Salad introduced a container engine platform hosted on crowdsourced edge infrastructure that allows the vendor to target high-performance computing (HPC) workloads.

CEO Bob Miles told SDxCentral Salad's unique infrastructure targets two workload niches: web-scraping and GPU computing. But "more exciting for the team and I are the workloads that require high-performance computing," he said. Salad's network of "powerful gaming PCs with high-end hardware and GPUs" is ideally suited to run engineering simulations, rendering, and artificial intelligence (AI) jobs. "We’ve seen a handful of customers from the synthetic data generation space using our infrastructure and combining those last two industries," he added.

On the supply side, Salad's network of PC gamers download and install its open source software and configure which resources Salad can use and when – like bandwidth, storage, RAM, and CPU/GPU. Salad's software detects the host hardware and pulls "the most profitable workload (container) to run," Miles explained. Salad then uses Windows Subsystem for Linux to create a virtual machine (VM) on that host hardware and runs its own build of Linux as the environment. That build contains libraries that track resource consumption, fraudulent behavior, and node health. Depending on resources consumed, the PC gamers that make up the network accrue a balance they can use to redeem rewards like games, gift cards, and Netflix subscriptions.

And on the demand side, Miles said the platform is "really no different to a managed container service from the hyperscalers." Developers can create and configure organizations and container groups, choose which resources are needed and the number of replicas, and see pricing estimates. Salad can also pull container images from public repositories like DockerHub or from private registries in hyperscalers like Google Cloud, Microsoft Azure, or Amazon Web Services (AWS).

"Over time, we plan to service any workload from any industry just as the other cloud providers do. Such is the nature of a managed container service, it is workload agnostic," Miles said. The platform is currently in its beta release, and Miles noted "plenty more functionality" is on the way.

Salad competes in the cloud market with two main advantages: the decentralized nature of its network and its low input costs.

Miles claimed that "by tapping into the worlds’ largest pool of compute resources (consumer devices, internet connected, and sitting overwhelmingly idle) our infrastructure allows us to grow to a scale that will rival the hyperscalers’ over time."

He also touted the company's low input costs that let the vendor compete on price and "set the floor" for compute resources like storage, bandwidth, and compute cycles. Since Salad doesn't manage the upgrade cycle, maintenance, or electricity costs, it can offer "a managed container service that costs a fraction that of the competition."

Computational resources will be the "core commodities of this century," and Salad expects its model will set the market.

"As Salad gains market share, the incumbents face the 'innovators dilemma' and will face a challenge protecting their margins. Only time will tell whether we end up supplementing one of the hyperscalers via an acquisition or whether we become an entity of our own similar to how Airbnb and Uber did within the hotel and transport categories," Miles said.