Microchip’s Proposal To Acquire Atmel

A proposal from Microchip to acquire Atmel has been deemed a ‘superior proposal’ by Atmel’s board of directors (PDF). This is the first step in the acquisition of a merger between Microchip and Atmel, both leading semiconductor companies that have had a tremendous impact in the electronics industry.

Microchip is a leading manufacturer of microcontrollers, most famously the PIC series of micros that can be found in any and every type of electronic device. Atmel, likewise, also has a large portfolio of microcontrollers and memory devices that are found in every type of electronic device. Engineers, hackers, and electronic hobbyists are frequently sided with Microchip’s PIC line or Atmel’s AVR line of microcontrollers. It’s the closest thing we have to a holy war in electronics.

Last September, Dialog acquired announced plans to acquire Atmel for $4.6 Billion. Today’s news of a possible acquisition of Atmel by Microchip follows even larger mergers such as NXP and Freescale, Intel and Altera, Avago and Broadcom, On Semiconductor and Fairchild, and TI and Maxim. The semiconductor industry has cash on hand and costs to cut, these mergers and acquisitions are the natural order of things.

While the deal is not done, the money is on the table, and Atmel’s board is apparently interested.

130 thoughts on “Microchip’s Proposal To Acquire Atmel

    1. Thanks for the clarification! I was preparing to ask here how Microchip could acquire Atmel if Dialog had acquired Atmel. I didn’t think it made sense for Atmel to be bought repeatedly by different companies. :)

    1. Yes, that’s where we seem to be heading. And seeing how well corporations have got politicians into their pockets, I doubt that there will be legislation to prevent that.

      Btw, is there any other 8bit RISC MCU architecture beside PIC and AVR? Or will Micromel be the only supplier then?

      1. 8051 is still going strong 35+ years later. Atmel, Silicon Labs, Cypress, NXP, Nuvoton (formerly WinBond), Intel and others still sell tons of them. I would bet it is close to AVR8 in cores shipped annually / collectively. HC05, HC11, and 8031/32s are still around in various incarnations and well supported.

      1. Looking at some market share figures (although dated a few years), Atmel seems to sell more parts overall. With the majority sold to businesses, not hobbyists. Axing existing Atmel parts wouldn’t make sense.

        1. Don’t you have it reversed? Microchip is about 2-3x larger than Atmel. Moreover, they don’t make only microcontrollers – Atmel makes only microcontrollers & memories.

          Here are some older market data – in MCUs Atmel is taking in more revenue (but their chips are a lot more expensive) than Microchip, on the other hand, Microchip isn’t selling only MCUs, so that isn’t covered in that survey:

          1. By “parts” I really meant MCUs, as that’s what folks posting here are going to be most concerned about losing. The Atmel MCUs are more expensive, though I wouldn’t say they’re “a lot more expensive”. Maybe Microchip actually does sell more MCUs by quantity, rather than by market share; but market share is more important, and it still wouldn’t make sense for them to start axing MCUs. Especially if they’re in a position to manufacture those MCUs more inexpensively than Atmel!

            Also, go to atmel.com, and click Products. They sell a lot more than just MCUs/memories. I recall hearing that their touch product line is a good seller.

    1. I see potential for a positive outcome, from a hobbyist point of view.

      1) Microchip has a strong policy of not obsoleting chips, so I don’t expect they’d discontinue any existing Atmel product.
      2) I get the impression Atmel is short on fabrication capacity. Seen many complaints of parts shortages. Merger with Microchip could help.
      3) And maybe, supporting the existing Atmel customer base would help push Microchip to be more hobbyist-friendly. They’ve improved on that, but rather slowly, and not as much as some would like.

      1. BS on 1. Microchip is obsoleting chips every few years. I know this from personal experience. I used to be a Microchip customer, but with their myriad of incompatible chips (even among versions of the same micro (try 16F874 vs 16F874A)), i switched to Atmel few years ago. If someone has to disappear, I hope it is Microchip, not Atmel.

        1. I looked at 16F874 vs 16F874A. The incompatibility between the two is:

          “Same PIC micro, but different versions (16F874 vs 16F874A) use different methods to burn internal flash.”

          With the quote being from *you*, on another Hackaday thread from more than three years back. That change of method is to allow it to be more easily supported by newer ICSP programmers, supporting a wider variety of MCUs, including more modern ones. Which is generally a good thing.

