Apple's Mac revenue grew about 30% year over year in the June quarter, faster than any other segment in the company, in a quarter when the iPhone was also growing above 20%.
Apple did not build the Mac mini for artificial intelligence, does not sell its chips to anyone, and does not break out who is buying. Developers running AI agents bought the machines anyway, in bulk, and now cannot get them.
"It's just so much cheaper running it right on your device where you already own the hardware."
Aaron Tilley covers Apple for The Information and published the reporting behind this segment that morning, including the bulk orders from OpenAI and from a neocloud nobody had heard of.
I listened to the full segment so you can skip it.
Here are the 6 takeaways that matter.
👤 Guest: Aaron Tilley, who covers Apple for The Information and wrote the story on the Mac's AI-era demand that this segment is built on
🎙️ Host: Akash Pasricha, who anchors TITV, The Information's live weekday show at 10 a.m. Pacific
📰 Published: 31 August 2026 on YouTube (The Information)
🔴 YouTube | ⏱️ 10 min
Key Takeaways
The Mac mini became the default machine for running AI agents, and Apple did not plan it
Developers leave the box running in the background as a personal agent, with no keyboard and no display
Running an open-weight model on hardware you already own undercuts paying for cloud inference
Unified memory is the specific reason Apple silicon suits this work
On a conventional PC the memory sits in separate places, which creates bottlenecks
Nvidia now treats Apple as a competitor in local AI, which is new
It shipped DGX Spark in the past year as a direct answer to the Mac mini
Mac was Apple's fastest-growing segment in the June quarter at about 30%
The iPhone grew above 20% in the same quarter and was still left behind
Whether this lasts is a supply question, not a demand question
Buyers who cannot get a Mac mini go and look at the Nvidia box instead
1. Mac minis run the agents
The segment started from the shortages people have been posting about. Pasricha asked what buyers are actually doing with the machines.
Tilley said agents are the use case, and the Mac mini and Mac Studio are the hardware developers settled on. "And the Mac minis, Mac Studios have really been the platform of choice, hardware platform of choice for these developers."
The appeal is the form factor: a box with no keyboard and no display that can be left running a task in the background
The second use case is local AI development. Developers run open-source models, including Chinese ones such as Qwen, directly on the machine rather than calling a cloud API
The cost argument is the whole argument. "It's just so much cheaper running it right on your device where you already own the hardware."
2. Unified memory is the edge
Asked what makes Apple's computers good at this, Tilley pointed at the chip design rather than at raw speed.
Apple has designed its own chips for well over a decade, and Tilley put the first one at around 2011
The architectural difference is where the memory sits. "They have this thing called unified memory where the memory on the chip is shared between all the parts and that makes the process of doing AI computing very fast."
On a traditional PC the memory is distributed across different parts of the chip, which Tilley said creates bottlenecks in speed
He added that Apple also has some of the best mobile central processing units, which suits certain kinds of AI work
3. Nvidia built a rival box
Nvidia sees local AI as a market worth defending and shipped a competing machine. "So they and you know the past year they released DGX Spark which is a very similar computer to the Mac Mini."
The competitive framing is unusual, because these two companies do not sell the same thing. "I mean, they make chips, they don't sell them. And Nvidia sees them as huge competition."
Nvidia dominates the cloud, Tilley said, and the growing demand to run work on local machines is where it wants to play next
4. Mac grew 30% and led Apple
The demand shows up in the reported numbers. "Despite like a lot of supply issues in the most recently reported June quarter revenue was up around 30% year-over-year and that's far faster than any other business segment."
The comparison that makes the number mean something is the iPhone. "You know the iPhone has continued to do quite well above 20% but Mac is currently as a segment is blowing it all out of the water."
5. The buyers are AI labs
Pasricha asked whether the growth is coming from the enterprise side of Apple's business.
Apple does not break out those financials, so the evidence is Tilley's reporting rather than a disclosure. "As are reported in the story, you know, OpenAI has purchased tens of thousands."
He said he has heard of other AI labs placing orders in the tens of thousands as well
One buyer is a company most people have not heard of. "There's this new Neocloud I reported on in the story called Mount Thor." He put its purchases in the tens of thousands of Mac minis too
His summary of the shift is that bulk orders from enterprises and startups are now consuming this hardware
6. Supply decides if it lasts
Asked whether this is a blip or a trend, Tilley made it Apple's decision rather than the market's.
The complaint he heard most was not price or performance. "The biggest complaint I heard from people in this story is lack of supply."
Running out of stock is how the demand goes somewhere else. "Apple has really let these products fall out of supply and these people can't get these products anymore." Buyers then look at Nvidia's DGX Spark instead
He expects the incoming chief executive to care, and named the reason. "I think an interesting move for Apple in terms of their executive staff is Johny Srouji, their chief chip person who's now chief hardware officer."
Srouji led the silicon program from the beginning, so highlighting the chips is the natural way for him to show its value
Bonus Insights
The segment aired on John Ternus's first week as chief executive and Tim Cook's last day in the job, which is how Pasricha framed the timing
The show opened with a separate scoop that bears on the same hardware question. Pasricha said The Information reported that morning that CXMT, China's largest memory-chip maker, has begun producing advanced high-bandwidth memory, which he called a potential breakthrough for China's domestic AI chip market
This was the first of five interviews in one episode; the other four are written up separately
Tilley's bottom line is that Apple has an accidental AI hardware business growing faster than anything else it sells, and the only thing that would end it is Apple failing to make enough of the machines.
Products, Companies & Tools Mentioned
Mac mini and Mac Studio (The two headless Apple desktops developers have adopted for running agents and local models; both are in short supply)
Nvidia DGX Spark (The compact machine Nvidia shipped in the past year that Tilley says is very similar to the Mac mini, and where buyers go when Apple is out of stock)
OpenAI (Named in Tilley's reporting as having purchased Mac minis in the tens of thousands)
Qwen (His example of the open-source models developers run locally instead of paying for cloud inference)
Claude and ChatGPT (The cloud services the local setup is meant to avoid paying for)
Mount Thor (A neocloud Tilley reported on whose Mac mini purchases are also reaching the tens of thousands)
CXMT (China's largest memory-chip maker, which The Information reported that morning has begun producing advanced high-bandwidth memory)
Books & Resources Mentioned
How Apple Stumbled Into AI Hardware Success With the Mac (Tilley's own story, published that morning, which this segment walks through)
If this was worth your time, send it to someone closer to the industry than you are.
Get the latest market chatter as it happens: