Qualcomm to buy NXP for $38B in largest chip deal | Hacker News Reader

Qualcomm to buy NXP for $38B in largest chip deal

reuters.com

165 pointsghosh72 comments

Everyone who maintains designs with NXP parts, please take notice: this is a good time to download and keep all the documentation on these components.

Qualcomm's track record on releasing and delivering documentation is among the worst in industry. If you are not a big buyer they vetted, they won't even answer your emails or send you the NDA. And after NXP bought Freescale, previously available documents went missing. So, all care must be taken.

This x100... I totally echo this experience..

I just finished up a design with references to NXP documents, blog posts, etc... Time to go through and capture all of that info.

I honestly can't believe they and Broadcom continue this idiocy. Makes no sense whatsoever. A lot of small gigs I've worked used freescale parts, and don't have the leverage to get the datasheets. How absurd a situation.

Totally agree. Freescale, NXP and TI are all great at documentations etc. Qualcomm and Broadcom and Marvell are all notorious on that regard, in most cases all you can get is a product brief unless you're their top 10 customers, otherwise they probably don't even bother sending you a NDA. Broadcom was bought up, Marvell is going nowhere, Qualcomm is the exception though itself is undercut by MediaTek/Huawei these days heavily, which is probably one major reason for this deal. Really sad to see Qualcomm swallows in NXP/Freescale.

MediaTek has quite a few leaked datasheets out there on various Chinese forums, and so do RDA and some of the other lesser-known SoC makes; I haven't seen the same for Broadcom nor Qualcomm. Maybe it's the different culture ( https://www.bunniestudios.com/blog/?p=4297 ) that facilitates this.

It's a shame. We have made many designs based on Freescale parts, and a large reason we chose Freescale was documentation. The NXP acquisition of Freescale seemed to largely maintain the status quo, but I am worried about qualcomm..

Maybe this is a good project for ArchiveTeam. Looking at some google searches, there looks to be about 100k PDFs to download from NXPs website. I'm sure the majority is data sheets and application notes.

I'd be willing to help download this data if there was an collaborative organized attempt to it.

I would also be more than happy to donate all 200mbit of my available bandwidth to this effort. Maybe some folks who are much more talented at scripting than I can figure out a way to scrape a list of URLs to the PDFs.

Where would one find an index of these PDFs?

NXP's docs are all over the place and they change when there are big changes to the product lines.

Working happily with NXP... Hope this won't affect things. The documentation and cooperation are all great so far.

Yes I too feel the same way NXP and freescale have been go to SoC vendors for me - their great and easy to find docs have been the big plus for me - frankly if I can't go to Digikey or Mouser and click on that "data sheet" link I'm probably not going to use your chip

I guess it's time to go an have another look at that NXP chip I just qualified

Would it be possible for you to pop into #archiveteam on EFNet? I'm not sure where to start with getting the PDFs - this may be a good project for http://archiveteam.org/index.php?title=PDF_2016 but they'd need a (large) list of PDFs.

"Everyone who maintains designs with NXP parts... this is a good time to download and keep all the documentation"

That's kind of funny because I've had some bad experiences with NXP when it comes to documentation for some specific chips. Although, NXP does provide a lot of documentation related to technologies in general. That stuff is easily available.

You can bet that the documentation for older parts has already been uploaded to the numerous datasheet sites out there, but I agree that redundancy is always better on the safe side.

When NXP was spun out of Philips in 2006, everybody around the businesses were a tad grumpy about it. Philips got paid a lot for NXP by the private equity folks who took it out, who wrote it down as a loss on NXP's balance sheet.

So NXP started its independent life deeply in the red numbers. Doesn't seem nice, does it? People were wondering whether the thing was just gonna be chopped up, sold on, and half the people fired.

Instead, however, people had to be much more scrappy than before. The new bosses weren't tech geeks but money sharks. Don't do things that don't have a clear path to profit. Don't waste time. Don't tolerate people wasting their time. Of course, this seems entirely obvious to all of HN, but anyone reading this familiar with Philips culture knows that things aren't always that way. Philips and NXP had (have) plenty of people who were rightfully referred to as furniture. They were there for decades, same desk same chair, and nobody really know what exactly they were doing. Somebody, somewhere, just kept paying the salary.

NXP's new need to get back into the black numbers was truly transformative for the company. Sure, it's still a bigcorp with all the quirks you'd expect, but they came a long way.

I'd not have predicted a Freescale-level purchase 10 years ago, and I definitely would not have predicted a $38B valuation. That's more than Philips' current market cap ($26B) and NXP was spun out because it was a loss leader that they couldn't turn around.

