Checkmate, Qualcomm: Apple in billion-dollar bid to gobble Intel’s 5G modem blueprints, staff – new claim

Apple and Intel are apparently in “advanced talks” over buying up the remains of Chipzilla’s defunct 5G modem business. Sources close to the deal told the Wall Street Journal on Monday that Cook & Co were offering the x86 processor goliath a billion dollars for the intellectual property and staff behind its cellular modem …

[Anonymous Coward]

Good Luck with That

The question is: is the Intel silicon as good as Qualcomm silicon? Or at least, nearly as good? An NG modem is somewhat of a black art combined with a thicket of IP. If Intel cross licensed with Qualcomm, that is valuable, avoiding another IP black hole for legal payments. Next is avoiding paying Intel profit margin, although this balances against the alternative of paying somewhat rapacious Qualcomm profit margin.

Finally, is Intel silicon good enough? Huawei works, and Qualcomm works, but does Intel work also? Did Intel burn enough cash paying for development to be comparable to the competition, or is Intel a wannabe that is not quite there... and did key engineering talent stick around hoping for an Apple buyout?

My guess is Apple will buy the division and eventually toss the dross and keep some of the talent, perhaps not losing too much of the investment due to trading IP with Qualcomm-- at least for 5G. 6G is another battle.

[big_D]

Re: Good Luck with That

That was part of the problem, they were slow, compared to Qualcomm - and Qualcomm had a go at Apple for artificially throttling the Qualcomm chips, so that they ran at the same speed as the Intel chips.

Likewise, Intel were struggling to get 5G up and running, which is probably the main reason they backed out of the market. Their technology just isn't mature enough, yet.

With Apple's money behind the project, it might make a competitor to Qualcomm, but only for Apple. It is a shame Intel are backing out, selling the technology to either Qualcomm or Apple is not good for the open market.

[Anonymous Coward]

Re: Good Luck with That

Realistically there's no one other than Apple who would be developing a NEW cellular modem in 2019. The only other customers for Intel's stuff were companies like Qualcomm, Samsung, etc. who would want the IP to add to their patent warchests. The open market was never going to gain a competitor to Qualcomm.

[big_D]

Re: Good Luck with That

Yep, that is the problem.

[ctr00001]

Re: Good Luck with That

Except for maybe Samsung who have launched commercial products already. And Huawei HiSilicon. And in 2020 maybe Mediatek. Spreadtrum not far behind that. There are others as well.

[Wade Burchette]

Re: Good Luck with That

If Intel, with their deep pockets and experience making cellular mobens, had trouble with 5G chips then what chance does Apple have? If Apple buys Intel's 5G division, they will be buying their patents and engineers. The same engineers that did not succeed with 5G. Apple would have to hope they can hire talent. They can certainly afford to. But will these people leave Huwai or Qualcomm? Buying up Intel's division is no guarantee of success.

[tip pc]

Re: Good Luck with That

its all about creating competition. If there isn't a Qualcomm competitor then Apple risks higher Qualy prices or being slow to market behind Sammy or WhoRU.

[Anonymous Coward]

Re: Good Luck with That

People said the same thing when Apple started designing their own CPU cores - how can Apple hope to compete with companies that have been doing it for decades. It didn't take long before their cores were easily the best ARM cores, and today despite their low power draw they're competitive with all but Intel's highest end desktop chips.

Designing modems is actually a much easier proposition hardware wise than designing CPUs, the DSP and baseband software is the overwhelmingly more difficult part of designing a good modem. The reason Qualcomm's modems work so well is because they are so dominant that they're the defacto standard for what the companies writing the software in the towers test against. It is like when IE dominated the web and everyone was writing for and testing against IE. It wasn't that IE was better, it is that it was dominant.

Qualcomm's baseband software is also littered with exploitable issues (some believe put there deliberately at the rest of the NSA) so Apple controlling their own baseband software is the most important part of this for Apple in the long run.

[Charlie Clark]

Re: Good Luck with That

It didn't take long before their cores were easily the best ARM cores

Strawman but also untrue. Apple's ARM-based chips do incorporate lots of features but the CPU stuff is the least interesting. Where they do win plaudits is the GPU. But basically they add stuff to the ARM design.

