A naukri.com initiative
TechCrunch
1M
251
Image Credit: TechCrunch
TensorWave thinks it can break Nvidia’s grip on AI compute with an AMD-powered cloud
- Chipmaker TensorWave is using AMD’s Instinct MI300X GPU to challenge Nvidia’s monopoly on artificial intelligence compute hardware.
- TensorWave, founded in late 2022, is a cloud service that only deploys instances of AMD’s MI300X and requires a minimum of a six-month contract with a range of hourly prices depending on the workload and the GPU configurations chosen.
- Co-founder Darrick Horton said the team wanted to democratise the AI industry, which had an unhealthy monopoly, by offering an alternative choice.
- Horton noted the MI300X is considerably cheaper than Nvidia’s H100 GPU – widely used by the industry – and believes it will appeal to developers seeking alternatives that can handle AI workloads effectively.
- With the cloud space as a whole enjoying huge growth, TensorWave is already generating $3 million in annual recurring revenue and is projecting $25 million by year-end.
- TensorWave’s pricing strategy means it is competing in a market that includes GPU infrastructure providers such as CoreWeave and Lambda Labs.
- The company’s burn rate is believed to be “well within sustainable levels” and it has recently received $43m in funding led by Nexus VC, with participation from AMD Ventures, Maverick Capital, StartupNV and Translink Capital.
- TensorWave is planning to bring online the MI325X GPUs once they are released in late 2024, and also plans to turn the GPUs into collateral for a large amount of debt financing.
- Horton noted there will be challenges ahead in taking on Nvidia. Its development software is more mature and easier to use than AMD’s, while the world’s largest companies continue to invest in their own custom hardware.
- TensorWave was founded by three entrepreneurs who had limited knowledge of the hyperscaler landscape but believed they could solve the GPU supply problem driving their formation.
Read Full Article
15 Likes
For uninterrupted reading, download the app