14/09/26: OpenAI, millennium prizes & US inflation reports
Monday Espresso Podcast - 14th Septemeber 2026
[00:00:00] Rory Dowie: Good morning. Today's Monday the 14th of September. I'm Rory Dowie, portfolio manager here at Marlborough, and today I'm delighted to be joined by one of our investment analysts, Nick Warmisham. Nick, good morning.
[00:00:12] Nick Warmisham: Good morning. Rory.
[00:00:13] Rory Dowie: Two big stories this week. First, a real milestone in the artificial intelligence race.
[00:00:18] Rory Dowie: OpenAI says, it has solved one of the great unsolved problems in mathematics. And secondly, the last US inflation reports before the Federal Reserve sets interest rates on Wednesday. But Nick, first things first. How did markets due last week?
[00:00:33] Nick Warmisham: Yeah, it was a pretty nervous week overall Rory, driven by the oil price increases. Renewed tensions between the US and Iran pushed oil sharply higher. Brent crude briefly topped $105 a barrel up from around $73 at the start of the year.
[00:00:47] Nick Warmisham: Dearer energy raises worries about inflation and that pulled equity markets lower for most of the week with energy companies being the only exception really.
[00:00:56] Nick Warmisham: The UK FTSE 100 held up better than most because it's full of big oil producers like Shell and BP, which gain when crude oil prices rise.
[00:01:06] Nick Warmisham: And by Friday, oil was easing back and markets steadied somewhat.
[00:01:11] Rory Dowie: Yeah, so oil being the major driver there across markets, and really a reminder for all of us is how important the oil price really is and how that affects inflation. And then obviously how investors interpret that oil price move and what that may mean for the future pathway of interest rates and inflation.
[00:01:28] Rory Dowie: But let's start with the first story, Nick, because that is quite something. OpenAI give us the latest.
[00:01:34] Nick Warmisham: Yeah, well last Tuesday OpenAI, the company behind ChatGPT said one of its AI models had solved the Navier Stokes problem. That's one of the seven millennium prize problems. A famous list of the hardest, unsolved questions in mathematics, each carrying a $1 million reward.
[00:01:52] Nick Warmisham: Since that list was drawn up in the year 2000, only one had ever been solved. The problem itself asks how water and air move and whether the equation that describe that motion can ever break down, in effect spin faster and faster until they stop making sense. And it has stood unsolved for about 90 years.
[00:02:13] Rory Dowie: Yeah, and put that significance in context, Nick, because it really is quite remarkable the pace of improvement that we've seen in some of these AI models.
[00:02:21] Nick Warmisham: Yeah, that's the real striking part about it. Just over a year ago, the benchmark for AI in maths was answering international maths Olympiad questions extremely hard, but questions that already have known answers.
[00:02:35] Nick Warmisham: This is a different order of difficulty, an open problem that no human had ever cracked. OpenAI used up to 10,000 AI programmes working side by side over roughly 88 hours, at a computing costs running into the millions of dollars. The finished proof was then checked by software that verifies every single logical step.
[00:02:56] Nick Warmisham: So mathematicians can be fairly confident it holds up. One leading mathematician, a past winner of the field's top prize called it undeniably a big moment.
[00:03:06] Rory Dowie: Two things to caution though, probably for our listeners. Firstly, it wasn't entirely clean. Actually a rival team of human researchers working closely on a related problem at the same time, and there's been a dispute over potentially who got there first.
[00:03:20] Rory Dowie: You know, those guys are OpenAI, so that's one angle to look at. And then secondly a result, like this still needs to be independently checked, obviously, by the wider maths community, but OpenAI have said that they're pretty confident in it. Other mathematicians said they're pretty confident and even with the solving of the problem, OpenAI have said it's not about the money and they're not gonna claim the million dollar prize.
[00:03:40] Rory Dowie: So what does that mean, Nick?
[00:03:41] Nick Warmisham: Well, I think that tells you that the prize money isn't really the point for a company like OpenAI. The value is in the demonstration showing its technology can do genuinely original scientific work, not just repeat answers that it's seen before.
[00:03:54] Rory Dowie: Yeah, and I think that's the main point for us as investors.
[00:03:57] Rory Dowie: It's another data point behind the enormous sums now being poured into AI and a reminder of why so many of the world's largest companies are spending so heavily and aggressively on it. Whether all that investment ultimately pays off is the trillion dollar question, but breakthroughs like this clearly are exactly what that spend is chasing.
[00:04:15] Rory Dowie: Right, Nick, let's move on from the frontier of technology to something that touches every household around the world. Inflation. What did the US figures on Friday show us?
[00:04:25] Nick Warmisham: So on Friday we got the August inflation figures, which measure how much the cost of a typical basket of goods and services has risen over the past year. Prices were 3.4% higher than a year earlier. That's the same as July and slightly above the 3.3% economists had expected. So a little hotter than hoped.
[00:04:45] Nick Warmisham: On the month alone, prices rose 0.4%, four times July's pace. The main driver was energy, with petrol prices jumping nearly 4% in the month and oil above a hundred dollars a barrel, that alone accounted for about a third of the increase.
[00:05:00] Rory Dowie: Yeah, okay. So headline inflation, energy prices, no surprise affecting it there, but there is a second figure that the Fed watches even more closely, isn't there?
[00:05:09] Nick Warmisham: There is indeed. It's called core inflation that strips out food and energy because those two are volatile and can swing around month to month for reasons that have little to do with the wider economy.
[00:05:20] Nick Warmisham: What's left is a cleaner read on the underlying trend. Core prices rose 0.3% on the month, a touch above the 0.2% expected. Over the year core inflation actually eased slightly to 2.4%, so a mixed picture Overall. The yearly trend is still gradually improving, but the monthly detail came in firmer than the Fed would've liked.
[00:05:41] Rory Dowie: And the reason why this matters so much this week, Nick, is?
[00:05:44] Nick Warmisham: Because the Federal Reserve sets US interest rates on Wednesday, and this was the last major piece of data before that decision. Its target for inflation is 2% and at 3.4% we're still well above that. A softer report might have eased the pressure, but this wasn't it.
[00:06:00] Nick Warmisham: On top of that, the US job figures a week ago came in strong, which points to an economy that doesn't obviously need lower rates, put it together and markets now put the chance of a rate rise on Wednesday at about 70%. That would be the first US rate increase since 2023. A real change of direction.
[00:06:17] Rory Dowie: Yeah, okay. So we'll be watching that Fed rate decision very, very closely this week. I guess, Nick, aside from that, what else have our listeners got to look out for this week?
[00:06:28] Nick Warmisham: Yeah, well, it's a big week for Central Banks. The US Federal Reserve, as you mentioned, decides on Wednesday, but the Bank of England and the Bank of Japan meet too.
[00:06:37] Nick Warmisham: So expect plenty on interest rate decisions one way or another.
[00:06:41] Rory Dowie: Very clear. Nick, thanks as ever. To our listeners, as always, please do reach out if you have any questions and wishing you all a great week ahead.