"Zonal" relates to the concept of "availability zones" which are the next-smallest unit below a (physical) "region."
Most instances of a cloud ___ created in a region are allocated and exist at the zonal level (i.e. a specific zone of a region).
A physical "region" usually consists of three or more availability zones, and each zone is physically separated from other zones, limiting the potential for foreseeable disaster events from affecting multiple zones simultaneously. Zones are close enough networking-wise to have high throughput and low latency interconnection, but not as fast as same-rack, same-cluster communications.
Systems requiring high availability (or replication) generally attain this by placing instances (or replicas) in multiple availability zones.
Systems requiring high-availability generally start with multi-zone replication, and Systems with even higher availability requirements may use multi-region replication, which comes at greater cost.
I don't know much about his family life, but it's very plausible that he spent more time working during his lifetime than with his family, and this is a good take. Considering the company's state of affairs, I would assume he was under a tremendous amount of work-related stress during the last few years, if not the majority of his career, peaking in this role.
I would assume he would be in work related stress for decades but daughter's wedding preparation, considering how stressful asian weddings are, pushed him over the edge.
JH Han came from Samsung's Visual Display (tvs, smart tvs) business unit which is part of Samsungs larger consumer electronics business which includes things such as home appliances. Mobile electronics such as Galaxy brand smart phones are another business, and Han was promoted to lead the businesses spanning all of those groups, which was (re)named to called "Device Experience (DX)". Samsung's semiconductor fabrication and component businesses have normally had a separate "CEO" from the consumer electronics finished products businesses (aka. "SET")
The company name of "Samsung" originates from the founder's admiration of Japanese companies such as Mitsubishi, known as "keiretsu" and their place in its economy; he hoped that his company would also endure as a shining symbol in the sky, like stars. Mitsubishi means three stars in Japanese language.
Samsung the company started as something resembling a dry goods market.
JG is famous for research teams doing "hillclimbing" but the organizations doing this seldom thought deeply about where they were hoping to end up, what would be good-enough, or when it would be time to find a new hill.
The Vision/Vision Pro product was also years behind schedule, based on original expectations, and some people feel that they had to launch something so that's why we got what we got.
John "JG" Giannandrea was hired over to Apple around 2018 to head Apple's new AI/ML division, which would include the Siri organization that predates Giannandrea's tenure at Apple. Giannandrea was previously the executive in charge of Google's Research division.
Mike Rockwell was hired into Apple from Dolby Labs around 2015 to head a group called "TDG" that would be the R&D and the product group for what we know today as Apple Vision, an AR/VR headset. The same organization had its own AI/ML applied research teams focused on computer vision problems, and had hired in folks from Microsoft's Hololens team.
From what the article describes, the Siri organization is being moved from Giannandrea's scope to sit under Rockwell, and Rockwell will be moving away from the Apple Vision organization.
Apple hired Giannandrea to build an organization like the one he led at Google, and appears to have found that that the AI/ML organization built at Apple struggled to manifest its work into successful product wins despite the massive financial investment. It's worth noting that at Google, Giannandrea was succeeded by Jeff Dean, and the organization was more recently reorged to become the "Google Deepmind" we know today also had the same struggles.
> "appears to have found that that the AI/ML organization built at Apple struggled to manifest its work into successful product wins despite the massive financial investment"
I would contest this statement very vigorously. Siri has been a disappointment but nearly every single Apple product has a load-bearing reliance on an AI/ML feature.
I'm continually frustrated by the (relatively recent, since LLMs) tic where just because a ML model isn't conversing with you or otherwise making its presence impossible to ignore, that it isn't delivering a massive amount of value. It is frustrating from laypeople, it is extra frustrating from industry insiders.
Just a few examples off the top of my head:
- AirPods with ANC and spatial audio. Both headline features of these products that are 100% AI/ML.
- Watch with heart arrhythmia detection, automatic workout detection, fall detection, etc. All are headline features (some literally life-saving) and are 100% AI/ML.
- iPhones have high-profile features that are 100% AI/ML: automatic car crash detection. Others are more subtle but IMO substantial differentiators - such as automatic image enhancements out-of-camera.
Again, I know in the age of ChatGPT we seem to have twisted ideas of what ML is, but "AI/ML" is not synonymous with LLM.
1) This appears to be what Apple has concluded by the action they're taking and implied by the premise of the article, versus my personal opinion.
2) The examples given may have been produced by AI/ML-based technologies developed at Apple, but may not have been the work of Apple's AI/ML Organization. Many "groups" (divisions) within Apple have their own teams using AI/Machine Learning (e.g. AI/ML Org coexisting with AI/machine-learning work and teams in TDG/VPG, SPG (special projects group: the one previously working on a car), etc.
It feels like Apple had a dream of a granting easy AI integrations for many different apps and workflows, only to discover that very few people wanted any of those integrations from apple intelligence. At least, not the current iteration of apple intelligence.
I suspect the most significant thing holding Apple back in the realm of AI is the fact that Apple prides itself as a company that delivers revolutionary products with ‘wow factor’ that is leagues above the competition. Apple just doesn’t have anything like that with AI.