          But if that’s difference isn’t acceptable for your application, then I’ve verified the original 16F874 is still produced, in stock, and easily ordered from Microchip. Despite it being a chip from 2001.

          So your example is false, as nothing has been obsoleted. Sounds more like many years ago, you failed to read and understand the documentation, bought a different chip under the assumption it was identical, and are still holding Microchip accountable for your mistake.

          1. What exactly is it you expect, for every new MCU to work identically to one from 2001, till the end of time?

            The original 16F874 is still available, and will be for years to come. If that’s what you wanted, you should have used it. See that “A” on the end of 16F874A? That means it’s a different part, so of course it works differently, and it’s *your* responsibility to determine if those differences will affect you.

            If you can’t understand that, or can’t accept responsibility for your own negligence, then please remain in ignorance silently and without spreading lies about Microchip obsoleting parts.

    1. Dialogs $4.6B offer was tied to their stock price, which has dropped ~42% since September. So their offer dropped to <$3B since then. They need to offer cash if they want to counter.

    1. That was news of a week ago and points out that the unsolicited bid by Microchip was made even further in the past, at some point that I can’t even find mention of. The real news of the day is that Atmel said, publicly, that they like the Microchip bid (which now may not even be a bid by Feb 2).

      Sometimes you can follow trends by following hobby sites; but getting the financial news up to the minute from hobby sites doesn’t work so well, regardless of which hobby it is.

  1. Microchip seems to be on a roll.. they bought Micrel last year, along with smaller companies ISSC, Supertex, and EqcoLogic. I watched the Micrel one closely — I’ve had a really good relationship with Micrel, and I use some of their power electronics and Ethernet switches in my designs. Never had much use for Microchip microcontrollers, but I suppose their serial EEPROMs are as good as anyone else’s :-) But the PIC did seem to be the go-to micro for hobby projects over the last 10 years or so, at least in the days before $10 ARM boards.

  2. This is a very bad news, microchip is the leader of “errata sheet” and false advertising datasheet with shitty development tools for devices that (almost) never involved. Of course it’s nice to see that 20 years later parts are still available, they made a lot of money on the back-in-time-already-obsolete PIC architecture but only because they massively sponsored schools and students, most of them don’t even try by themselves any alternative architecture. When I see a supposed experimented engineer using a microchip I immediately assume he is a fucktard losing all credibility.

    1. Agree. I always found the Atmel datasheets within the top best practice. I especially like the fact that all the informations, including all chips version errata and document history, are always included. This have saved me a lot of time in many situations. Freescale datasheet are probably even worst than Microchip.

      1. Exactly my experience too. Atmel datasheets are really well produced and thorough. Microchip ones are not as good by far.

        Frescale ones tend to be ‘here’s a bunch of registers and bits and a bit of waffle about how it might work’. I actually think that Freescale’s problem is that the hardware blocks have been upchucked through various designs and grown bits and modes making then ugly and buggy.

        By contrast I always find the AVR hardware blocks simple by design and nice to use.

  3. Here’s finally a reason to try out the STM8 line microcontrollers! They seem to be more cost efficient in volume production anyway, enough so that I’ve seen them in several Chinese devices.

    1. Try Stm8 with cosmic compiler and stvd IDE, nice, powerful and cheap. Stvd is light but fast and without fancy GUI.
      Compiler is size limited but free and the code size limitation is enough even for complex project.
      DIE are mature and well documented with a growing community, avoid stm32 because not mature and many DIE revision in the wild.

  4. AVR vs PIC IS the holy war of electronics! As everyone sane person agrees, Microchip is the evil empire with mplab as their means of torture. We are all doomed if they acquire Atmel. Doooomed I’ll tell ya!

  5. This is bad news, not just for the lack of competition. As techies we could dream of a merger of the technologies, but in the end I fear they will just push whatever brings the more money and kill the rest while fighting for intellectual property rights to ensure nobody else in the western world will produce anything similar. – Chinese clone makers, are you listening? –

    1. PICs only run at 80Mhz or so because they use 4 cycle instructions. Doing a quarter of the work per clock as an AVR.

      20Mhz AVR and an 80Mhz PIC are very roughly equivalent… and acutally AVRs can run at 40Mhz in many cases with enough voltage it just isn’t guaranteed over the temperature range.

