I think it may be a stretch to conclude from this that the acquisition of Infineon's wireless chip unit was a bust. Chip companies are always looking to bolster their capabilities when they can. If Strauss is right that this unit's mobile communications transceiver and design team is probably the best standalone RF team in the world, perhaps Intel just wanted to grab the talent.
@mcgrathdylan -- because having the best RF wireless modem in the world isn't going to sell unless as a minimum you've also got a competitive baseband DSP and the software stack to go with it, and preferably also a competitive application processor.
Motorola (and then Freescale) didn't have these, then Fujitsu didn't, now maybe Intel will ;-)
This is the best RF IC on the market for 4G LTE. Intel will next ask Fujitsu who makes the best LTE modem in the world and try to buy them. The advantage of the Fujitsu RF IC is low power. I think part of the purpose is not just to get the best RF transceiver on the market, but to get inside information on everybody else using this part.
When Motorola ran this group they mismanaged product development and had rapid sales decline. I was surprised they had anyone left to do this part. This part was developed with money from that Japanese consortium for 4G development. The consortium was supposed to do an LTE modem but that was going nowhere. Some companies have already spent over a billion dollars and come up empty.
A lot of companies are hyping their LTE products but it is not going well for them.
That would be on the LTE modem part and not on an RF IC. This Fujitsu chip can handle a lot of RF bands but some applications don't need so many bands. There will likely be smaller versions of the chip in the future for simple data applications.
In some of the articles about companies shutting down their R&D efforts on 4G, a lot of bloggers wondered why the companies would do that. I am sure that the efforts were not as far along as advertised and the companies heard that better parts were coming to the market than what they offered. If you are not there after a billion dollars, you will never be there. If you did a decent LTE modem for less than $200 million, then you did well.
Intel had tried WiMax and then trying to enter into mobile market with LTE modem. Yups agree that mobile RF is getting very complicated and success rates are very low. Hope Intel make good progress with this aquisition.
Fujitsu Semiconductor Limited and entire Japan semi industry is downsizing. The LTE transceiver was a valuable asset. Maybe FSL did not have wherewithal to compete with QCOM (high end) and MTEK (low end), and 'decided to better sell the group'. Also rocking FSL is the ongoing (failed?) attempt to merge with Panasonic Semiconductor.
Multi-mode multi-band modems and front end RFICs are the new "platforms" of the industry. Such solutions for smartphones are a Kings game. Only the cash rich need apply. Even for Intel, success does not come easy or quick.
>> If Strauss is right that this unit's mobile communications transceiver and design team is probably the best standalone RF team in the world, perhaps Intel just wanted to grab the talent
I am not sure they are that good. Otherwise, they would have done so well that the company might not have to sell. This is an established firm and not a startup where the argument for acqui-hire could be made.
The key point may be the x86 integration. Most ARM implementations these days integrate DSP coprocessor capability. Intel may be looking for something to tie together with an x86 core to compete more effectively in the mobile space.
Intel had bought Infineon's division called Comneon about three years back. This division was primarily doing software for wireless modems (GSM and UMTS). Intel also bought a company in Egypt, and a company in Germany that was working on LTE. This was together called Intel Mobile Communications. Intel acquired about 200 people from Motorola last year. My guess is, as Fujitsu Wireless lineage goes back to Motorola, someone from the newly acquired people suggested this purchase to speed up the product.
There is a standard called DigRF, which defines the interface between processors and transceivers. Intel could have been using Infineon's transceivers till now (Infineon still has this division with itself). Intel may be wanting to start using its own set of transceivers after this purchase.
A wireless product is made of many things. A good power efficient RISC processor (Which Intel is still developing), set of DSPs, a transceiver, Power Amplifiers, Software. And that is only a wireless modem. A development modem board is a green PCB with lots of ICs and wiring, and an antenna.
A smartphone usually has another set of application processor + the operating system (Android, IOS).
A lot of cost is involved in fabrication of the development boards, software development, test equipment, field test. All this just for the basic modem.
Intel is very very late in this business. It is now trying to put its hands into anything that it can get. Only a few companies have been successfully able to do most of these things: Qualcomm, Nvidia (To a certain extent), ST Ericsson ( now Ericsson).
(Samsung has its complete solution too, just that it is not too evolved. It does, periodically try to push its OS or its processor in some products, in an effort to reduce its dependency on QComm and others).
ARM, intel's competitor only licenses its architecture. Intel is now trying to become a Qualcomm which is higly unlikely. Intel is an amazing company, I want Intel to succeed, because competition is always good. Hope it works out for them.
