Qualcomm agrees to acquire semiconductor biz NXP for $47bn

Chip designer Qualcomm has agreed to purchase Dutch rival NXP Semiconductors for $47bn. As rumoured, the San Diego-based business and its Eindhoven-based target announced on Thursday that they had come to a definitive agreement to see Qualcomm acquire NXP in an all-cash deal worth $47bn, including debt. NXP Semiconductors is …

[BillG]

Reasons

NXP Semiconductors is the fifth-largest non-memory semiconductor supplier in the world, and the primary provider of chips for the automotive market, which Qualcomm is considered to be eyeing in expectation of growth in the self-driving car market

Doubtful this is the reason, as despite what you read in the news, we in the semiconductor market do not expect the self-driving car market to ever be mainstream - that's what our automotive customers are telling us.

This is just part of the continuing consolidation of the semiconductor marketplace. With microcontrollers being sold at commodity pricing and so those margins getting thinner, the fastest way to expand while preserving margins is to acquire the competition. It's not about self-driving cars, it's about survival.

[short]

Re: Reasons

There's plenty of silicon in automotive even without all this self-driving bollocks. Cameras (and image warping, because you can't put the cameras where you want), an ever growing amount of back seat entertainment, motorised everything, loom weight reduction, voice and data comms, location services, advertising, post-crash emergency calling, insurance tracking, government tracking, doomed attempts at securing all the above, ever-tighter emissions and efficiency control (and motor control for traction motors)

Part of me thinks it's a decent fit. Part of me looks at the scanty documentation Qualcomm publish at the moment, and fears for the future of NXP/Freescale stuff. Qualcomm are used to a small number of devices and customers, but massive volume. NXP, not so much. Time will tell.

[Adrian 4]

"The acquisition dwarfs that of chip designer and supplier Avago, which bought chip designer Broadcom in a cash-and-stock deal worth $37bn,"

I wouldn't consider $47bn to 'dwarf' $37bn. Surely dwarfing requires at least 100% higher.

[short]

<rummage>

couldn't find the average height of dwarves, fictional or dwarfism-style However, there seems to be an accepted USA threshold of 4'10" (58"). The average USAian is 5'7" (67")

67:58 is 1.16

$47b:$37b is 1.27

Unless my maths / stats / input data is wrong, dwarfed-and-a-bit. More so as the average dwarf is likely to be shorter than that threshold.

And yes, thanks, I am indeed trying to put off doing some unappealing work.

[DropBear]

"couldn't find the average height of dwarves..."

Ah, but when you say the word "dwarf" do you think of "person not quite tall enough to be called normal" or one those adorable munchkins surrounding Snow White...? Because those are definitely less than half her height...

[Anonymous Coward]

Good luck with that self driving car thing...

Good idea, if the Self Driving Car thing takes off. There's more to it than kit that works. Laws will have to be changed to allow self driving cars on public roads. Don't expect that to be quick or easy to accomplish.

[vincent himpe]

i spent 23 years playing with silicon. and jumped ship just as the big consolidation wave hit in full force...

Here is my take on it: the semiconductor world is struggling. The 'good ole big boys' are having a hard time turning their expensive fabs into cash. Their fab technology has fallen behind what can be done by the fab-only houses like TSMC, Chartered and UMC. So they are stuck in a dilemma :

- do we design cutting edge that can only be run outside forget our own fabs ?

- do we restrict our designs to what can be handled in our own fabs.

We have evolved from a do-all to fabless design houses shopping around for the best fabs. Those guys are highly successful. The old boys that own their own fabs are stuck in a quandary...

When IBM wanted to get rid of their fabs they had to give 1.5 billion dollar in cash for Globalfoundries to TAKE them .... Yes, you read that right . IBM gave 1.5 billion to globalfoundries so they would take the fabs.... http://www.extremetech.com/computing/192430-ibm-dumps-chip-unit-pays-globalfoundries-1-5-billion-to-take-the-business-off-its-hands

That's how bad it is.. and IBM's fabs were cutting edge ...

[Dexter]

Qualcomm is already a fabless semiconductor company. They use TSMC, Samsung and others.

NXP has a few (old) Fabs, but nothing cutting edge.

[Mage]

Automotive

Maybe, but I'd bet it's more about acquiring IP than making chips. Qualcomm are far more interested in royalties, unlike many others, often you have to pay a royalty on sales as well as buy their chips, which are means to an end. They buy smaller outfits and shutter them, just to get Mobile related IP.

There is a lot or receiver IP, NXP used to be Philips.

As others have said, it's more about consolidation.

National Semi -> Texas

Altera -> Intel

Atmel -> Microchip

Linear Tech -> ADI

Others too. I think NXP already ate one of the two Motorola Semi Spin offs.

HP got rid of chips and Instruments and kept the simple but mass market printers and PCs.

[bazza]

Re: Automotive

Qualcomm may want to close down some of NXP's product lines, but they'll struggle to do that completely. NXP have Freescale, and hence chips like the PowerPC 8641D, etc. These have been widely used in many really quite important military systems in the US, and Uncle Sam won't take kindly to the supply being cut off. Some parts have found a use in the F35 program, and there's no way on earth anyone will consider doing a silicon swap anytime soon.

That happened to Apple when they bought PA Semi all those years ago. Apple bought the company for the staff, not the product line, and promptly announced discontinuation. The US gov told Apple that they had to keep PA Semi's PowerPC SOC chip going, because it had already been incorporated into some fairly significant military systems. To cap it off, the staff (there were a lot of ex-DEC silicon engineers involved who had started up PA Semi in the first place) didn't like being Apple drones and quit, setting themselves up as yet another startup called Agnilux which then got bought by Google.

At the time that SOC was unmatched by anything else on the market, and it would have been impossible to migrate the designs on to alternate silicon. By the standards of the day it was phenomenal - dual core, 2 GHz, 64 bit, dual Eth NICs, fast memory interface, well suited to real time applications, lots of GFLOPS, all in less than 13 Watts.

[Anonymous Coward]

NFC? It's the one for me.

So this is the semiconductor company that suddenly pulled out of the NFC chipset market leaving their customers high and dry now buying one of the world's largest NFC chipset companies. Plus cà change...

[Anonymous Coward]

How long before we gain a real ChipZilla

TI bought Luminary Micro. Intel bought Altera. (well, Intel can have 'em, Quartus II is awful!) Microchip bought Atmel (look out AVR, you have been PICed). NXP bought Freescale (RIP Motorola). Avago bought Broadcom. Now Qualcomm has bought NXP. ST have been left alone (as far as I know) thankfully.

It'll only be a matter of time when there's just a duopoly of mediocre companies to buy chips from instead of the competition we have now.

Viva la change!

[short]

Re: How long before we gain a real ChipZilla

The barrier for entry to the microcontroller game seems to be slowly lowering. Design tools and blocks improve, fabs take smaller wafer MOQs, packaging houses likewise. I'm sure that there will be a steady trickle of innovators to burn brightly and get borged. Espressif Systems, Nordic, blah. If there's more need, more will appear. In 20 years, the barriers might be much lower - sketch design, get silicon (in 2 months).

[DropBear]

Re: How long before we gain a real ChipZilla

Omni Consumer Products. Are we there yet...?