Well done, folks! As a fellow Eindhovener, I can't help but be a bit proud. Too bad you didn't buy Qualcomm instead but you can't have it all I guess!

I do wonder what this means for the wider industry. The fact that they finally got bought means that other methods to grow revenue are failing (compared to, say, 2-3 years ago), and now directly buying revenue has become the dominant way to grow.

When reading financial textbooks they talk about how this happens in the last stages of the business cycles, because it makes sense at those points : economy steadily worsening, but only slightly (easy to mask on company results, so companies appear attractive, but management knows they will be significantly less so in a year's time or so), combined, especially today, with the massive availability of cheap credit, and the idea that "rates will be lower for longer" (history, of course, teaches us they won't).

The purpose here has to be to conquer more of the market by limiting options and moving towards more consolidation.

I'm not so sure this should be allowed. If Qualcomm buys NXP that means they'll own Freescale. That would reduce the number of mobile phone processor manufacturers down to just Qualcomm, ST Micro, Nvidia, Apple, and Samsung. Realistically this means it will be just Qualcomm and Samsung covering almost the entire market of Android devices as the other manufacturer's chips are more niche.

I suppose maybe Marvell could catch up and wind up in more phones but they're already the bargain basement chip that winds up in cheap junk tablets. They'd have a long way to go.

Intel could also make a sudden change of course and enter the market. Their foray into tablets didn't work out well for them (I own several Intel Android tablets) but at least their performance was on par with the competition at the time.

There's also Mediatek [1] who are very big with Chinese phones and a couple of the cheaper Sony phones such as the XA [2].

There's also the Kirrin line that are owned by Huawei [3] and used in their newer phones.

1. http://www.mediatek.com/

2. http://www.sonymobile.com/gb/products/phones/xperia-xa/#spec...

3. https://www.cnet.com/news/huawei-unveils-blazing-fast-kirrin...

> I'm not so sure this should be allowed.

I agree. Competetive reasoning aside this would leave ST Micro as the last wholly European player in that market (and NXP already is the last major one in the EU, ST being based in Switzerland). In theory Infineon could re-enter the mobile processor and memory market, but I think they left that turf for good, 10 years ago.

So I think EU authorities might have a word in this.

You are missing Mediatek. Which is hitting Qualcomm hard in low to mid range SmartPhones.

What products does ST Micro have in the phone processor space? Wasn't NovaThor cancelled?

I can't recall any recent design wins for NXP and/or Freescale in the smartphone market.

NXP really had no Cortex-A solutions before acquiring Freescale, and Freescale's i.MX series hasn't been in a phone for years, if at all. They did have some early wins in mobile books, but not much more.

NXP is now targeting the i.MX line for automotive HMI/IVI use. The automotive-class chip domain has always been a specialty of NXP (nee Freescale nee Motorola Automotive nee Motorola Semiconductor)

I believe NXP has an NFC chip in a few different iPhone revs.

But I think NFC chipsets are noise in this acquisition, I think this is all about QCOM missing the boat on automotive and buying their way into this market via NXP.

The real loser here is Intel who really doesn't have a compelling automotive story (aside from maybe Altera) and risks loosing to QCOM in handsets AND automotive.

I agree that automotive is the play here.

Automotive isn't just about the application processor showing rear-camera video and 3D maps, there are a lot of other pieces needed and Moto/Freescale has had that market share for a very long time (at least with the American automakers).

QCOM is currently making most of its profits from 3G and 4G licensing, and those royalties have been threatened by poor manufacturer cooperation and protectionist regulatory actions in China and South Korea.

Buying NXPI (and its many technologies) gives QCOM an escape path to diversify away from a maturing mobile handset market, with many foreign manufacturers in countries with a different view on IP.

Which does not matter.

Which matters more is, that NXP and Freescale (now part of NXP) are strong in automotive, which is not so much niche you think it is. While there is a smartphone comparable processor in the head unit, it is surrounded by about 40 to 60 Cortex-Mx processors and comparable.

Then think about the growing market of IoT, which also benefit from those automotive Coretx-Mx processors and their cheap prices.

The smartphone market is done and settled, with small improvements. But the IoT market is the one which will be disruptive.

There was massive amount of questionavle mergers lately: Altera + Intel, Fairchild + Onsemi, now this. We stepping into the age of monopolies, dear readers.

Altera + Intel? I get it, Altera was a 17B dollar deal, but come on. Intel buying AMD would be more cause for concern about monopolies, and that would be a far smaller deal.

