AI is Moving to the Edge
Cloud-based AI is broken. It’s too slow, too expensive, and insecure. So why are people still trying to move massive amounts of data to the cloud for AI computations?
According to Bloomberg, artificial intelligence (AI) will add $15.7 trillion to the global economy by 2030. PwC says global GDP to be 14% higher in 2030 as a result of AI.
The future for AI is bright. Unfortunately for cloud providers, AI growth will not be happening in the cloud. Cloud latency is too slow for realtime computations. Moving large datasets to the cloud is also too expensive and creates security risks.
AI growth will be happening where the data resides or where it’s generated — on-premise (behind corporate firewalls) and on edge networks such as driverless vehicles, drones, and locations where IoT devices and sensors are prevalent. Until now, there hasn’t been enough processing power to compute in data centers or on edge networks.
Computes Inc. is resetting our expectations. Computes not only harnesses idle computing resources across workstations, servers, and VMs in organizations and edge networks but our decentralized Lattice technology can route computations to the right machine in the right location at the right time. This means that Computes Lattice can move AI computations to the datasets and data sources!
Lattice makes it possible for every machine to swarm on a secure software-defined mesh network and connect, communicate and compute together as if they were physical cores on a single software-defined supercomputer.
Lattice is a next generation computing platform. It’s not cloud and it’s not edge computing — everything computes everywhere! Lattice can run on Windows, Mac, and Linux (x86 and ARM) as well as VMWare, XenServer, and KVM. Lattice can also run on AWS EC2, Spot Instances, and even Lambda (as well as Azure and other cloud platforms). In fact, using Docker containers, Lattice has created the sense of universal computing. The same AI algorithms can run on AWS and Azure as well as on-premise and on edge networks (as well as on some IoT devices) without any additional infrastructure. We refer to this new computing paradigm as infrastructureless computing.
Lattice is decentralized which means that there is no central control point or single point of failure. In fact, it’s fault tolerant by design. Lattice is built for general purpose computing — meaning that it knows how to distribute serial computational tasks as well as massively parallel tasks.
Lattice is organic. Computes nodes with the same mesh crypto key can automatically swam and discover like nodes and connect with one another to form a secure private mesh computer. Lattice continues to manage AI computations in a decentralized manner even as nodes connect and disconnect from the mesh computer. Lattice routes computations to nodes closest to the datasets and data sources. If the computational time-to-live (TTL) setting expires, Lattice can move the same compute to the next best machine in the next best location.
To make things even more interesting, Lattice tasks can create additional tasks on their own. Check out the chaining tasks section of our Getting Started guide.
Since Lattice is not built on blockchain technologies, it has a very high TPS throughput (today thousands of transactions per second — tomorrow millions). It also does not have PoW or PoS mining requirements — no wasted computes. This also means that our transactions are essentially free making the Computes platform perfect for micro-payments in the near future. Last but not least, Computes nodes do not need to maintain a ledger making this platform ideal for smaller IoT devices and machines that frequently connect and disconnect from the network.
Lattice is built on top of IPFS. We use IPFS for the Computes P2P file system as well as our mesh network, machine-to-machine communications, DAG-based blockless ledger, and our decentralized queuing technology. This allows for faster TPS, no wasted PoW mining, and smaller footprint since our nodes do not maintain the ledger.
Reclaim your wasted infrastructure computing resources today. Interested in learning what computes.com can do for your business? Contact us at hello@computes.com.
Stay tuned to this blog for more exciting information about our new technology stack and development progress! You can also reach us on Twitter, Facebook, and GitHub.
