Advertisement

Blog

Mohamed Kassem: Q&A With the Company Founder

I always love it when I am the first to talk about a new company, and this is the first time I have been able to speak about one in the analog design space. A short time ago, I spoke to Mohamed Kassem, the founder of efabless Labs. What makes this company special in many ways is that it is not an EDA company or a design company, although there are elements of both with in it.

As its name implies, efabless Labs is a fabless semiconductor company that has opened its doors to allow other people to use their flows, their relationships with the fabs and their collective knowledge to build the bits and pieces necessary for someone just like you, with an idea, to be able to put a complex analog idea into silicon without breaking the bank. During our talk, Kassem walked me through the founding of the company, his love of engineering, and what the future of the industry holds. Here’s our Q&A conversation:

Mohamed Kassem: I have been in the area of analog and mixed-signal design and have worked in both design automation companies and in semiconductor development, with a focus on mobile and wireless devices. At TI, a lot of innovation was required in order to meet complexity challenges with aggressive time to market pressures.

When I started there we were working on 180 nm modems and by 2009 we were down at 28 nm, a process that was not yet mature. To meet the requirements we had to use innovative processes that contained more automation, new analog architectures, and we had to do it with first pass silicon and only one metal change. While doing this I believed there was even more that could be done and that led me to wanting to start my own company.

BB: The new company is to be launched at the IEEE Custom Integrated Circuit Conference — CICC (September 23-25, San Jose, Calif.). Tell me a little about the company.

MK: Fundamentally we are a semiconductor company that believes in open innovation through community. We build chips collaboratively by allowing someone with an idea who wants to take it down a level or two to actually design and fabricate chips. efabless Labs brings the concept of the garage back to analog designers.

First we allow them to design and verify their designs. Then we have predefined pad frames, test vehicles, and things that make it a complete flow. When the chips come back, we provide them with a small test board that plugs into a computer and they can bring the chip up in a home garage. We make it easy for people to focus on design rather than the logistics that usually surround the creation of a chip. We have our own design teams creating chips, but we also enable other designers to use those same flows.

BB: What tools are you going to provide?

MK: I don't like calling them tools. They are prepackaged solutions. Some are tuned to the design exploration phase, such as schematic entry and simulation tools. In the second phase they use a prequalified pad frame into which one can implement a design. The next level provides access to manufacturing and to the partner foundries we work with. Finally we have a test board that enables you to wake up the chip.

BB: Who is the customer for this?

MK: Anyone who is design passionate and capable. It could be a company wanting to create a prototype or prove a concept, it could be an independent person who has an idea and wants to look at the possibility of using it commercially, or it could be a graduate student who is working on some new ideas or using it for research.

BB: Assume I have a new idea for an op-amp. How long would it take and how much would it cost me?

MK: The cost is going to be dependent on the process technology, but let's assume it can be in 0.35 or 0.18. Most designs are probably going to be done for about $500 plus silicon costs which could be in the hundreds or low thousands of dollars. You will get your chips in about eight to ten weeks.

BB: What can people expect to see at CICC?

MK: We will be with XFab, one of our foundry partners, and people will actually be able to get a feel for the system. We will be demonstrating a Raspberry Pi terminal that costs just $150, and this enables you to access everything online. More information about this can be found here.

For more information about the company in general, take a look at the company's home page.

So, if you like to tinker and have some ideas for analog designs, then maybe efabless Labs will give you the opportunity to turn those ideas into reality and maybe your own new semiconductor company.

Related posts:

22 comments on “Mohamed Kassem: Q&A With the Company Founder

  1. Scott Elder
    September 27, 2013

    Brian, How does the company make money if the business plan is <$1,500 to provide a few custom ICs, evaluation boards, tools, and software--all in 8 weeks or less?  It seems there is something missing here.

  2. eafpres
    September 27, 2013

    Hi Brian–this sounds very interesting.  However, in such a short space there is a lot of detail left out.  It sounds like efabless is providing sort of a project management service where they can shepherd your design from idea to prototypes.  I'm surprised by a $500 price tag, even it that is just one person managing the project.  With your experience in silicon, do you think that perhaps that $500 is kind of a talking figure, but real costs might be higher?  If they are renting out software etc. it seems it would have to go up.  

