds. I think you know I think the same way about the internet um and how much it has freed us how much information has unlocked how democratizing it has been um obviously I feel that you know a closed source model has big problems but we're seeing more and more open-source AI models that allow you to learn allow you to build allow you to experiment and I feel like if if more of those are thrown into the mix and more of those prove to be a bit more prevalent um we'll be in a really good place I think because it will unlock creativity and democra ize our ability to unleash AI and use it for really cool things and I think that can only be a nice thing but I just want more examples of that and quick please and quick
sorry just very quick very quickly you mentioned Davos yeah I saw this morning I cannot remember the name of the interviewee but the BBC did a small interview at Davos and uh one of the things they got to they said oh yeah the UK certainly has the potential to be an AI superpower in the future I mean I don't know I live in the UK I'm you do as well. That seems a long way off to me.
I think it depends on how we play um sovereign AI, how we play building infrastructure and how we play most importantly uh promoting businesses that are trying to develop this technology. Like we don't have um an open AI or a Google or a Microsoft competitor. And you know, we've been talking about this and people have been talking to me about this for many many years that we're we have an ecosystem where it's the best time to be an entrepreneur where you have huge amounts of opportunity and huge amounts of talent. And there's a reason why President Trump and you know all of his tech friends came over here uh trying to promote you know their projects and it's because there's the talent is there. It's for the picking. And I think, you know, what the UK has historically been kind of bad at is is, you know, pushing forwards that that technology, pushing forward those people and allowing them to grow domestically and not flock over to the US. And I think if we're able to do that better, which people have, look, both governments, both the Labour and the Conservative governments in the past have had examples of people who have told them exactly what to do. This is not rocket science. This is not a mystery. you that there's a very clear reason why people and entrepreneurs decide to do what they do and decide to leave and decide to sell up and decide to do all these sorts of things and why we don't have these big powers here and there are there is a way to change that and it's in our power to do so um there's still time is what people were saying at Davos there's still time to do this like we we still can do this we've got the ability we just need the the means to do so and for it to be easier to be an entrepreneur and easier to grow a business without being penalized ized and without feeling on the back foot because we don't have enough investors and VCs here that are willing to do so and instead you've got a lot of VCs in America who are only focused on filing there. So there is there is absolutely an opportunity. Um I feel it would be really sad if this whole push for sovereign AI ended up being us as a bis as a company and as a business community uh kind of acting as an ecosystem through which all these big players come and come and use the resources and then take their profits elsewhere. I think that would be a really sad state of affairs and it is certainly on the cards. So um I think we need to do something and quick um we can follow the example that Deep Seek set last year and it was it was it caught them by surprise. They didn't see it coming. People who are watching AI though they knew what was coming. It wasn't it wasn't a massive surprise. This is they're not reinventing the wheel. They're just showing they can do something that's more effective and smaller scale that uses less energy. thinking about literally what George Noble was saying like are there people in the UK that could solve that conundrum of of the energy crisis and AI. These are the kind of things that could really make a difference. You don't need to you need to be the next open AI. You need to be the the company that can sell them a solution to their LLMs. Um it's not it's not that hard. It's just it's the hard thing is is is to make give people the opportunity to do it in my view. I think that seems like a very good place to end it today. Natasha, we always appreciate having you on. Thank you for coming on the Tech Report.
Thank you for having me.