But that's totally irrelevant when it comes to modems. Intel has great chip designers and failed with Infineon because radio chips are completely different beasts and this is about making chips from scratch rather than adding hardware to the ARM baseline. And, when it's done, you don't even try to sell the chips to the competition.

If this report is true, then it's probably all about whatever cross-licensing deals Intel had with Qualcomm. Of course, if it wanted to, it could just buy Qualcomm.

[rcxb]

Re: Good Luck with That

If Intel, with their deep pockets and experience making cellular mobens, [sic] had trouble with 5G chips then what chance does Apple have?

Apple has deeper pockets, a deep-seated interest in preventing a monopoly in this space, and a guaranteed market for their output. When Intel was the supplier, Qualcomm just had to undercut them on price to win over Intel's customers. That's quite a bit harder for Qualcomm to do now that the supplier and the customer are one and the same.

[Anonymous Coward]

Re: Good Luck with That

Qualcomm publicly accused Apple of handing over Qualcomm's trade secrets to Intel. Intel probably got spooked. I have myself worked in a company where tainted IPR got into the company, twice, and these things are far messier then most people realise (hence anonymous). Intel got a lot of IP from Apple, but now could not know what was legitimately theirs to give. Under the circumstances, selling the lot to Apple to deal with was probably the easiest way for Intel to get out of what could have ended up in a decade long litigation.

[aks]

Re: Good Luck with That

It's presumably not an open market. If it were, a Chinese company would buy these IPR's.

[Anonymous Coward]

Re: Good Luck with That

At the last count, Apple had a $245 billion cash reserve. If they want this to happen because they've had it with Qualcomm then it will happen. The only question is will Qualcomm go quietly or will they start throwing grenades out of their pram.

[Anonymous Coward]

Re: Good Luck with That

Apple had already been working on their own modem, and knew that Qualcomm was going to go after them when they finished it. By buying Intel's cellular IP they have a much better chance to defend themselves. They won't have as much as Qualcomm, but they'll have a lot more than they have today (they acquired Nortel's old IP some years ago, which includes some LTE patents but obviously no 5G)

[Anonymous Coward]

Re: Good Luck with That

As a former Intel minion, I can tell you the biggest problem for Infineon, was Intel !

It drove them nuts the way Intel demanded that Infineon spend time doing things, 'The Intel Way', rather than using the limited time they had to tape out a 4G design, on getting the damn product working !!

Intel kept on asking to add feature that were zero value, & just caused delays.

It didn't help either that smack in the middle of a critical phase of design, Intel had another round of layoffs, & a bunch of experience people, on the Intel side, took the money & ran.

[Anonymous Coward]

Apple buying IP and talent. What happens if the talent decides not to opt-in? Is the deal scuppered?

[Anonymous Coward]

That's a risk you always take when buying a "team", but I'm pretty sure Apple is mostly interested in the IP. They already had their own team working on it, getting Intel's IP and designs will help accelerate that process by a few years - the hardware is easy it is the baseband software that's the complex part of a modern modem.

Probably the reason why there were rumors Intel was going to "auction" the business is because they wanted to get Apple worried that they might not be able to get that IP. Maybe they got Apple to up their offer and that's why the deal is supposed to be close to done.

[phuzz]

I suppose it partly depends on where Intel's modem team are/were based, and if they fancy moving to Cupertino (or Apple opens an office where they are).

[Anonymous Coward]

Yes they will

I know of one specialist team that defected from another company to Apple, who built an office for them in Texas. They did plenty of commuting to Cupertino, but their main office is still at "home".

[doublelayer]

I imagine the current team is in limbo at Intel right now. Going to work for Apple is a pretty good guarantee that they'll continue to have a job. They can quit, but there are not a ton of other modem manufacturers to go to. If someone only bought the IP, I imagine Intel would probably transfer a few of them to other divisions, and the rest of the team would be seen as unnecessary. So I think most of the team will probably be happy enough--they still have jobs in modem design if they want them.

[Anonymous Coward]

As long as they don't use their CPUs..

The performance hit of countering all that Spectre malarkey (which backdoor, it hardly needs saying, nobody asked for) is significant so I would very much welcome Apple et all to tell Intel to F off. We're actively investigating to switch procurement to AMD exclusively, but still need to work through consequences. Fewer suppliers is one thing, longevity is another (no long term exposures yet, so we'll have to look at spares and risk management).