  6. Ive used PICs for over 10 years and will most likely never change to a diferent chip/architecture.

    I PAY for my compiler and the tools serious PIC users use are FAR superior than AVRs free stuff.
    People dont go with PIC cause it involves fronting money…200 bucks for a compiler… peanuts when you think about how much money people pay for those shields.

    In the early Arduino days, i decided to try it. It was an intereating offer….terrible datasheets, and code that hides everthing pushed me away. There is nothing easier than a PIC datasheet.

    There are no libraries for PIC’S yet i wonder how good and stable those libraries really are. Most people using Arduino ( the only reason AVR has any acceptance) are not programers, yet they blindly use and spread libraries. BLINDLY.

    This has gotten so far out of hand that the new app BLYNK (which looks awesome) says basically theve never heard of PICs… google “BLYNK PIC”…..their official reply is absurd.

    I dont deny AVRs are good, but peoples Ignorance of PICs is not based on technical comparisons but on pure media coverage bias.
    How many PIC articles have you seen in the last 5 years in HAD?

    1. Could not agree more. I’ve used PICs for a long time now and haven’t found myself in need of a PIC that couldn’t do whatever I need it to do. It’s way easier starting off with a blank page and writing my own software than trying to fuck with poorly written libraries and having an IDE that wants to hold me back to spite myself. I too pay for my compiler/IDE and IT ACTUALLY WORKS LIKE A REAL IDE!!! Meaning it can do everything the free POS arduino compiler cannot do without having to waste my time getting some POS third party solution to work. AND the development suite I use has already paid for itself hundreds of times over. I can buy raw PIC for a buck and blow an arduino out of the water in every respect. I only pay for what I need and can ditch that silly ass little board and it’s footprint AND its lame ass expansion header layout, the bloat loader and the crappy non-IDE AND still program in a high-level language OR switch to assembly in the same program if I need something tighter for a certain routine. And knowing how to handle things from this level makes learning other architectures soooo much easier.

      While you won’t find as much pre-written code for PIC’s out there, in other words, libraries, it is actually better for you, seeing how you will actually know how your code works, and over time you will learn how to make better code. I don’t see the problem with that.

      If I was a newb and needed my hand held and didn’t want to actually learn how to program, I’d suffer the arduino lifestyle. Or if I just wanted to rapid prototype and wasn’t really that interested in quality or performance or bothering to actually learn how to code, but typically the projects you see aren’t that innovative that you couldn’t just go buy a pre-made solution anyway. I find that a bit like going out and buying a paint by numbers canvas and then claiming to be Rembrandt.

      1. uhhh, the Arduino “compiler” is just a platform to simplify programming for beginners based on the Wiring platform (with GCC under the hood)- use the full GCC tools if you want a professional tool chain. You obviously do not understand the difference between the microcontroller itself, and a third party development platform designed for people who are not professional embedded developers. There are tools like pic basic for microchip products as well.

      2. You seem to be conflating AVR with Arduino and then comparing that to PIC (originally the Peripheral Interface Controller, not designed from the ground up as a general purpose microcontroller).
        You can do all of what you said on an AVR target device as well. When I first heard that PIC compilers are generally paid for and locked to Windows, I laughed, and laughed, and laughed. The alternatives are best described as “boutique”, where standards compliance and interoperability take a back seat and everyone does something a little differently.The entire Atmel microcontroller lineup is fully supported by GCC and the GNU binutils, a standards-compliant, extremely widely used, high-performance and free as in freedom toolchain.

        Why in 2016 I still see people opting for any of the old PIC range for anything but pure cost reasons is beyond me. Likely the same group that are holding on to their Windows XP installs.

    2. Wow, you pay for a compiler that can’t even do C99 properly in the year 2015 and has erratas on math operations every month. I hope you feel proud on using a piece of shit they are forced to write because nobody at GCC or Clang is insane enough to support the non-linear piece of shit memory architecture of a PIC.

      I literally have reported more XC8 bugs than i care to fucking deal with anymore which is why I ditched PICs altogether a year ago. I don’t know Indian based tech support telling me wrong and then wasting a week to make them realize their compiler is a piece of shit that doesn’t even know the correct endian of the micro.

