Based on an initial scan of the publications below it seems like it would but I haven't had a chance to review the robustness of these publications yet so don't want to officially claim this yet :)
< disclosure, i run the firm that lead this series A >
if you mean integrating into Arduino shields, then YES, absolutely.
the beauty of Spark is that it you can program using Wiring (the same language as arduino), so a lot of the example sketchs and libraries are easily transportable.
So basically, if you love an arduino shield, you can prototype using this (http://www.makershed.com/products/shield-shield)
then you port the library over to spark's cloud, and you have a wifi enabled Arduino Shield!
once your prototype is where you want it, you can roll-up a version of that shield with the spark components built on the board. because the reference design of the spark core is open, its actually quite easy (assuming your shield's schematic is available).
and when you are ready for production, getting the cc3000 and stm32 is as easy as throwing a rock down the street in shenzen.
i could not be more excited about leading the investment in spark....and building products at home on the platform.
They don't make a chip. They use a CC3000 from TI -- which costs about $10 -- and provide a thing wrapper around TI's own provided driver.
You can take the same CC3000 from TI -- around $11 in low volume -- put it on your board and write your software with their driver.
They basically make a good dev board but for anyone wanting to make anything in a volume beyond 1K they'd use the TI module directly. You don't put a dev board in an end product.
I fail to see how their business is going to scale beyond this.
The main product isn't the board -- CC3000+STM32 -- but rather the glorious and inscrutable cloud service that goes with it. The board designs are even being given away for free, it's OSHW.
Their cloud software is also open sourced, and based on IEFT CoAp[1]. I see this being useful for a newbie maker, but for product shipping in any volume you'd integrate the CC3K directly ($11 vs $39) and you'd host the cloud server yourself, or pay them a nominal fee.
I am not seeing a cloud "operating system". Maybe I'm missing something??
But most of the magic is the cloud app that each device maker writes that does the heavy lifting, and that's pure web service; i.e. if it's X o'clock send command to turn on the sprinklers, unless it is raining in the user's zip code.
The OS part is that you're running the Spark firmware to manage comms and security, which talks to the Spark Cloud and makes the rest of it easy. Building the cloud app to be fast, stable, scalable, and secure from scratch is difficult, so this is a SaaS play. Plus, how are you going to manage your 100,000 devices in the field? With Spark you can even do it all with API calls.
They recognize that it's more for prototyping and wouldn't be used at a production level:
"We use industry-standard components — chosen to be powerful, inexpensive, and widely available — that can be designed directly onto your own circuit board, bringing your Bill of Materials down for volume production."
We tried hard to make sure the components on the Spark Core dev kit are really easy to source and buy in low quantities, so you can build the board yourself if you really want. Large companies often already have their own supply chain, so this makes things much easier for them, but you can also buy in bulk and get great discounts if you're a startup or building in lower volumes. :)
Right, I agree. That is what I gathered before but the article and their new splash page makes it seem like they want to supply this to large device makers (i.e. anyone not building things out of their basement) and I don't see how this could work.
They can't make money on the hardware, because TI makes the chip and provides the reference design and drivers for free. Spark boards are all OSHW anyways. You'd spend the extra week and integrate the TI part into your device anyways.
Their cloud stack is open source, and is basically a transport. As the device maker you'd still have to write the cloud app that gathers data from different points and sends commands to the end unit.
Beyond the dev kits I'm having a hard time seeing the revenue stream and/or value proposition.
What am I missing? Maybe guys from Spark want to chime in?
If I use Spark to prototype, is there a product they are offering me that I would need to pay for at scale?
With more than 20,000 units in the field, the Spark community is the largest consumer of the CC3000 module, so you get the benefits of driver testing and patches, and great community support. Large businesses could certainly use the open source modules and run with them, but if you're a large business you generally want to pay for support and licensing. Next for us is building a suite of fleet management tools and controls that larger companies want that also benefit hobbyists.
If you're a business or a maker, using the Spark Cloud saves you from having to worry about building your own protocol, dealing with socket programming, spinning up servers, as well as worrying about encryption on microcontrollers, or wearing a pager. Having the source code available means you can keep your private things private by not going over the internet, as well as building things we haven't thought of yet.
Couple of questions:
- Are you always sticking w/ CC3000 or are there other WiFi modules you intend to support?
My main question, might have gone unnoticed in my long post... Is there a service of yours that I need to pay for to get these features, or do I get the benefits you outlined above by running your firmware and deploying your open source cloud code on my system?
Do you have plans to go closed source in the future?
Zach here, also from Spark. We're definitely working on next-gen hardware now, haven't selected a final chipset yet.
Our hosted cloud is somewhat different from the open source version, and the differences will grow over time. As an open source company, the question always comes to "how do we make money?" Our open source thesis is to give away basic technology and sell advanced technology. Right now the biggest differences between our hosted cloud and the open source version is scalability; our system is designed to scale horizontally whereas the open source one is a single instance (although it could, of course, be modified). We are still deciding which features that are still in development will be open sourced, and which won't. We've learned from other for-profit open source companies (like MongoDB, for instance), that even after being on the market for years they're still trying to figure out which features to give away and which to sell.
Thanks for answering Zach! Your answer certainly helps make it more clear.
To use your feature rich cloud, would one have to use your spark core board or one can roll own CC3000 based board + your open source firmware but pay you and have it talk to your closed cloud for scale? When is your plan to announce pricing for this feature rich cloud?
You've got it- you can build your own board using our reference design and pay us a minor fee to use our lovely cloud. Pricing isn't public yet, but we can discuss it privately if you have a need. Thanks!
Can you explain in more detail how well it works with Arduino? Could I take 3 arduino components and 3 spark components and have them integrate to do something useful?
No. The spark dev kit includes an STM32F microcontroller. You can think of this as kinda like your Arduino. You can write programs for it, but you'd have to use their online IDE I think since Arduino native doesn't support STM32F.
It is basically an Arduino dev kit w/ the WiFi module integrated. If you want to use an existing Arduino shield you'd have to buy their Shield Shield which you can plug your spark core into and then plug in the other Arduino shield.
1. Meat vs Vegan Diet
2. Intermittent Fasting vs No Fasting