Golioth wants to be a cloud designed for hardware nerds

What would a cloud platform designed for hardware engineers look like? That’s the question entrepreneur Jonathan Beri asked himself more than a year ago, after a career building connected devices and struggling to connect them to the cloud. His answer is Golioth, a startup building a cloud for the IoT.

Beri used to design networking protocols at Nest and at Google. He then worked at IoT platform company Particle. And after that, he went to WeWork, to build out a smarter building infrastructure for offices where access control and building management needed to shrink and flex to accommodate a rapidly changing collection of users. Along the way, he realized that when building connected devices, the hardware teams and the cloud teams didn’t communicate very well.

And that lack of communication made it really challenging to build a cloud backend for IoT devices, especially for devices that need to run on really limited physical infrastructure with little memory and computing power. Or those that need to use a specific connectivity protocol that isn’t supported by existing cloud platforms. Beri’s company is starting with a cloud that works for microcontrollers running popular real-time operating systems, is protocol-agnostic, and can adapt to whatever connectivity for which the hardware team has designed.

Beri says that existing cloud providers for the IoT — whether they’re a big company like Amazon or a specialist platform such as Ayla Networks or Tuya — are designed by cloud engineers and can force hardware designers to choose physical devices that compromise a design. For example, many cloud platforms have deals with specific chip vendors that make connecting that particular chip to their cloud infrastructure easier.

But hardware teams might have a different chip in mind. If they want to use a chip without an established connection, they may need to run a software development kit or other software that can suck up valuable memory on it. At that point, they may as well compromise and choose something else. Similar challenges can arise when trying to pick connectivity and network protocols.

This is where Golioth comes in. The company’s platform will support all of the needs of an IoT device, including device messaging, software updates, security, fleet management, computing and storage, networking, manufacturing, and provisioning. Customers can run it on premise or in a cloud provider of their choosing, which will be helpful for those customers who might have a directive to use Microsoft Azure but want to use a board or networking protocol that isn’t supported by Azure.

I initially questioned the need for another cloud platform for the IoT. But after my conversation with Beri, and later conversations with device developers, I came around. Beri’s big challenge will be to add support for all of the connectivity, all of the networking protocols, and all of the other customizations that clients will want. There’s a reason that other clouds have limits associated with them.

However, Beri is starting small by focusing on microcontrollers and real-time operating systems, because that is where the most pain is. The first networking protocol he’ll support is CoAP, followed by MQTT and Lightweight M2M.

Golioth is still in the early stages; Beri expects to close a seed round soon. He’s also testing the software in pilots, but doesn’t yet have a paying customer.

There are dozens of companies trying to make it easier to connect hardware to the cloud, and all have made some inroads with customers. I think the opportunity is massive, fragmented, and confusing, which means there’s room for another startup to give it a whirl and see if what it has works for enough customers to make it a viable business.