    3. Arduino is an IDE for people who don’t want to deal with registers and all other internal wild things of the core, don’t confuse with AVR.

      You can work without library with AVRStudio and with a free GCC compiler, recently a colleague switched to Avr after I recommended it, I was so pleased with that he abandoned pic for all future design, there is a big big big step between them, pic is only popular because of the marketing and lobby by massively sponsoring schools and students. But I’m not an Avr Jéhovah witness I’m working with Stm8 and freescale, I’m also considering nxp arm mcu.
      Explore and try a maximum of architecture, never focus on only one architecture, open your mind !

    4. Why are you comparing a pricey PIC dev system to an Arduino board with its free stuff? The first is a professional tool, the second basically an evolved toy. There are expensive tools for both platforms, a true comparison should involve stuff aimed at the same market.

      – People dont go with PIC cause it involves fronting money…200 bucks for a compiler…

      Many moons ago when I bought my MikroC compiler and dev board from mikroe it wasn’t that expensive. IIRC, for like $250 I got the compiler AND the development board with periperhals, displays etc. Now I see their new compilers prices skyrocketed.

      To me that’s one more reason to support what works with free compilers.
      Of course unless one is writing firmware for medical devices, airplanes, missiles, ABS controllers etc, which is not my case.

      1. I develop using MikroE boards since it gives me the flexibility of a “Shield”, and Ive found them to be of great quality and performance. Their compilers on the other hand i have not tested, but looking at their example codes I dont think their example man is very good and i have no need for a different compiler.

        Since i discovered CCS Compiler in 2001 ive never looked back. I just purchased the compilers for the cores i use and its awesome.

        For others above, everybody knows AVR!=Arduino the reason why its commonly tied together is because if it wasent for Arduino, alot of people would have never heard of AVR… which is probably half the people on this flame war.

        As a PIC user, I agree with every body saying the Microchip compilers are awful.. i tried, never again.
        CCS Compiler is where the magic is at.

      2. Forgot to make my other point: Why is paying for hardware OK but not for a decent compiler?

        Paid hardware: “OMG Sparkfun new product post friday arduino extravaganza! YESS! GIVE ME ALL”

        Paid Compiler: “PIC Sucks, Arduino is Free and awesome”

        People have money for expensive Dev boards and shields but Theyd rather die before purchasing a decent compiler.

        1. 1 acronym & 2 words. LLVM Byte-Code. MoFo binary Rosetta stone.

          I deplore apple and the fanboi cult they cater to. Only good thing that came from them was internal subversives and the Clang project. Granted Woz is the man. But to have that vision corrupted by the “fruits and nuts eater” of course published standards for IBM XT advanced us.

          But “decent” compilers? Not on this planet, a certain layer of cruft, laziness, corp culture, operating “standards” and the death knell of the organization is “A one off quick fix becomes the convention then the standard” lives inside all purchased compilers.

  7. I switched camps from Microchip to Atmel few years ago, because:
    1. You get much more bang for the buck on Atmel chip (more RAM without fckng bank switching, etc).
    2. Free and open source tools that work well on Linux (fck u MPLAB and Windoze).
    3. Atmel does not produce myriad of incompatible micros, obsoleted every few years.
    4. Microchip has even minor version numbers that are incompatible (16F874 vs 16F874, different Flash write sequence, wtf?).

    I could go on, but most people have seen the light way before me, and developed Arduino using a sane chip.

    I pray that this “deal” does not go through.

      1. Another set of corrections:
        1, RAM bank switching is domain of old 8-bitters. Are you aware of their newer offerings?
        2, I use FOSS tools for Microchip MCUs daily, under Linux. GCC for PIC32, SDCC for 8-bitters.
        3, The old chips are marked as obsolete, but still available. You should do new designs with new chips, but you are perfectly OK with getting the old ones. The 16F874 you mentioned is normally available.
        4, Yep, you have to read datasheets and migration documents.

        1. Try locating an 8-bit PIC with more than several KB of ram, without bank switching. Atmegas are just beautiful designs compared to PIC wtfckery with banks and registers constantly changing names BTW models. You try to adapt and select one model, but wouldn’t you know, very soon a new subversion is out and your code works no more … Why this constant rush into deliberate obsolescence I really don’t know.