Overall it looks that it's pretty much thumbs up for renewals to go AMDs way, but we can't do that with Apple gear..

[big_D]

Re: As long as they don't use their CPUs..

We are talking about cellular modems here, nothing to do with Intel's CPU business... :-S

[iron]

Re: As long as they don't use their CPUs..

Plus Apple already use Intel CPUs in their desktops & laptops and obviously are not about to ditch their home grown ARM CPUs in mobile devices.

[Robert Jenkins]

Re: As long as they don't use their CPUs..

But possibly not for long - don't you read The Register ? :))

Apple Macs will, supposedly, be using ARM CPUs within a year or two.

https://www.theregister.co.uk/2018/10/18/apple_to_dump_intel_again/

[sw guy]

Re: As long as they don't use their CPUs..

Off course there are CPUs involved, though rather of the DSP family.

Very hidden, very specific, but a modem (whatever kind) needs lots of computational stuff.

[Duncan Macdonald]

Is it worth the cost ?

One billion now - probably at least the same again in development costs to get a 5G modem that matches Qualcomm. Modems derived from the Intel technology would probably not make it into Apple phones before 2022. Is the cost of the Qualcomm modems high enough to make this cost worthwhile.

There is also a question - how many of the top designers left Intel when they announced that they were exiting the modem business?

[Anonymous Coward]

Re: Is it worth the cost ?

For the IP alone it probably is, because Apple will need a big patent warchest to defend against Qualcomm's inevitable lawsuits when they ship their first phones containing their own modem. Apple has already been working on designing their own modem, and had already hired some of the Intel people as well as some former Qualcomm employees.

I doubt they are all that interested in most of Intel's employees, those 'top designers' who left Intel when they announced this probably sent their resume straight to Apple already. The run of the mill guys they can get anywhere - most of Intel's modem design team was in India and I doubt Apple is interested in having the work done there.

[JetSetJim]

Re: Is it worth the cost ?

Apple now has $245 billion cash on hand, so shouldn't be a problem for them - although whether the Intel designs could be brought up to compete properly with QC designs (IIRC Intel designs were a bit shit when they were stuck in an iPhone, as others have pointed out).

Then QC will sue Apple for patent infringement, and about 15 years later that case will be decided after repeated appeals & refilings.

[YourNameHere]

Re: Is it worth the cost ?

From scratch, ignoring the patent issues, its many many billions of dollars and years to design, test, roll out to all of the different countries with different frequencies, different versions. This is a complete nightmare infrastructure.

[Anonymous Coward]

Re: Is it worth the cost ?

They don't need to be the equal of Qualcomm's designs, just "good enough". Does anyone bother benchmarking the different wifi chips in various phones to see which ones are the fastest? So why should they bother with cellular, because as with wifi once you are in the hundreds of Mbps any difference no longer matters.

[ctr00001]

Re: Is it worth the cost ?

Fastest means feck all in cellular baseband design. It the squillions if other things that matter and take time and money. Lots of it. Much lots.

[Anonymous Coward]

Slow...

Does Intel being slower than Qualcomm’s modem matter?

Who is getting anywhere the theoretical or promised speeds anyways? I know I’m not... that’s on the carrier.

Sign me up for real LTE speeds, then who needs 5G.

[el kabong]

Lower latency

Network latency matters as much as network speed, for highly interactive applications latency is even more important than speed.

5G will be able to achieve much better latency than LTE. 5G will require lower power too.

[Justthefacts]

Re: Lower latency

Will 5G achieve significantly lower latency in practice? It’s not obvious.

To start with, the radio latency of 4G from scheduling is up to 20ms, reduced to 1ms for 5g. BUT clever scheduling can be achieved in 4G down to 2ms, and that has been demo’d already by Huawei.

In practice, the bigger contribution comes from the delay within the core network and public internet. So the question is how much difference the change in core network architecture will make. Many more nodes, more virtualisation, potentially fewer or more node hops. So the classic engineering answer is “it depends” as usual.

[adam payne]

Sources close to the deal told the Wall Street Journal on Monday that Cook & Co were offering the x86 processor goliath a billion dollars for the intellectual property and staff behind its cellular modem business

£1bn to buy the company, another couple of billion in R&D to get the modems up to a similar level as Qualcomm. Then of course a couple more billion for the lawsuit settlement after they apparently infringe on some Qualcomm patents.

