What Apple's new CEO is cutting isn't just cost. It's distance
Originally written in Chinese. Translated with AI, reviewed by Gary. Read the Chinese original →
Apple's new CEO John Ternus took over on 1 September, and less than a month in, his first move is to cut.
About half a dozen director-level engineering program managers in the hardware engineering group have been told to find other roles inside Apple within weeks, or they may be let go later this year.
Teams including Siri, Vision Pro and AI software engineering have also seen layoffs.
Apple had even studied laying off about 5,000 AppleCare support employees and handing part of their work to AI agents, though that plan is on hold for now.
In planning for 2027, budgets are tightening, headcount is expected to stay flat, marketing spending is being cut, and even plus-ones at the company holiday party have been cancelled.
At the same time, Ternus wants to break Apple's long-standing spring and autumn launch rhythm and release new products more often throughout the year.
Basically, everything he has done since taking over points in the same direction:
Cut people, cut spending, move faster.
Why is one of the most profitable companies cutting first?
What's interesting is that this isn't a company in trouble.
In the third quarter of fiscal 2026, Apple's revenue was US$109.4 billion, up 16% year on year, with a 50.1% gross margin. It was the best June quarter in its history.
One of the most profitable companies in the world, and the new CEO's first move isn't to talk about expansion, but to start cutting.
Why?
Reading Bloomberg's report again, I think what John Ternus really wants to cut may not just be cost, but distance.
The distance between engineers and decision makers, between the front line and management, and between spotting a problem and getting to someone who can actually decide.
The report says he wants to remove some layers of middle management so engineers are closer to senior executives, to get more done with fewer people, and to use AI to speed up product development and automate part of the work.
The cost that rarely shows up on the P&L
As a company grows, there's one cost that rarely appears on the P&L: decision latency.
When a company is small, a customer complains and the founder knows quickly. If a product has a problem, one WhatsApp to the warehouse sorts it out the same day. If ads aren't working, you look at the numbers that night and change them the next day.
As the company grows, the front line tells a supervisor, the supervisor tells a manager, the manager writes it up for the head of department, then there's a meeting, a report, and the management meeting. By the time it reaches someone who can say yes or no, weeks may have passed.
Even more dangerous, information gets tidied up at every layer.
Tidy it once and there's a bit less noise. But tidy it too many times, and sometimes the truth gets tidied away too.
Recently I happened to talk with Jing, who leads BilaBila Mart. They now have more than 150 convenience stores. When I asked what worries her most now, she said something that stayed with me:
She worries that one day, she won't know what customers actually want.
For a founder to still worry about that at this scale is rare.
When a company is small, we are next to customers every day. As it grows, we start managing "the people who manage customers". Bigger still, we manage "the people who manage the people who manage customers".
In the end, what the CEO sees every day is more and more complete: dashboards, PowerPoints, KPIs, weekly reports. More and more numbers, while the customer may drift further and further away.
That's what is so frightening about bureaucracy.
It isn't necessarily because people are lazy. Often everyone works very hard, and every process had a reason to exist in the first place.
It's just that every time the company grows, a process is added. Every time something goes wrong, an approval is added. When a department grows, a manager is added. Another problem, another SOP.
Nobody does anything wrong, yet the whole organisation slows down, layer by layer.
In the AI era, what's slow is the organisation
The speed of the AI era is completely different.
A product cycle used to take a year. Now a competitor ships a new feature in a few weeks. Work that used to need ten people to gather, analyse and report may soon be done by AI in a few hours.
If technology is already ten times faster but the approval structure is still the one from ten years ago, what slows the company down may no longer be technology. It's the organisation.
Looking at Apple's changes, even one of the world's most profitable companies has started re-examining its own organisation: which layers can be thinner, which decisions can be faster, and how the front line can get closer to the people who actually decide.
Even Apple is changing.
If a company has already started slowing down and drifting away from its customers, yet still feels "it's fine as it is", the real question may no longer be whether to change.