Furthermore, the challenge with AI is that even though it can deliver some shockingly impressive results, it also generates some bafflingly stupid responses, as well as answers that are seemingly correct but are never less wrong in ways that are hard to determine. Those hallucinations really shatter the illusion of quality and reliability, which ultimately undermines confidence in the functionality as a whole.
Given that, Apple is stuck trying to catch up to the rest of the products in the AI space while also realizing that everyday AI is nowhere near as polished as Apple would expect it to be. So the more Apple pushes AI, the more customers are likely to pick up on the errors and inadequacies of the product.
Even though other AI assistants suffer from the same fundamental problems, the ones from Apple Intelligence are going to seem more embarrassing because of Apple’s brand image.
That 'wow factor' is a thing of the past, very firmly. When I put next to each other say Iphone 16 pro max and Samsung S25 ultra, none sticks out, none has anything 'wow' for regular users. Just different styles of working with device and small + and - all over the place. Same for tablets, notebooks, and so on. Vision pro seems like a failed product even according to Apple.
Its doesn't mean that each product doesn't have something a bit special but competition caught up long time ago, sometimes went ahead (better batteries, bigger cameras etc). These days there is much more 'yawn factor', and its across whole industry.
They could add some “wow factors”, but the rate of error is still too big to ship it. Like, I don’t think it would be impossible for them to have a live language converter using LLMs when you call somewhere in a different language. But being responsible for stuff is just a can of worms that nobody wants to sign up for.
Gemini has access to your Gmail, Calendar, Docs, Drive, Keep, and Tasks, as well as Flights, Hotels, Maps, YouTube, YouTube Music, OpenStax, and of course the entire web more generally.
At the end of the day, personal context is king for useful AI.
Which means Google or Apple. Or Microsoft on the business side.
Nobody sane is going to share all their personal data (about everything) with a fly-by-night random AI app/company.
Apple has time, in the same way it did with Maps, because despite the marketing push no one is buying phones solely because of AI features these days. They're still an "Oh, that's nice."
>I suspect the most significant thing holding Apple back in the realm of AI is the fact that Apple prides itself as a company that delivers revolutionary products with ‘wow factor’ that is leagues above the competition.
I don't remember them ever doing that. Usually they start with something lacklustre then incrementally improve it until it is strong.
Maps is a great example, it was poor on launch compared to Google Maps but has now got to a level where most people don't bother with Google on the iPhone.
Google Maps largely exists because of Apple.
Do you remember when the iPhone was released?
AirPods? The iMac? The Mac? The Apple II?
Broaden scope a bit to understand what this comment meant. You’re right that Apple botches things in the small scale and especially in software, but they’ve been a great aggregator of amazing tech in the total lifespan.
Google Maps had already launched and had really surprised everyone at how much better it was than any other mapping tool a few years before iPhone launched. and just the interactivity of it in general independent of it being a “map” was a paradigm shifter.
Yes, the iPhone put it in everyone’s pocket, but if Apple had built a handheld device that was powerful enough to run any interactive mapping tool Google Maps was already the obvious choice.
It’s a minor quibble, but to say Apple played a majority role in Google Map’s success doesn’t ring true to me. In 2007 Google Maps was easily overtaking all other mapping tools with or without the iPhone.
It was great when I was visiting China. I only tried it because Google Maps was unavailable (nothing Google worked at all except push notifications, because those come from Apple), and was pleasantly surprised. Every train station and mall had detailed maps.
Apple Maps is also pretty good here in Canada, although I would never trust it to tell me with any accuracy what a given business's hours are, or whether it even still exists. Google is much better at that.
That is now, there was very little wow factor when Copland was being developed and they were at the edge of bankrupcy, long were the days of Mac Classic and LCs.
What is saving them this time around is the way they have been priting money with iDevices.
The devil is in the details – you do shit clippy-like integration = nobody will use it, you do it well – everybody will, even though nobody currently does in any meaningful way beyond setting a timer.
Well, the current iteration of Apple intelligence doesn’t enable personal context or 3rd party app related features at all. (https://www.macrumors.com/2025/03/07/apple-intelligence-siri...) So who knows whether people people want it or not; it’s not even out in the world! What Apple has actually discovered is that this is a hard problem to solve. But people do want a smarter Siri; they complain about it constantly.
Given the size of Apple and these orgs, plausible. Siri also had a bunch of folks that came in from outside, through acquisitions, and did their own thing or tried to identify themselves and their roles as something different from Apple norms.
Apple had teams involved with AI/ML in several divisions also, from AI/ML (including Siri), TDG/VPG, SPG...
„[…]struggled to manifest its work into successful product wins […]“
working in AI/ML, is there any recommended literature one may read on how to ensure efforts result in products? Would be interested to access more experience to benefit both an engineer and a manager pathway.
Most instances of a cloud ___ created in a region are allocated and exist at the zonal level (i.e. a specific zone of a region).
A physical "region" usually consists of three or more availability zones, and each zone is physically separated from other zones, limiting the potential for foreseeable disaster events from affecting multiple zones simultaneously. Zones are close enough networking-wise to have high throughput and low latency interconnection, but not as fast as same-rack, same-cluster communications.
Systems requiring high availability (or replication) generally attain this by placing instances (or replicas) in multiple availability zones.
Systems requiring high-availability generally start with multi-zone replication, and Systems with even higher availability requirements may use multi-region replication, which comes at greater cost.
reply