[DontFeedTheTrolls]

"... dump Qualcomm and its ilk, and save itself a packet on royalties and licensing payments"

Isn't one of the problems that Qualcomm own the necessary patents, so Apple will still need to pay even if they own the manufacturer?

[Anonymous Coward]

Correct. The journalist didn't understand how patents work.

Qualcomm owns both standard essential patents as well as non-essential patents that relate to performance and battery improvements. Apple cannot afford just the standard essential patents from a premium vendor point of view. Especially after battery gate they want to make sure batteries last a long time.

Anonymous, since I had a connection to one of the parties.

[YourNameHere]

Repeat, do again.

Apple is late, they could have bought Intel's last modem business which the sold to Marvell about 13 years ago for 25 million or less. This is the cell phone business that Intel sold to Marvell after spending billions on it and then decided to sell it for 625 million. Wonder how many billions Intel spent on this version of the modem? Marvell sold it a couple of years ago for something like 25 million or so.

[Anonymous Coward]

Re: Repeat, do again.

Why would they have purchased Intel's Marvel business before the iPhone was announced? They hadn't even bought CPU design companies back then, it took a few years for them to reach the scale where it started to make sense to design their hardware. Sure it would have been nice for them if they had been designing their own modems from day one, but it isn't realistic.

[LeahroyNake]

I don't get it.

I feel a bit like Tom Hanks in Big saying this but I don't get it.

5G is a documented standard just like umm 4G, 802.11.a/b/g/n/ac, ethernet 10 100 1000 etc. A lot of companies can make them work so why is 5G that hard in comparison and if it is that hard then why is there not another easier to implement, cheaper, version?

Or maybe Qualcom is on the standards board and pushing thorough specifications that it knows it has IP rights with that others cannot copy to make it work 'as good' as they can.

Its an honest question, I have almost no idea how RF / electromagnetic waves work beyond making an AM receiver / radio in school electronics class over 2 decades ago and I'm quite sure I didn't violate any IP in the process.

[Anonymous Coward]

Re: I don't get it.

The baseband software is extremely complex. Hardware wise a modem isn't that difficult, it is far simpler than designing a CPU, it is the software that makes a modem so much harder than wifi.

While yes there are 'standards', a lot of it is ad hoc. It is more like writing a browser back in the IE6 days when webservers treated standards as "what IE6 does" rather than what the specs say. A lot of it ends up being "what Qualcomm does".

[YourNameHere]

Re: I don't get it.

The variation in the spec is huge from company to company, repeater to repeater, country to country. Then each country/company has to certify your chipset/Comm SW stack. And each country is somewhat different. It's a nightmare

[Anonymous Coward]

Re: I don't get it.

Good point. The modem is an accelerator for a complex, real-time software spec that was more or less invented by Qualcomm in the 3G/4G LTE standards and openly licensed.

Qualcomm could do a better job improving the transparency of their patent thicket and licensing granularity to better account for what technologies are being used by third parties and account for expired patents, but you sound like you're saying that doing "what Qualcomm does" is inherently gatekeeping and devoid of value. Their methods do conserve spectrum, improve reliability, and conserve power; they also can be implemented in a way that can be crammed into a small space into a client device.

The software methods are what's being licensed and various enforcement authorities and judges utterly failed to understand this in their rulings. This is all very much preferable to not declaring specs into standards and sitting on that IP. Intel and Apple do this for example with processor instruction sets and app stores, stifling competition and driving prices sky high. (One wonders if regulators forced an iPhone into existence with an open store standard whether it would retail for more or less than an Apple App Store exclusive mode.)

[HmmmYes]

The story isnt Apple or Qualcomm here.

Its Intel and Intel's inability to create new silicons.

Intel area one or two trick pony. Sure, their fabs n processes are great but shit design in, shit silicon product out.

The question of Intel stockholders is - Why cannot Intel design good silicon products? Where are all the chip designers? In an age of fabless semi's Intel is rapidly approaching irrelevance.

Intel are painting themselves into a corner, clinging to Xeon.

[aks]

The up and coming chip designers are in China and India.