You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
(from Reuters) U.S. chipmaker Intel (INTC.O) agreed to buy driverless car-technology firm Mobileye (MBLY.N) for $15.3 billion on Monday, positioning itself for a dominant role in the autonomous-driving sector after missing the market for mobile phones.
I certainly see why Intel did this but I think they paid WAY too much and will end up with another Altera (underperforming acquisition). I also believe that Mobileye is a stop gap for fully autonomous vehicles so the future is not so bright.
I would liken this to Apple building their own SoCs and operating system to control the customer experience. In my opinion car companies will also build their own systems to control the driving experience to better differentiate their product. Tesla is leading this effort of course and the other car companies will have no choice but to follow. Thus far here is the autonomous partnering landscape:
BMW/INTEL/MOBILEYE UBER/VOLVO/TOYOTA
LYFT/GENERAL MOTORS/CRUISE AUTOMATION
LYFT/DIDI CHUXING/UBER/APPLE
GOOGLE/FIAT CHRYSLER
VOLKSWAGEN/GETT
FORD/VELODYNE/BAIDU
AUDI/MERCEDES/BMW/HERE
NISSAN/NASA
So you have to ask yourself: Self, just who is Intel going to sell this product to other than BMW?
On the plus side Intel letting Mobileye take over their current autonomous group is a net positive because Intel execs are not even close to being qualified in this highly specialized market.
Intel again is skating to where the puck is and not to where it is going. Just my opinion of course......
Agree with Nenni, Intel paid 75X the revenue to acquire Mobileye, so within 3 years we will likely see Mobileye spin out of Intel again.
Smaller companies like Mobileye are used to making rapid decisions in order to predict the market direction, while Intel has a long and ponderous decision-making cycle, so the cultures are not compatible and will force the Israeli employees to wish for the good old days when they were in control of their own destiny.
Intel missed the last big thing (mobile), and now seems to be throwing globs of money at anything that could possibly be the next big thing, wearables, drones, autonomous driving, AI. Being a second tier player in a bunch of different speculative markets seem like a great strategy to me, it appears to demonstrate a lack of vision on Intel's part.
I wonder if mobileye was freelanced in the production of 10nm and Intel decided to snatch them. I could see their system-on-a-chip for driving could be modified for more precise manufacturing. That's my fictional story.
Agree with Nenni, .................... so the cultures are not compatible and will force the Israeli employees to wish for the good old days when they were in control of their own destiny.
Some people at INTEL and Mobileye seem to see this differently:
In a letter to its employees, cosigned by Ziv Avriam, Mobileye’s co-founder, president and CEO, and Amnon Shashua, co-founder, CTO and chairman, the two executives stressed, “The transaction is unique in the sense that instead of Mobileye being integrated into Intel, Intel’s Automated Driving Group (ADG) will be integrated into Mobileye.” Under the agreement, two Intel executives currently overseeing ADG – Doug Davis, Intel senior vice president, and Kathy Winter, vice president – will report to Shashua after the transaction’s closing.
We probably don't know the whole story. Maybe there are some patents that could be important in self-driven cars ? Or maybe some good customer(car companies) trust ? Or maybe the ADAS experience has usefulness verification and regulation of SDC's , which are very tough subjects ? or maybe some bunch of data ? Also those guys at mobile-eye are quite talented.
And it could be that Google/GM wouldn't really care much about what processor site in the car - as long as it got the flops ? because what's am extra 100W for a car ? and what's some extra silicon margin, for a car that would be quite expensive and probably be offered as a service ?
Could it be that in that situation the only thing that matters is who gets there first, and all the factors i talked of before will be what matters, and what allows Intel to become the silicon supplier ?
Or let's ask it in another way: considering all the money Intel has, how could it find a better use for that $15B ?
Intel could try and buy the #1 foundry in the world TSMC, but it has a market value of $156B. At least Intel understands what a pure-play foundry like TSMC produces because of similarities with the IDM business model.
In the #2 pure-play foundry ranking we have UMC with a market value of $4.98B, so it is more affordable than TSMC.
We probably don't know the whole story. Maybe there are some patents that could be important in self-driven cars ? Or maybe some good customer(car companies) trust ? Or maybe the ADAS experience has usefulness verification and regulation of SDC's , which are very tough subjects ? or maybe some bunch of data ? Also those guys at mobile-eye are quite talented.
And it could be that Google/GM wouldn't really care much about what processor site in the car - as long as it got the flops ? because what's am extra 100W for a car ? and what's some extra silicon margin, for a car that would be quite expensive and probably be offered as a service ?
Could it be that in that situation the only thing that matters is who gets there first, and all the factors i talked of before will be what matters, and what allows Intel to become the silicon supplier ?
Or let's ask it in another way: considering all the money Intel has, how could it find a better use for that $15B ?
My suggestion to Intel was to buy eSilicon and become the #1 ASIC provider. TSMC has GUC, GlobalFoundries has Invecas, UMC has Faraday, etc... And eSilicon is doing some very big chips for some very big companies, absolutely. Same advice for Samsung Foundry.
Daniel, i like the eSilicon idea, but i would add that on top they should attack the high NRE problem.
For example, buying eAsic/Baysand(structured mask programmable asic), Zeno semi(2T/1T sram), Maybe someone with multi-beam e-beam litho, and work(i.e. ackuire talent) on the next generation of design tools , something above HDL's. Add the EMIB bridge , some prefabbed dies. And don't say no to making ARM processors. Combined with eSilicon , They may be able to weave a platform that pushes 28nm chips to 14/10nm Intel, or even some new node guys from TSMC etc.
But (a)does that make sense? (b)does that make sense to the stock market ?
What about NVDA? They claim cars area a software problem while INTC and MBLY are merging for "lower cost and faster integration". Are they doing this because they are behind? Is that what the TSLA move to drop MBLY and adopt NVDA is telling us? Meanwhile INTC needs to keep those fabs full at all costs in order not to fall behind TSM as they seem equal in process to me.
Dan, when I did some research on Mobileye a few months back they seemed to be working with many more customers than just BMW.
Per the article I wrote, "Mobileye’s use of IP from Imagination and NetSpeed IP is a valuable feather in the caps of both these IP providers as Mobileye is known to be a pioneer of heterogeneous SoC designs and they know how hard it is to get it right, especially in a coherent environment. Their SoCs are used by a majority of the world’s automakers including Audi, BMW, Fiat, Ford, General Motors, Honda, Nissan, Peugeot, Citroen, Renault, Volvo and Tesla. Mobileye’s use of these IPs is a testament to the strength of the offerings from Imagination and NetSpeed.".
[FONT="]The lack of GPGPU IP is another weakness in the Mobileye deal that could potentially hurt Intel chances of fully utilizing the acquisition[/FONT]
I further reviewed company comments on the matter and NVDA management claims to have with Xavier a 1-chip L4 and several chip (2 maybe) L-5 solution with a ton of working software on top including AI features at every level (perception, decisions, mapping, blah blah blah). Apparently car vendors just call a few APIs and they are done. Meanwhile INTC management shows up with a bag of chips, even with MBLY, with the first integration "years" away. They seem positioned at the very low end of the market, dare I say L2 or L2.