What about Allwinner?

qualcomm financials for reference:

		+----------------------------+-------------+-------------+----------------------+
		|                            | Fiscal 2015 | Fiscal 2014 | Year-OverYear Change |
		+----------------------------+-------------+-------------+----------------------+
		| Revenues                   | $25.3B      | $26.5B      | (5%)                 |
		+----------------------------+-------------+-------------+----------------------+
		| Operating income           | $5.8B       | $7.6B       | (23%)                |
		+----------------------------+-------------+-------------+----------------------+
		| Net income                 | $5.3B       | $8.0B       | (34%)                |
		+----------------------------+-------------+-------------+----------------------+
		| Diluted earnings per share | $3.22       | $4.65       | (31%)                |
		+----------------------------+-------------+-------------+----------------------+
		| Operating cash flow        | $5.5B       | $8.9B       | (38%)                |
		+----------------------------+-------------+-------------+----------------------+

wow.

Wow is right. I had no idea they were doing so poorly!

Did some of their key patents expire or something? Or maybe we've hit "peak phone/tablet"? Haha

I am so shocked you call this poorly. Well I am sure americans are used seeing incredibly big numbers but for me this is hardly something I would call a poorly results.

Gosh...the entire company has more money than some countries.

Year over year they lost practically a third of everything across the bank.

For a technology company big negative changes like this can be very bad, they can in fact signal that the company will go bankrupt or essentially so, e.g. forced to sell itself at fire-sale prices, in at little as a year or two.

Their revenue went down 5% which is never a good thing but not a "stop the world" moment either. We don't know why their income crashed, but that might as well be R&D. It's still in a healthy place relative to revenue..

If their margin went down because of massive advertising/discounts yet they still lost 5% of revenue, that would be worrying. But we can't tell from the table alone.

there was this $Billion penalty in China

http://www.wsj.com/articles/qualcomm-settles-china-probe-142...

followed by 'voluntary' investments and 'partnerships'. Intel faced similar fate and was forced to open fab and inject almost two billion into chinese consortium Spreadtrum.

Year-OverYear Change Minus signs are missing

A number in parentheses is negative according to accounting standards. (34%) = -34%

TIL a number in parentheses is negative :)

Pretty standard practice in accounting + book keeping. You'll see pretty much all balance sheets represented in this way.

Do you know what's the reasoning behind this convention? I had no idea too that percentages in parenthesis mean negative. It's a whole new world.

Parentheses are easier to see than negative signs when scanning through large tables of numbers. Also, in the days when financial documents were transcribed or faxed, it was harder to lose a set of parentheses than a negative sign when sharing documents.

You have taken the first step towards mastering your 10K analysis abilities. Report back once you've mastered the concept of how assets, liabilities, and equity balance out [1]. Then we can move on to the principles of revenue and expense recognition so we can begin to read income statements :)

[1] http://www.principlesofaccounting.com/chapter-1/accounting-e...

It has a hard year but I think Fiscal 2016 has to be better -- nearly every major phone has a new Snapdragon in it except for Apple's.

So Qualcomm is finally getting pressured by the other vendors in mobile space. Finally.

I'm glad Intel, HiSilion, Rockchip, Allwinner, Spreadtrum, and MediaTek are being competitive.

For those not aware:

NXP subsidiary Freescale was originally the semiconductor division of Motorola, which developed the first ever commercial high-power transistor, supplied much of the computing technology to the Apollo program, and developed the MC6800 and MC68000 processors that helped spark the modern computing, gaming and consumer electronics industries.

Quite a legacy Qualcomm is aquiring.

Freescale comes from Motorola

NXP was originally Philips

They merged on 2015

It's funny I just learned what Freescale was 3 days ago. I wanted to integrate this system I'm designing with employee presence so I bought and rooted this time clock https://www.amazon.com/dp/B0155NL3MA Inside was a MicroSD card on an ARM board that contained a copy of "Freescale Linux"

Qualcomm intends to fund the transaction with cash on hand and new debt.

As a layman, I'm surprised a company can simply raise more than a third of their market cap. At this level, isn't debt almost an investment in the company? If revenue falters, how can the lenders make sure they're made whole?

The debt can be secured against the assets its buying, and if not explicitly secured, debt holders are still pretty high on the pecking order in bankruptcies.

A major reason to do it is that in the US debt is massive tax advantaged compared to things like stock, and a major reason for entities to lend out money for something like this is how poorly other less risky vehicles are paying in this era of financial repression (http://lexicon.ft.com/Term?term=financial-repression or https://en.wikipedia.org/wiki/Financial_repression).

