What does the World Cup, Barbra Streisand, Levi's, and Heinz have in common?

**Leanna Byrne** (0:00)
Micron packs a mighty punch on the markets.

**Chris Miller** (0:03)
Micron does make the memory that AI systems require. So when you use AI, ask it a question, it requires a lot of memory.

**Leanna Byrne** (0:11)
It's World Business Report from the BBC World Service. I'm Leanna Byrne. Micron says demand from artificial intelligence is sending profits surging. So what exactly does the company do, and where does it sit in the global chip race? And FIFA tries to hide brands that didn't pay to sponsor the World Cup, but has covering them up only made them more famous.
So Micron isn't a household name the way Nvidia is, but this week it became one of the companies to watch in AI. It had blockbuster results which sent its shares soaring by as much as 19%. So what exactly does Micron make? I put that question to Chris Miller, author of Chip War and an expert on microchip technology.

**Chris Miller** (1:00)
Micron is one of the world's three big producers of a type of memory chip called DRAM, which is found in phones and computers, but is also a critical ingredient in the most advanced AI servers.

**Leanna Byrne** (1:13)
So is it right to think of Nvidia as making the brain of the AI systems while Micron makes the memory those systems need to work?

**Chris Miller** (1:21)
That's right, Micron does make the memory that AI systems require. And AI is very, very memory intensive, and especially the inference workload. So when you use AI, ask it a question, it requires a lot of memory to produce the answer. And this has driven a surge in demand for the types of memory chips that Micron manufactures for AI companies. And it's caused shortages in other parts of the memory industry because there aren't enough chips to go around.

**Leanna Byrne** (1:50)
Now, where exactly does it sit in the chip supply chain?

**Chris Miller** (1:54)
Well, Micron is both a designer and manufacturer of memory chips. It's comparable to companies like Korea's SK Hynix or Samsung. And these companies provide chips for a wide range of applications. But for AI, they're embedded inside of the servers that are produced by NVIDIA or other AI companies. So they're deep inside of the data centers that are enabling AI capabilities.

**Leanna Byrne** (2:21)
So we know what the company does now. So what has given Micron the sudden boost?

**Chris Miller** (2:27)
Well, over the last couple of months, it's become clear there's a real shortage of AI chips, especially the high bandwidth memory chips that Micron is one of just three companies in the world to produce at scale. And this has driven up the price for these chips and most other types of memory chips as well. And so for companies like Micron that have production capacity of these chips, they've raised their prices to deal with the supply crunch. And that's driven much higher profitability across the memory sector.

**Leanna Byrne** (2:57)
I also saw it signed a series of long-term customer agreements. Why is that such a big deal for a memory chip company?

**Chris Miller** (3:03)
In the past, memory chips were fairly commoditized. All the key memory players made roughly comparable types of memory chips, and they would sell them on short-term contracts. The price would swing up and down based on whether the market was oversupplied or undersupplied. But now, for the first time, we're seeing really substantial long-term contracts that are locking in customer relationships and prices. And this is going to help smooth out the memory company's revenues so that they're less susceptible to the ups and downs of the market, and it's going to lock in higher prices for them because the current prices are really elevated relative to historical trends.

**Leanna Byrne** (3:44)
So does that mean customers are effectively paying up now to guarantee that supply later?

**Chris Miller** (3:49)
Well, that's exactly right. And companies are realizing that there are only a couple producers of these high-end memory chips. It's very difficult to bring new production capacity online. It'll take multiple years to build a chip making facility. And so if they want to lock in capacity, they've got to sign longer term agreements.

**Leanna Byrne** (4:06)
And Chris, who are those customers?

**Chris Miller** (4:08)
Well, customers for memory chips include smartphone manufacturers, computer manufacturers. But the specific type of high-bandwidth memory that's used for AI are the most advanced AI and data center companies. So NVIDIA, certainly an important player in this segment. But companies like Google and Amazon that design their own chips are also major users of this type of high-bandwidth memory.

**Leanna Byrne** (4:33)
I want to bring Apple into this because, interestingly, Apple says memory and storage costs are rising so much that it's having to put prices on Macs, iPads, and its other products up. So is this the same AI boom showing up in consumer prices?