Broadcom knew this day would come. Broadcom has literally state they will rather sell their WiFi Business in 2019. Anyway to date Apple is still only making ultra low power W chips. WiFi 7 isn't scheduled to arrive until 2024 / 2025. So there is plenty of time for them to work on it if they decide to do so.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Apple Building Chip Team to Bring More Wireless Component Production In-House
- Thread starter MacRumors
- Start date
- Sort by reaction score
Broadcom knew this day would come. Broadcom has literally state they will rather sell their WiFi Business in 2019. Anyway to date Apple is still only making ultra low power W chips. WiFi 7 isn't scheduled to arrive until 2024 / 2025. So there is plenty of time for them to work on it if they decide to do so.
I see why this makes a lot of long-term sense for Apple. Just own the entire thing, end to end, and stop getting caught at the mercy of other chip-makers who want to sell to other people besides Apple.
My concern is that, first off? This could be really bad for the open source community. (Are you going to keep struggling to build a Linux distro that runs well on a Mac if you can't get any info on how to make the wi-fi chipset work, or the audio chipset, or ??) At least when chips are used across the whole industry, there's a strong incentive to get them reverse-engineered, if not to get their own creator to provide drivers for multiple OS platforms.
Second? I'm not sure if Apple will manage to keep the number of top-tier chip development experts employed they'll need to stay competitive with all of this? I mean, already, we're seeing where their M1 design seems to be inherently slower handling USB data transfers than Intel's solution. I think there's a lot of potential for more of these performance deficits in the new, proprietary Mac chips. Again, Apple only needs to design them to go in whichever Macs they need them in to sell them. Traditional chip-makers are selling them as solutions to ALL builders who need what they're offering. More incentive to spend a lot on the R&D up-front when something like your wi-fi chip might be relied upon in hundreds of different products from many manufacturers.
My concern is that, first off? This could be really bad for the open source community. (Are you going to keep struggling to build a Linux distro that runs well on a Mac if you can't get any info on how to make the wi-fi chipset work, or the audio chipset, or ??) At least when chips are used across the whole industry, there's a strong incentive to get them reverse-engineered, if not to get their own creator to provide drivers for multiple OS platforms.
Second? I'm not sure if Apple will manage to keep the number of top-tier chip development experts employed they'll need to stay competitive with all of this? I mean, already, we're seeing where their M1 design seems to be inherently slower handling USB data transfers than Intel's solution. I think there's a lot of potential for more of these performance deficits in the new, proprietary Mac chips. Again, Apple only needs to design them to go in whichever Macs they need them in to sell them. Traditional chip-makers are selling them as solutions to ALL builders who need what they're offering. More incentive to spend a lot on the R&D up-front when something like your wi-fi chip might be relied upon in hundreds of different products from many manufacturers.
The open source community LIVES for this kind of stuff, though.
I was looking over the notes regarding what they were doing to get Linux running with graphic drivers on an M1 and it’s REALLY impressive. And, they’re continuing to make progress.While there is some chip design work done in Los Angeles, it is San Diego and Orange County that really has the legacy (thanks to the Navy) of wireless electronics / semiconductor design. The real bulk of chip design is done in the Silicon Valley, Austin, and some in Research Triangle Park in North Carolina.
(Of course, there are teams scattered around the whole country, I'm taking about the "centers of mass"...)
(Of course, there are teams scattered around the whole country, I'm taking about the "centers of mass"...)
highly doubt it, qualcomm is another beast, outperforming intel wasn't a hard challenge consider they been sleeping since sandybridge, if folks think apple magically storm out with a modem that can keep up with QC they got another thing coming.
Hope that in-house means not in China.
Apple is creating a new office that will focus on wireless chip production as the company works to bring more of its chip development in-house, reports Bloomberg.
Apple is hiring a few dozen engineers for an office in Southern California to develop components that may eventually replace parts that are currently sourced from companies like Broadcom, Skyworks, and Qualcomm. The office is located in Irvine, California, which is close to Los Angeles and where major chip makers are located.
Based on job listings, Apple is seeking employees who have expertise in modem chips and wireless semiconductors, and at the facility, employees will work on wireless radios, radio-frequency integrated semiconductors, and semiconductors for connecting to Bluetooth and WiFi.Apple in 2020 signed a multi-year deal with Broadcom, which was set to last for three and a half years, which means it will expire in 2023. Under the terms of the deal, Broadcom supplies Apple with "range of specified high-performance wireless components and modules."
When the contract expires, Apple will no longer need to use Broadcom components and can instead rely on its own components.
