May 25, 2022

Self-driving cars and other applications developed with AI are known as “deep learning”, the basic concept of which originated in the 1950s. This requires training models based on patterns similar to the human brain. This in turn requires a lot of processing power provided by TPUs (Tensor Processing Units) or GPUs (Graphics Processing Units) that are used for a long period of time. However, the cost of this computing power is out of reach for most AI developers, who mostly rent it from cloud computing platforms like AWS or Azure. What can you do there?

Well, one approach comes from British startup Gensin. This [email protected] And took the idea of ​​distributed computing power from older projects like COVID-19-focused [email protected] and applied it to this deep learning AI developer quest. The result is a way to get powerful computing power from a distributed network of computers.

Gensyn has now raised a $6.5 million seed under the leadership of Eden Block, Web3’s venture capitalist. Also participating in the round are Galaxy Digital, Maven 11, Coinfund, Hypersphere, Zee Prime, and the founders of some blockchain protocols. This is in addition to a previously unannounced $1.1 million pre-investment in 2021 to be led by 7 Percent Ventures and Counterview Capital, with input from Entrepreneur First and ID4 Ventures.

In a statement, Gensyn co-founder Harry Grieve said, “Rising demand for hardware – and high margins – is why big names like AWS and Azure are vying to capture such high market share. As a result, the market has become expensive and centralized… We have developed a better approach: better in price, with unlimited scalability and without gatekeepers.”

Jensin says that to do this, he will launch his own decentralized computer network for training AI models. This network uses the blockchain to validate deep learning tasks by triggering payments with tokens. It then generates unused processing power in a verifiable manner. Gensyn also claims it’s a greener solution, as that processing power would otherwise go unused.

Lior Messika, Managing Partner of Eden Block commented: “GenSyn is committed to truly democratizing computing through decentralized technology, which is arguably the most ambitious enterprise we have ever faced… The team aims to become one of the largest and fastest growing markets. Industry.” world, dramatically reducing the costs and friction associated with large-scale training of neural networks.”

Speaking to me, Grieve said, “Our estimate is up to 80% cheaper per unit of a standard Nvidia GPU, or 45 cents per hour, on average, compared to about $2 per hour for other cloud services.”

Leave a Reply

Your email address will not be published.