          1. So you are telling me that you have code for device X, they release device Y and your code running on device X magically stops working? What kind of sorcery is this?
            No, you don’t have to use new device, if your programming skills are that poor you can’t simply adapt no new ones – which is usually trivial procedure. You can freely use the old ones, they are normally available and nobody forces you to change anything.

            I reviewed a few assembly projects from the past and in one of them I used PIC18F26K22. It is 8-bit device, with 4kB of RAM and there is not a single bank select instruction in the whole code (~ 8 thousands lines of code). A made a few assembly projects with PIC24FJ64GA002 too – I must tell you, beautiful architecture compared to AVR.

            I wonder why so many negative comments regarding Microchip devices are coming from people who don’t know a thing about it.

        2. Last I heard, the license of Microchip’s libraries for PIC32 forbade you from using them with GCC. Why? They want you to use their rebranded version of GCC, which is almost exactly like the upstream one except with a license manager added that disables higher optimisation levels unless you pay out $$$ to Microchip. I am not joking.

    1. You as a true AVR Programer can then appreciate the jewels found here:

      This is the crowd my comments are directed to…
      Those who view PIC as a Rare thing.

      Im sure AVR has some great things, Ive never denied it… i even say it on my first post.
      I will never enjoy AVR greatness because i simply haven’t met a challenge that requires me to go away from PIC.

        1. You’re kidding me right? I ridicule a turd and you recommend me an obsolete turd? Let’s try again:
          “I’ll try to email you but I installed MPLAB V8.92 and the debugger uses a molested HID-driver so occasionally my keyboard stops wo o89fiuh34 ndeiu3 icwiyeg ing”

  8. Well, I have been a pic user since the late 80’s / early 90’s. We used to use National COPS, Motorola 68HC’s, and believe it or not…z80’s. I have tried to jump off the microchip bus many times, among them Cypress PSOC’s. I work in an industry where it is all about taking that 0.01 dollars off the cost of a product in volume.

    If in the real world, your company is big enough, you don’t pay for dev tools, they are given to you. You get good tech support with engineers that most likely have worked in the industry. They tend to have application engineers that are hired from the industry.

    They were smart enough to branch out into other things, just like Cree isn’t just about LED’s.

  9. German:
    08.43 Uhr: Der Apple-Zulieferer Dialog Semiconductor hat den Kauf seines US-Konkurrenten Atmel abgeblasen. Zuvor hatte Atmel mitgeteilt, dass es das Angebot von Microchip Technology akzeptieren wolle. Wie Dialog am Donnerstagabend per Ad-Hoc-Mitteilung mitteilte, verzichtet das Unternehmen darauf, ein neues Angebot vorzulegen.

    08.43 Uhr: Der Apple-Zulieferer Dialog Semiconductor hat den Kauf seines US-Konkurrenten Atmel abgeblasen. Zuvor hatte Atmel mitgeteilt, dass es das Angebot von Microchip Technology akzeptieren wolle. Wie Dialog am Donnerstagabend per Ad-Hoc-Mitteilung mitteilte, verzichtet das Unternehmen darauf, ein neues Angebot vorzulegen.


    1. arrrgh st***id missing edit button:

      08.43 Clock : The Apple supplier Dialog Semiconductor has blown off buying its US rival Atmel . Previously Atmel had indicated that it would accept the offer by Microchip Technology . As dialogue on Thursday evening announced in an ad hoc announcement , the company decided not to submit a new offer

          1. of course but at least they made nice and competitive product ! I’m also supporting Freescale because the debug protocol (DBM) is open and there is plenty of cheap tools in the wild, ST with STM8 is one of the most mature and powerful MCU currently on market with single wire debug/program, ARM device looks promising but I don’t want to focus on only one architecture, so that’s why I’m also playing with old Mitsubishi MELPS7700, hitchai H8/500, etc…

  10. Everyone’s such on what each other company makes our provides. But the burger questions is Microchip planning on closing Atmel once they take cover. That’s over 1,000 people over of work.

  11. PIC32 is a shit! and MPLABX is another fuckers game! Pay money to microsoft, pay money to intel and pay money to microchip for stupid-copy-pasted-gcc compiler. There are thousands of bugs in compilers, IDE and microcontroller is just a nonsense mistaken copy-paste.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

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