    Regarding the eval board, I'm guessing that by constraining the design to certain pad layouts and packages, they have (I wonder if it is socketed?) an eval board they have tooled up and probably costs under $50–it is just a board, some pads, a couple connectors, and possibly a socket for the IC.  It may be that for many designs that won't work–you need a bunch of passives and other stuff around a more complex chip even for eval and testing.  So the price might go up there as well.

    In some sense this seems like an extension of the Maker phenomenon, into IC design and fab.  The assumption is there is a market for that.  Since there are many fabless IC companies and process flows available, but perhaps not to “garage” operations, it will be interesting to see if they can survive.

     

  3. BrianBailey
    September 27, 2013

    @Scott – I did not go into details with Mohamed about their business plan, but I think it is based on 1) they design and build their own chips and they are renting out their flow to other 2) if other people are using their flow and the foundry space, others will be making it cheaper for them because of volume 3) they hope to start collaborative designs that may net them some money 4) they hope that Universities and tinkerers will flock to something like this.

  4. BrianBailey
    September 27, 2013

    @eafpres – yes I find it an interesting concept that I am not sure has been tried before. I think you are right about the board. Perhaps it also includes some room for adding passives. Not sure. I also suspect that it will cost more if you need lots of hours on the tools to get the design right, so this is a starting cost. When I first asked Mohamed if this would enable a sub $10,000 design he said – I can't conceive of many chips even being close to that, although he was also reluctant to tell me about the lowest costs as well.

  5. Scott Elder
    September 27, 2013

    I suppose this could be a modern day Heathkit.  I once thought about providing a service where only one die was manufactured and packaged and all of the participants got a few pads/package pins and some space on one die.  Think of this like another layer on the MPW flow…an MPD (multi-project die) on top of an MPW.

    This could probably be done for $1500, but I don't think there is anything left for profit.  And all of the users would be able to see each others work because a copy of the one die went out to each participant (read: patent nightmare).  I guess we'll have to wait and see how this actually works out.  Sounds more like a service a non-profit university would provide.

     

  6. Davidled
    September 27, 2013

    It looks like company website does not provide their all process to make chip in detail. If they are running the business based on customer projects, I am concern a bit for quality of chip. I expect that Website should provide detail information of chip process at least to let customer understand before contacting them.

     

  7. jkvasan
    September 28, 2013

    It seems the discussion is centred around the business model of the company rather than the platform they 'may' be able to provide for millions of chip-design-aspirants. With this kind of a platform, I think, the textile industry is one of the major domains to get benefitted. Textile machinery use a lot of ASICs to minimise the size and probably to maintain IP.

    If Mohamed Kassem sincerely follows what he means in the interview, I guess, this is going to be a game changer in an industry which scares away common designers and patronises only big brothers.

  8. Netcrawl
    September 28, 2013

    @daej I agree with a company should post at leat some detailed information about their chip, its the only way to know exactly what's their doing, customers want to know more about their chip. Its avery important in today's business, information like this need to be shared public in the company website.  

  9. BrianBailey
    September 28, 2013

    @Jayaraman – I think you are right – he wants to create a game changer and to provide something that would enable aspiring analog designers to actually try their hand at it without emptying the bank.

  10. samicksha
    September 28, 2013

    @Jayram: This is very true that designing Analog/Mixed Signal IC & Systems is not very common expertise infact if you are good at this you can run your own business but at the same you need to keep eye on market that whats missing where you can intervene and earn and i guess MK is doing same thing here he knows that designing is not easy job but if you design, the deal is all yours moreover they claim high-yield Mixed-Signal IP/Chip Design (65, 45 & 28nm CMOS)…

  11. jkvasan
    September 28, 2013

    Brian,

    I may be terribly wrong but I would like to confess that I hate this volume game as it scares away aspiring technoprenuers. The very idea of inventing a great technology stops at that; never comes to the market only due to non-availability of what Mohamed is trying to offer today.  

