Transcript: Israel’s economy slumps

This is an audio transcript of the FT News Briefing podcast episode: ‘Israel’s economy slumps’

Josh Gabert-Doyon
Good morning from the Financial Times. Today is Tuesday, February 20th, and this is your FT News Briefing.

[MUSIC PLAYING]

Israel’s economy shrinks 20 per cent. Ukraine loses a key strategic outpost in the east.

Ben Hall
As you said, it was the first big battlefield setback for the Ukrainians in several months.

Josh Gabert-Doyon
Plus, is OpenAI’s business model sustainable? I’m Josh Gabert-Doyon, in for Marc Filippino, and here’s the news you need to start your day.

[MUSIC PLAYING]

Israel’s economy has been hit hard by the war in Gaza. The country’s total economic output shrank by nearly 20 per cent on an annualised basis in the last three months of 2023, that’s according to official state data. While the government has been pumping funds into the conflict with Hamas, Israel’s also dealing with more than 300,000 reservists who had to leave behind their day-to-day work. And restrictions that Israel imposed on Palestinian border crossings caused a labour shortage in the construction sector. Moody’s, the rating agency, downgraded Israel’s credit rating earlier this month over fears that the conflict could drag on.

[MUSIC PLAYING]

Ukrainian forces have suffered their first big battlefield defeat since Russia’s capture of Bakhmut last May. They pulled out of the eastern town of Avdiivka at the weekend. In spite of these losses, Ukraine has still managed to recapture and hold about 50 per cent of the territory taken by Russia when it launched its full scale invasion in 2022. But optimism about the strength of Ukraine’s military force is fading as the country runs short of weaponry and ammunition. To discuss the state of the war, I’m joined by the FT’s Europe editor, Ben Hall. Hey, Ben.

Ben Hall
Hi.

Josh Gabert-Doyon
OK, so, Ben, how significant is the loss of Avdiivka in this conflict?

Ben Hall
I think it was significant in two senses. One, as you said, it was the first big battlefield setback for the Ukrainians in several months. This is a town they’ve been defending actually, since 2014. And it probably has more strategic value than Bakhmut, which was the town that they lost back in May last year, because it’s sometimes deemed the sort of gateway to Donetsk, which is a Russian occupied city. So it has a sort of strategic purpose. But the second reason, I think, was that this was a test for Ukraine’s new top commander, Oleksandr Syrsky, and the political leadership under Zelenskyy, because it was a test of to what extent are they willing to expend Ukrainian forces whilst defending territory now that their strategy is essentially a defensive one. Grinding down the Russians might be a perfectly good military strategy, in fact it is their military strategy. But you don’t want to do it if you’re grinding yourself down.

Josh Gabert-Doyon
OK. You mentioned this particular battle has been going on more or less since 2014. What changed? Why did Ukraine lose it now?

Ben Hall
Well, they were clearly outgunned. There are analysts who say that Russia now has a 10 to one artillery advantage. It can fire 10 shells for every one that Ukrainian forces can fire, which obviously makes it harder to defend territory and gives Russia a huge ability to essentially bombard buildings. Russia also poured in enormous numbers of forces. They’ve been very successful over the last year, and more of raising manpower for the army and effective manpower. So the Ukrainians were outgunned and outmanned, and ultimately they just couldn’t hang on to this little pocket of Ukrainian territory without just sacrificing enormous numbers of their troops. And they rightly, I think, concluded it wasn’t worth it.

Josh Gabert-Doyon
And what about Oleksandr Syrsky, the new military commander you mentioned? I know he came in after Ukraine’s kind of failed counteroffensive. Are we seeing any change in tactics now that he’s in charge?

Ben Hall
Well, it’s difficult to know whether the change in tactics is because of him or whether they would have changed anyway, even if they had stuck with the same command of his predecessor, Valeriy Zaluzhny. The thing about Syrsky was that he is held to a degree responsible by a lot of Ukrainian soldiers for the Battle of Bakhmut, because Syrsky before was in command of Ukraine’s land forces. And he is thought to have played a role in keeping Ukrainian forces in Bakhmut longer than perhaps they should have been there to defend the territory, although that’s possibly also what President Zelenskyy wanted. So that’s also why Avdiivka in a way was seen as a test because was Syrsky gonna defend it like Bakhmut at all cost, or was he actually going to prevent what could have been quite damaging Ukrainian losses. But they’ve taken the view that it was better just to pull back.

Josh Gabert-Doyon
And have there been any successes at all that could boost Ukraine’s morale?

Ben Hall
I mean, we haven’t talked at all about the long range strikes by drone and missile in Crimea, which is their one kind of real bright spot. And they have been remarkably successful, given that they don’t even have a navy of any size, and they are carrying out long range drone strikes on Russian air bases and Russian oil refineries. Those could, over time be stepped up and could weaken the Russian military machine, but probably not on the scale that Ukraine really needs.

Josh Gabert-Doyon
Ben Hall is the FT’s Europe editor. Thanks so much, Ben.

Ben Hall
Thank you.

