Arena Technologies Inc., the developer of a cutting-edge artificial intelligence platform for hardware testing and optimization, today announced it has raised $30 million in early-stage funding to scale the development of Atlas, the company’s core AI platform.

Arena was founded in 2020 by Chief Executive Pratap Ranade and Chief Technology Officer Engin Ural with the mission to solve problems enterprise-level problems by applying machine learning to the physical world.

The development of Atlas, an AI hardware engineer grounded in applied physics with an understanding of the real-world test environment and electrical engineering, became an outgrowth of this vision. “Atlas is really like an expert assistant,” Ranade told SiliconANGLE in an interview.

To start, Atlas can absorb a vast amount of data from manuals, PDFs, pinout schematics, design files and other sources using large language model capabilities so that it can assist a hardware engineer with their work. In this way, Atlas functions as an extremely brilliant engineer who knows the hardware inside and out.

“There’s this part of Atlas that’s very useful where it can look at your design and tell you how it should work,” explained Ranade.

For example, the model enables engineers to run multiple simulations across different conditions that would traditionally take an afternoon in just a short time. It can also automatically assist with generating test plans based on the hardware design, giving engineers a comprehensive testing strategy.

“If you think about Atlas that’s where the other part comes in, which is you can actively interact with the system and learn,” said Ranade. “That’s why the advances in LLMs have been helpful, because you can take the body of knowledge we have around electrical engineering, all of the schematics and start from somewhere, but that’s not where you finish. That gives us the starting point.”

Atlas integrates data from various sources, such as oscilloscopes, logic analyzers, multimeters and thermal/optical imaging in the testing lab, to provide a comprehensive understanding of system performance. That allows it to assist with diagnosing issues by correlating test data with expected behavior and identifying potential root causes of failure and poor component performance.

Test engineers can talk to the AI in plain English while they’re performing diagnostics and it will apply its expert knowledge to speed up the process.

According to Ranade, it doesn’t stop there. Atlas can also help engineers optimize device performance by exploring a much wider range of parameters than would be feasible manually. Across the company’s customer base, Arena said, users have seen a 35% reduction in engineering person-hours, 65% faster time to market and a 3% improvement in complex product quality.

In the last 12 months, Atlas has been brought into full-scale production in Fortune 500 companies such as microchip maker Advanced Micro Devices Inc. and vision care company Bausch & Lomb Inc. Atlas was piloted at AMD in 2023 for the company’s Radeon graphics processing unit testing and optimization, which allowed engineers to focus on other tasks.

“Atlas helps streamline the manually-intensive and lengthy process of configuring and testing today’s complex GPU technologies, giving our engineers a leg up in maximizing the optimizations for next-generation products,” Andrej Zdravkovic, senior vice president and chief software officer at AMD.

Arena’s Series B round was led by existing investors Initialized, Fifth Down Capital and Goldcrest Capital. The round also brought continued support from investors including Peter Thiel’s Founders Fund, Shield Capital and Garuda Ventures. The current funding brings the total raised by the company to $62 million.

Ranade said that with the funding, Arena intends to scale up the deployment of Atlas across customers and further develop underlying models and agentic capabilities. Arena also announced the development of Atlas Edge, where the AI platform will be deployed directly onto machines, enabling self-diagnosis and self-repair.

Image: SiliconANGLE/Microsoft Designer