Their ip business is ridiculously profitable. Every cell phone sold whether they make it or not they get a cut. So no one would have issues lending to them.

I have every expectation it's "ridiculously profitable" (they don't, for example, reveal all the details of achieving good power consumption in their patents), but it's clearly not a big enough part of the company to avoid the terrible results baq highlighted in this comment: https://news.ycombinator.com/item?id=12804392

And it's very common for a company with a great cash cow to bankrupt themselves by going out of their area(s) of core competence ... which this acquisition might be a doubling down on. I.e. isn't Qualcomm fabless, while NXP has 8 sites according to this list: https://en.wikipedia.org/wiki/List_of_semiconductor_fabricat...

  "they don't reveal all the details of achieving good power consumption in their patents"

In general, that would mean the feature is not patented: the choice is usually between patent an invention (but it must be explicitly described), or keep it secret (but if anyone is able to copy it, they're allowed to do so).

Patents are essentially an award of a short-term monopoly, as recompense for publicly disclosing an invention. If the invention is not sufficiently explained, then it's merely a trade secret, and can be copied without challenge.

Even so, we're seeing more and more companies use ridiculous amounts of debt for acquisitions.

I think this is pretty clearly a consequence of dirt cheap debt, which may lead to another economy crisis in 5-10 years, because not all of these multi-tens of billions of dollars debt deals for acquisitions are going to be successful, and when they'll crash, they'll crash hard.

Bond rates and interest rates in general are ridiculously low, so I suppose taking on debt isn't going to be a problem.

The scale of these acquisitions is really incredible. Qualcomm is acquiring huge company that itself recently acquired its own huge company.

How does management successfully integrate these merger of mergers?

So it's going to be Qualcomm and Nvidia going head to head for automotive chipset market?

Whomever spins smartphone/tablet ASICs will likely be a viable contender in that race.

Don't forget about HiSilion, Rockchip, Allwinner, Spreadtrum, and MediaTek.

I guess many vendors will have ASIC co-processors for some ML operations pretty soon.

The ones you listed are giving Qualcomm a run for its' money in mobile space, but automotive is a tough one to get into. Nvidia and especially Freescale/NXP already have automotive market share (and investment in the space)

I'm just curious what happens to Freescale's i.MX product line...i.MX directly competes in the automotive SOC space with QCOM.

I hope it survives, I've always liked the i.MX line.

Feels like deep learning is here to stay so I don't see how they are going to have any fighting chance.

Also Renesas (itself a combination of NEC, Hitachi and Mitsubishi's semi division) is popular among the Asian manufacturers. Both auto OEMs (Honda, Toyota, etc.) but tier-1 suppliers (Pioneer, Fujitsu Ten, Panasonic).

There is a sizable NXP facility in the city I live in. I'm really hopeful that this winds up strengthening their business here, rather than resulting in yet another closure.

This is some extremely disappointing news. Hopefully Qualcomm's awfulness won't seep into freescale.

Damn. I wish this was AMD instead. I guess they see IoT as a bigger/easier opportunity than fighting Intel.

> fighting Intel

At the close of the 19th century, the Jules Verne novels occasionally featured an 'arms race' for the biggest possible cannon.

Then in the 20th they just slapped smaller ordnance on tanks and planes. Mobile platforms ...

I understand, but there's still a whole lot of value Qualcomm could capture going after Intel's much more profitable businesses, like the server market. Going "up-market" as Clayton Christensen put it. If everything goes well with IoT and this acquisition, Qualcomm may still do that, but not expecting it to happen for another 10+ years now.

Qualcomm buying AMD would be awful, they are a 100% closed vendor, and wont even supply spec sheets/chip details if you arn't a $1mil+ a year customer.

They completely ruined Atheros wifi chips for instance. Pre-buyout they had the best featureset, and even open source firmware. Now they have a massive closed firmware driverlike-blob and you can't even set your MAC address. Pathetic.

AMD can't be bought. The x86/x86_64 cross licensing agreement forbids the sale of AMD (or Intel) to a third party.

If this happened, AMD wouldn't be able to make x86 and Intel wouldn't be able to make x86_64 anymore. [1]

Also Intel is very very keen to have AMD continue existing and manufacturing x86 CPUs because if they don't, the antitrust boffins will come after Intel.

[1] www.kitguru.net/components/cpu/anton-shilov/amd-clarifies-cross-license-with-intel-change-of-control-terminates-agreement-for-both

Reply on news.ycombinator.com