Apple has been working to bring more of its chip production in-house to reduce its reliance on third-party suppliers. Apple is fairly far along on its development of a 5G modem chip, for example, and when work on that chip is complete, the company will be able to stop sourcing 5G chips from Qualcomm.
Current rumors suggest that Apple's modem chips will be ready for use in the 2023 iPhone models, so Apple will continue to use Qualcomm chips for the iPhone 14 lineup.
Longtime Apple supplier TSMC will manufacture the Apple-designed 5G modems for the 2023 iPhones, and Qualcomm has already acknowledged that it expects to supply only 20 percent of the modems for the 2023 iPhone, with Apple largely relying on its own 5G chips.
Article Link: Apple Building Chip Team to Bring More Wireless Component Production In-House
Apple already has their own WiFi chip, the W3.
This supports a/b/g and n at 2.4GHz. (And BT5.)
The aWatch6+7 also supports 5GHz. This suggests an updated W4, but no-one has discovered anything definitive to that effect. Cracking either watchOS [to look for code and driver names] or the physical watch SIP [to look for chip markings] are both non-trivial!
No doubt they do some impressive things to figure these problems out. My point, though, is that this just creates more hurdles for them to overcome. And in the past with Linux, there have been long periods of time where such things as 3D video card support for certain cards was really poor, just because they couldn't replicate all of the things in the proprietary device driver that allowed the chip to use its full potential.The open source community LIVES for this kind of stuff, though.
So there are a lot of gains from incorporating this on an SoC, but I think a big one is to remember that removing third-party dependencies is a big win—especially when supply chains are stressed like they are now. I mean, can you imagine how frustrating it would be for Apple to have an almost-ready MacBook Pro sitting around waiting for supply of some small component that only has a cost of a few dollars? Getting a lot of this tech in an SoC removes that kind of issue.
Qualcomm's "lead" in this instance is largely based on the fact that most of the 5G carrier side gear, is also their own. Not saying they don't come up with some new stuff on their own that is better than everyone else's, but their market position and performance is largely because they're on both sides of the equation. LTE vs WiMAX is a time where they went the wrong direction at first, and it showed others could do some really nice LTE modems as well. With 5G, between all the anti-China stuff and other politicking, they're trying really hard to make sure all the carriers are only running their gear, so that their modems are a lot easier to make "perform better than others."
Again, that's not saying their chips aren't the best for use right now, or anything of the like. Just that I think the context matters of why they're the "best."
The fact that they are actually hiring is news, but the fact that they would insource this work is not. When they inked the settlement with Qualcomm last year, the feeling was that Apple would eventually (4-6 years out) insource these chipsets. Seems like they are more or less on that schedule. Be interesting to see how much of the M1-type magic they can bring to this part of the stack.
Apple in-house 5G modems won’t be able to compete with Qualcomm. They just can’t . Stock up on all Qualcomm iPhones now
I love the pessimistic nature of this comment lol.
Anyways, next step I’m thinking is will apple branches out their modem business and build the so-called Apple 5G using apple modem and apple owned cell towers, directly competing against giants like T-Mobile, AT&T etc, and only offer basic feature for non-Apple devices. That sounds nice.
I’ve been wondering how the timing of these new in-house chips will coincide with the release of a VR/AR headset. Maybe this new device will be the first to use them? I’m not expecting this to be out until 2023.
I foresee an exodus of people from Broadcom to this. They put this in Irvine for a reason, stones throw from Broadcom. They did this too with the old Intel locations in San Diego, it was one exit up from Qualcomm.
One interesting note about Apple building its own chips: generally speaking the broadcom stack is really inefficient. There are layers over layers over layers, and nobody there really cares about performance per se.
Apple could get a performance win here. Maybe it's only going to be 15-20%, but that's better than nothing.
Apple could get a performance win here. Maybe it's only going to be 15-20%, but that's better than nothing.
Going by all your comments, you certainly don't come across as an engineer. If I'm wrong, then that would be a surprise, and make you a rather fascinating character. However, I'm calling b*llsh*t.
Ah good, a Broadcom employee.
Your guys’s drivers are terrible. Windows and Mac.
Oh, certainly there will be new more complex hurdles, but I think we’re getting down to the point where the actual driver requirements may be different, but what the chip is doing will be easier to reverse engineer for these “commodity” type functions. At least while Apple has no crypto lockdown on them.
I’d reckon they’ve got the initial baseline implementation up and going and they’re hiring the folks that will be expected to iterate on their performance over time. So, while this effort won’t yield near term results, definitely by generation 2 or 3.