[MUSIC PLAYING]

Josh Gabert-Doyon
OpenAI has seen massive investment and user growth since it launched ChatGPT in late 2022. But those models are very, very expensive to build and operate. And now there are questions about whether OpenAI can make enough money to sustain the powerful artificial intelligence it offers to customers. Madhumita Murgia is the FT’s AI editor and covers OpenAI closely. Hi, Madhu.

Madhumita Murgia
Hi there.

Josh Gabert-Doyon
So most listeners will be familiar with OpenAI by now, and the fact that it’s received a lot of financial backing from Microsoft and some others. But before we get into some of the potential problems, give me a sense of OpenAI’s business model.

Madhumita Murgia
Sure. So their main ways in which they’re making money today is selling access to ChatGPT, either at scale to enterprises or small teams, where they charge a monthly sort of fee of $25 to 30 per person per month. They also sell access to the underlying model that kind of underpins ChatGPT. And developers, programmers, companies can access this via something called an API, which is a plug in, and then they can use that to build their own applications on top of that. The most recent kind of idea they’ve had for making money is the GPT Store. And this is essentially similar to the Apple App Store. They allow people to build little applications for specific purposes using ChatGPT. So these are the kinds of main ways in which they’re currently making money today.

Josh Gabert-Doyon
OK, so that’s how it makes revenue. What kind of cost is OpenAI dealing with? How much does it actually cost to build and run this AI?

Madhumita Murgia
So that’s the billion or trillion-dollar question as it may be, the biggest kind of expense of building these large language models, which are the technology or the software that runs a ChatGPT, for example, is the training cost. So this means when a new model is being developed, it has to be trained on huge amounts of data. And the cost of that is what sort of what’s the expensive part of it. And estimates for training can vary anywhere between hundreds of billions to, you know, recently an estimate put it at 7tn as the sort of capital needs for OpenAI to get to AGI and to train more and more sophisticated models. So these are just, you know, astronomical figures. And really the question that we were trying to answer is, you know, are they ever going to be able to build that with the kind of huge capital costs before they ran out of money. Because the revenues they’re making aren’t going to make a dent really on how much money they need to build it.

Josh Gabert-Doyon
But what presumably lots of companies are using ChatGPT and willing to pay for it, right?

Madhumita Murgia
So on the one hand, you know, they do have a lot of take up. Over 90 per cent of Fortune 500 companies, for example, are using ChatGPT, according to the company. But it’s not clear if they’re just sort of experimenting and playing in the margins or whether they’re actually gaining huge value from it, building products on top of it. We spoke to a couple of, you know, different sized companies. We talked to Salesforce, and they’ve developed a sort of customer service chatbot using these models. And they found huge uplift for their clients on customer service call handle times, you know, gone down by double digits. It’s just way quicker because the chatbot can find you the right answer and tell you how to fix it. So it’s transforming what your existing workforce can do, pushing towards more sales, more growth. But it is very nascent, and people are still kind of hesitant about rolling it out at full enterprise scale. And I think that’s what this year will be about, about figuring out how it’s gonna actually tangibly change how business is done.

Josh Gabert-Doyon
Right. So it sounds like we’re still in the early phases of the uptake of this technology. Presumably, OpenAI and Sam Altman are gonna have to rely on investors if the revenue isn’t filling that gap. How have OpenAI gone about raising money?

Madhumita Murgia
Yeah. So I think traditional venture capitalists, some of whom have invested early in OpenAI like Thrive Capital and Sequoia, they’re gonna start to be priced out. Increasingly, Sam Altman, the CEO, and the company are gonna have to look elsewhere, and Microsoft is one of their biggest backers. And similarly, they’ll have to go to new avenues to find this money. And I think sovereign wealth funds are another area. Sam Altman has been speaking to investors in the United Arab Emirates, for example, as a way to kind of inject a huge amount of cash into growing the business and developing new models.

Josh Gabert-Doyon
I know OpenAI has this bigger goal of building transformative AI systems, changing the world with human level AI is what they like to talk about. Do you think they’re gonna be able to reach that considering these dynamics with their business model?

Madhumita Murgia
I think the one thing that I’ve kind of learned from my reporting around this area is that nobody knows the answer to that. I think what companies like OpenAI, Microsoft, Google and others want is for companies and consumers to buy into this dream because they need us to build with the technology, to get value from the technology, economic value, through products that we build for our businesses, for there to be wealth generated, which can then kind of be further invested into this, you know, moonshot that they’re all trying to get to. I think the jury’s out on this, but there’s gonna be a lot of money, data thrown at this problem to see whether this goes exponentially upwards or not. So we’ll need to be watching out for that over the next five to seven years.

Josh Gabert-Doyon
Madhumita Murgia is the FT’s AI editor. Thanks, Madhu.

Madhumita Murgia
Thank you so much.

[MUSIC PLAYING]

Josh Gabert-Doyon
You can read more on all of these stories at FT.com for free when you click the links in our show notes. This has been your daily FT News Briefing. Make sure you check back tomorrow for the latest business news.