  12. jkvasan
    September 28, 2013

    Samicksha,

    If you take analog components, just combining two operations can make things more convenient. For example, I needed a few opamps and analog switches for an application, a year back. My best volume would be 1000 pcs per year. If I wanted an ASIC to be made for such an application, I would be spending a huge fortune in the present scenario. Mohamed's solution would have helped me, if was available at that point of time.

  13. David Maciel Silva
    September 28, 2013

    The process of manufacturing an integrated circuit, is not so simple to be done, I believe we will be spending many, many dollars ..

    But as I understand the idea and viable products / projects and to build the CI that has opening / space market and investments are paid.

    Something very close to the model Chinese Manufacturing: participatory cooperation, a kind of consortium of production.

  14. Netcrawl
    September 29, 2013

    China is getting too much attention these days, its getting bigger in the IC market because companies are moving their production plants in China for economic reasons. But there's one major problem here: the lack orf Intellectual Property protection has become a major deterrent for foreign companies seeking to build or establish a state of the art IC facilities in China, the lack of IP protection is also a reason why many tech companies have not brought their industry-leading IC designs to China.    

  15. Netcrawl
    September 29, 2013

    I like the Chinese model they're good in low cost production but  when it comes to Intellectual Property protection China is not an ideal place for this, companies are not bringing their high end design in China.   

  16. Scott Elder
    September 30, 2013

    @JKV  <>

    Why wouldn't something like a pSoC be acceptable for these types of problems?

    The path to a custom device that actually shows up in a product is not about what it costs to see a few parts.  That is very misleading.  These sub-$3,000 prototyping costs ignore the cost of test engineering, quality assurance, and hardware (i.e. burn in boards, etc.).  

    There is an old saying that if you can't afford to buy a Rolls Royce new, you shouldn't buy one used.  That mentaliity applies here as well.  That's not a statement of arrogance.  It is simply an evaluation of the practicality.

    Do it for the experience or the fun, not because it is an economical way to go at 1,000 pieces.  It certainly isn't a convenient development path.

     

  17. SunitaT
    September 30, 2013

    companies are not bringing their high end design in China.  

    @Netcrawl, I agree with you that IP is not safe in China. When you say companies are not bringing their high end design in chian do you mean companies are not manufacturing in China ? 

  18. SunitaT
    September 30, 2013

    Something very close to the model Chinese Manufacturing: participatory cooperation, a kind of consortium of production.

    @Maciel, thanks for sharing this info. I never knew such systems are already implemented in China. Can you please share more details about chinese model of manufacturing.

  19. SunitaT
    September 30, 2013

    If Mohamed Kassem sincerely follows what he means in the interview, I guess, this is going to be a game changer in an industry which scares away common designers and patronises only big brothers.

    @Jayaraman, I totally agree with you opinion. I think it would difficult for the company to make money, but if they can generate revenue from this model then this would definitely change the way designs are implemented. 

  20. Netcrawl
    October 1, 2013

    SunitaT What I mean that these companies would not bring their high end design, the industry-leading technology to China because of great fear that the chinese might steal their technology. These companies are building and moving plants in China for economic reasons- low cost manufacturing overhead and low cost labor. 

    But in term of technology transfer, from foreign companies to China-based companies I think there's still some big challenges and some fear.   

  21. jkvasan
    October 1, 2013

    @Scott,

    Point well taken. However, there was no pSoc knowledge and the learning curve (however short it may be) would have delayed the deliveries.

    It makes sense to go for pSoc based technology.

     

  22. Netcrawl
    October 1, 2013

    They fear of Chinese developing its own product or chips, and might one day compete with them. This is exactly the main reason why these companies won't bring their high end designs in China.

    Few in the electronic industry though that was remotely possible ten years ago but since then the chinese have harnessed some of the nation's best engineers and the government's deep pockets to make something big and huge progress. China graduates 500,000 good engineers a year compared to America's 250,000, that a wide gap. The China's single-mindedness is one big reason why these companies doesn't dare to build fab there or bring their high end designs. A good example is Intel, the company doesn't want to bring their high end chipmaking equipment, why? chipmaking plants are where the bulk of Intel's intellectual property lies. 

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.