Skip to content

Startup raises £1.5m to repurpose crypto mining rigs to train AI

Tromero
Tromero co-founders: Charles Higgins and Sophia Kalanovska. Image credit: Tromero

Tromero, a startup founded to repurpose crypto mining networks to train AI, has raised £1.5m in a seed funding round.

The startup is aiming to meet the growing demand for computing power to train AI systems. Tromero, which was founded last year, said it plans to solve this issue by combining the workload of AI training with the existing compute-intensive process of crypto mining.

Crypto mining involves solving complex maths problems in exchange for cryptocurrency payments. Many crypto miners invested in expensive GPU setups when cryptocurrency prices were high, but now find them less lucrative to run since the value of cryptocurrencies like Bitcoin have drastically fallen from their peaks.

Those same GPUs are also well-suited for training machine learning (ML) models. Tromero is looking to exploit these underutilised computing assets and make them available via a blockchain platform.

This would be an alternative to centralised cloud-based GPU clusters that companies currently use to train AI models. Crypto miners would earn income in cryptocurrency or fiat for providing computing power.

A spokesperson for Tromero told UKTN: “As long as it involves a sufficiently difficult ML model (i.e. a large neural network), we can work with it”.

“In an industry that struggles to keep pace with the soaring demand for machine learning training, our new way of mining cryptocurrency ensures that GPUs are no longer wasted on meaningless hashing,” said Tromero co-founder Sophia Kalanovska.

“Our system breaks down some of the barriers that inhibit the creation of impressive AI models, fostering a future where AI is created by everyone for everyone.”

The funding round was led by BlueYard Capital, a Berlin-based investor in decentralised startups, and will be used by Tromero to build a team to complete the company’s upcoming beta version.

The company’s website states it plans to launch its service in 2024.

“It is certain that the demand for training and fine-tuning ML [machine learning] models will continue to rise, all while compute costs remain high,” said a spokesperson for BlueYard Capital.

“Tromero provides a solution to these cost and supply pressures by supplying compute providers and miners a stream of revenue to optimize their return on capital. Given the industry’s inability to keep up with demand for machine learning hardware, we expect crypto-economic incentives to ensure there are no underutilised GPUs.”

Topics

Register for Free

Get daily updates and enjoy an ad-reduced experience.

Already have an account? Log in