Intel wants to win in the mobile ecosystem just as they won in the desktop era. Never though Japanese iconic firm will be the pathway to execute that strategy. I think Intel has all the tools it needs to compete in this sector.
A spokesperson from Intel has emailed me to say: "I can confirm for you that Intel acquired certain assets of Fujitsu's U.S. RF group to expand our mobile capabilities. The transaction closed in July."
This business is moving towards a real oligopoly after all the consolidations. It is becoming about scale than anything else. Someone was analysing the new Moto smartphone from Motorola and how the $500m ad buget pales to the estimated $6b that Samsung spends. In other words, in a crowded system, you can have a great technology and still fail. You need that size to spend to win not just in tech but the marketing blitz.
With so many acquisitions, how is Intel going to manage putting all the bits and pieces together? Its not a plug and play world. I am guessing they will now need some more engineers on the baseband side to interface with the new chips from Fujitsu. RF customisations take a hell lot of resources.
It is a tough life for Intel Management. First, they failed to see the growth of smartphones and tablets. Then, they failed to make a decent power efficient processor for the non-desktop products. Intel is in the processor business, they want to sell more processors. Unfortunately, this isn't happening. No company will replace ARM processors with Intel's in their products. So, Intel has to do it on its own. Thats why this shopping spree, it better not be chaotic. Someone at Intel needs to efficiently manage the two verticals now, one is the high end LTE A chipset (Fujitsu + Motorola acquisition), and other one being the low end Comneon + others that work on gsm-umts-lte. There is still no Intel product with LTE in Market (Samsung's Galaxy tab 3 with Intel LTE may just be a prior committment to Intel or just to test waters for a non Qualcomm based modem).
Intel also needs to demonstrate a really good processor for smartphones, without faked benchmarks.
>> With so many acquisitions, how is Intel going to manage putting all the bits and pieces together? Its not a plug and play world.
Acquisition is about both offense and defence. Intel can buy to get the jewel out of the hands of competitors. That is what happens most times. But in this case, Fujutsu has some real IPs that Intel can use to improve it wireless segment business.
Intel said it would not have integrated LTE until some time in 2014, 1+ years after rivals such as Qualcomm.
Apparently the Infineon wireless folks Intel acquired were not on a fast enough path to integrated, multimode chips so Intel had to change horses in mid-stream. Kudos for them having the courage to double down at this late stage in the game, but good luck ctaching up at this point.
It would seem like Intel has to hope for the next big inflection point beyond LTE if it is going to have a shot at a significant market share here.
Yes I was aware I had written that and should have linked to it.
In that context it looks like the Infineon acquisition, which did so well for Infineon, is somehow not delivering for Intel.
Is this a case of Infineon selling at the top of the value proposition because they could see problems ahead or is [yet another] case of big acquisitions being an almost guaranteed way to destroy value because of cultural incompatibilities?
I guess this team will be the 'tock' of Intel's tick-tock model. They have always wanted 2 parallel teams working on different generations of technology. With all the money it takes to make a good chipset, lets see how well they execute it.
I guess the two-phase manufacturing/design cycle could benefit from multiple feed-ins.
It is quite common to have processor IC implementation design teams dwell on a project for two or three years and leapfrog each other as they help the company bring out more frequent product introductions.
I guess the same arguments apply to LTE modem chipsets?
But I didn't know Intel always aspired to that, or couldn't generate it out of their Infineon acquisition.
With the Infineon acquisition, they got just the 'tick'. Pure R&D needs about 500 people which they planned to hire later on for tock. Infineon lagged badly in LTE which is evident in Intel's desperation now.
This is interesting indeed. I wonder how the Intel Corp. plans to expand its mobile capabilities. Yes, I read the information about mobility equipment such as Smartphones and tablet computers, but many people would like to know more about Intel Corp. business plan and direction for this project. Peter Clarke, we will be looking for a follow-up article to this one in the future.
Interesting acquisition indeed. Low power is the name of the game and while Intel processors are not the most energy efficient by any means, the energy consumption in mobile platforms is not dictated just by processors' efficiency. RFIC and Modems consume a big chunk of the overal power budget. The key is to manage the overall power budget efficiently for various use cases. That is where vertical integration can come in handy!
Even after Intel's Infineon wireless acquistion, we still dont have a baseband option in most ultrabooks . I am waiting for my macbook air to come with inbuilt 4G option. Hope they speed things up a bit with this acquisition.
Blog Doing Math in FPGAs Tom Burke 21 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...