Recently, an agent-based AI was able to design and verify a RISC-V CPU from scratch in under 12 hours. How is AI becoming a critical engineering tool, what did the AI achieve, and what does this mean for the future of AI in engineering?
How is AI Becoming a Critical Tool in Engineering?
For many, the very idea of AI brings in thoughts of hatred and repulsion. This makes perfect sense when considering how AI has reduced employment opportunities, stripped us of creativity, and caused harm in unusual ways. But at the same time, it is also becoming a powerful tool that, if used correctly, could revolutionize how we live.
The field of engineering is no different, and AI is already making a massive impact. One area where AI is quickly gaining traction is predictive maintenance. By analyzing large amounts of data from sensors, AI systems can identify patterns that indicate potential equipment failures before they occur. This allows engineers to take proactive measures to prevent downtime, minimize costs, and enhance overall efficiency.
Another area where AI is making a significant impact in engineering is in the design process. Engineers are able to leverage AI algorithms to generate unique designs that are optimized for specific requirements. For example, special use RF antennas with designs and shapes that humans would never dream of can be designed entirely through AI.
Additionally, AI is also finding its way into design software, assisting engineers in making better decisions. From identifying potential issues in the design stage to recommending alternative components, AI tools can provide valuable insights that would otherwise be time-consuming and difficult to obtain.
AI is also increasingly supporting engineers in writing firmware and code, automating routine programming tasks and offering suggestions that can speed up development. However, it is important to note that while AI can accelerate coding, it can also make serious mistakes if not carefully supervised, making human oversight essential for safety and correctness.
Despite the considerable promise, AI systems are not infallible. Engineers who rely solely on AI risk introducing errors or missing important design considerations that require human judgment and experience. The best results are achieved when AI works alongside human experts, augmenting their abilities rather than replacing them.
AI Designs RISC-V CPU in Under 12 Hours
Recently, a new startup company called Verkor.io, has managed to use an agent-based AI system called Design Conductor to design a complete RISC-V processor core in just 12 hours. The system, which operates like a virtual engineer, is able to cover all stages of the design including the specification, verification, testing, and error correction without any human interaction whatsoever.
The result of this experiment was a functioning RISC-V processor called VerCore that features a five-stage pipeline, in-order, single-issue architecture, and a target clock speed of 1.48 GHz. While the score obtained by the chip in CoreMark was comparable to older low-end CPUs (around Intel Celeron SU2300), the fact that the chip was designed from scratch, and developed in just 12 hours, is an impressive feat of engineering.
According to the engineers who developed the AI, the 12-hour time frame comes from the fact that the AI was only given a 219 word brief to create the RISC-V processor, something that would typically see a team of engineers working for 18 to 36 months. At the same time, the engineers also noted that the current state of AI does not allow for more complex processors with multiple pipelines and out-of-order execution. It's important to note that the VerCore processor is not a physical chip yet. The design currently exists as a high-fidelity simulation, which was tested using the Spike emulator. The design was created using the ASAP7 7nm process design kit, and while fabrication has not yet taken place, the achievement demonstrates the ability of AI systems to rapidly iterate and validate complex digital designs up to the GDSII layout stage.
This marks a significant advancement compared to previous AI-assisted design efforts, which were often limited to generating code or optimizing small design modules. Here, the AI coordinated the entire workflow autonomously, from specification through design and verification, without intermediate human intervention.
Despite these advances, the AI-generated processor remains limited to relatively simple, in-order architectures, and does not yet rival the latest human-designed CPUs in performance or complexity. Nonetheless, the project highlights the growing potential of AI to serve as a powerful prototyping tool and to dramatically accelerate chip design cycles, potentially reducing time-to-market for new products and enabling rapid exploration of novel architectures.
What Does This Mean For the Future of AI in Engineering?
There is no doubt that AI isn’t going to go away anytime soon, and its involvement in engineering will only continue to grow. Even though the processor designed by the AI has a poor performance, the fact that it was able to create a working design from scratch demonstrates that it has massive amounts of potential.
In the near future, AI-driven tools are expected to become standard components of engineering workflows. By automating routine and repetitive tasks, these tools will allow engineers to focus more on creative problem-solving and innovation. Faster design cycles will help bring new technologies to market more quickly, supporting advances in fields ranging from consumer electronics to scientific computing.
Of course, this doesn’t mean we will be living in a Utopian society where robots do everything, and we lounge about waiting to be doted on. No, we will still be expected to do hard work and slog through life, but spend less time doing tasks that hinder our creativity and thought processes.
Ultimately, the demonstration of AI designing a RISC-V CPU from a short specification in under 12 hours is an early indicator of what these technologies may soon make possible. While AI is unlikely to replace engineers entirely in the foreseeable future, it is absolutely destined to become an indispensable partner, enabling greater productivity, reducing human error in complex tasks, and opening new opportunities for innovation.
As these systems mature, their role in engineering will continue to expand, shaping the future of design and development across multiple disciplines.
