Looking to break NVIDIA's monopoly! Startup company TensorWave's cloud service only uses AMD chips

The AI ​​boom is in full swing, making NVIDIA the undisputed leader, with its GPUs becoming the preferred chips for AI companies of all sizes. However, media reports that TensorWave, founded at the end of last year, is making waves by launching a cloud service that only offers AMD hardware, a competitor to NVIDIA, to handle AI workloads.

Darrick Horton, CEO and co-founder of TensorWave, said in an interview with TechCrunch:

"We realize that this is an unhealthy monopoly phenomenon, one that prevents end users from accessing computing resources and stifles innovation in the AI field. In our desire to democratize AI, we have decided to provide a viable alternative to restore competition and choice."

Settling in Sin City

According to reports, Horton, along with TensorWave's other two co-founders, Jeff Tatarchuk and Piotr Tomasik, met through the sport of Pickleball and later reached a consensus on the issue of supply restrictions due to GPU computing power monopolies, ultimately leading to TensorWave.

Based in Las Vegas, TensorWave's headquarters is an unusual choice for a cloud infrastructure startup. However, Horton stated that the team saw an opportunity there. "We believe Las Vegas has the potential to become a thriving ecosystem for technology and startups."

This prediction is not entirely detached from reality. According to Dealroom.co data, Las Vegas is home to over 600 startups, employing more than 11,000 employees, and attracted over $4 billion in investments in 2022.

Compared to many major US cities, Las Vegas has lower energy costs and operating expenses. Additionally, Tomasik and Tatarchuk have close ties to the local venture capital community.

Tomasik previously served as a general partner at the Las Vegas-based seed fund 1864 Fund and now collaborates with the non-profit accelerator StartUp Vegas and Vegas Tech Ventures. Tatarchuk is an angel investor at the incubator Fruition Lab.

These connections have helped TensorWave become one of the first providers in the market to offer cloud services handling AI workloads with AMD Instinct MI300X instances. TensorWave offers GPU computing resources for rent by the hour and requires a minimum contract term of at least six months.

Centered around AMD

Media reports that the low-cost, on-demand GPU cloud market is booming. Horton, however, states that TensorWave has a clear competitive advantage.

First is the price. Horton points out that the AMD MI300X is much cheaper than the currently popular NVIDIA GPU H100, allowing TensorWave to pass on cost savings to customers. He did not disclose the specific pricing of TensorWave's instances, but to beat the H100, pricing must be below $2.50 per hour Next is performance. Horton cited benchmark test data showing that the MI300X outperforms the H100 in running (but not training) AI models, especially in text generation models like Meta's Llama 2.

In addition to TensorWave, other companies betting on AMD AI chips include startups Lamini and Nscale, as well as larger, more established cloud providers such as Azure and Oracle.

Analysis suggests that the current favorable factors for AMD chip users are the ongoing shortage of Nvidia GPUs and the delay of Nvidia's upcoming Blackwell chip. However, as the manufacturing of key chip components (especially memory) accelerates, the shortage may soon ease. This will allow Nvidia to increase shipments of the H200, a new product with significantly improved performance.

For emerging cloud service providers relying on AMD hardware, another challenge is how to overcome Nvidia's competitive barriers built around AI chips. It is widely believed in the industry that Nvidia's development software is more mature and easier to use, and has been widely deployed. Even AMD CEO Su Zifeng admitted that adopting AMD chips "requires effort."

Looking further ahead, as mega-companies increase investment in custom hardware, the challenge of focusing on price competition may become increasingly difficult. For example, Google offers its TPU; Microsoft recently launched two custom chips, Azure Maia and Azure Cobalt; and Amazon AWS has Trainium, Inferentia, and Graviton.

But Horton said:

"As developers seek alternative solutions that can effectively handle their AI workloads, especially as memory and performance requirements increase, coupled with delays caused by production issues, AMD will maintain a leading position for a longer period in the future and play a key role in the popularization of computing in the AI era."

Expected eightfold increase in recurring revenue by the end of the year, valuation already at $100 million

TensorWave currently has $3 million in annual recurring revenue. Horton said that once TensorWave increases the capacity of its MI300X to 20,000, recurring revenue is expected to reach $25 million by the end of the year, an eightfold increase.

Assuming a cost of $15,000 per GPU, 20,000 MI300X would mean a $300 million investment, however, Horton claimed that TensorWave's cash burn rate is "completely sustainable." TensorWave had previously stated that it would use its GPUs as collateral for large-scale debt financing, a method adopted by other data center operators, and Horton said this is still part of the company's plan.

He continued:

"This reflects our strong financial health. By providing value where it is most needed, strategically positioning ourselves to address potential headwinds." However, when asked about the current number of customers at TensorWave, Horton refused to answer due to "confidentiality reasons," but emphasized the partnership that TensorWave has publicly announced with network backbone supplier Edgecore Networks and AI inference startup MK1 founded by former Neuralink engineers.

Horton stated that the company is rapidly expanding its computing power to meet the growing demand. He also added that TensorWave plans to use the next-generation MI325X GPU from AMD, which may be launched in the fourth quarter of this year.

Investors are satisfied with TensorWave's growth trajectory. Nexus VP revealed on Wednesday that the institution led a $43 million financing round, with participants including Maverick Capital, StartupNV, Translink Capital, and AMD Ventures. This is TensorWave's first round of financing, and the company's valuation reached $100 million after the financing