Recently, I’ve been seeing news about “AI automatic routing efficiency skyrocketing,” and many engineers who design circuit boards (PCBs) are starting to worry: Is my skill about to be replaced by AI? Don’t panic, let’s discuss what AI can and cannot do; the situation is quite clear: AI is a quick and efficient “helper,” not here to “take jobs.”
AI has indeed made board design a bit faster:
Time-saving “skilled worker”: Those with fixed patterns and simple component placements and routing, AI can do quickly and well, shortening the initial time for board design.
Quality-checking “inspector”: It can automatically check if the lines are drawn too close or too thin, or whether the designed board can be manufactured (DFM), reducing the hassle of rework later.
Providing suggestions “advisor”: For example, how to improve heat dissipation or how to adjust line impedance, it can provide engineers with some reference suggestions.
However, AI’s “shortcomings” are also quite obvious:
Cost is not feasible: Google invested $200 million in an “AI automatic circuit design” project, but it didn’t sell well! Why? Because customers find using AI for design more expensive than hiring engineers! Currently, the industry generally charges per pin, about $3 per pin. In this model, replacing human labor with AI on a large scale is considered unprofitable by bosses.
Can only serve as a high-level “assistant”: Current AI tools are essentially smart assistants. They excel at optimizing within existing, familiar patterns. However, when faced with completely new designs that require innovation from scratch, they become stumped—because they haven’t “learned” it!
Therefore, the core value of PCB engineers remains strong!
True “hard skills” in circuit board design lie in: Finding the best balance among various conflicting requirements (power saving, heat dissipation, stable signals, low cost, ease of production…) and breaking conventions to achieve innovative designs. These two aspects are precisely what AI cannot handle!
Why can’t AI completely replace human engineers? The key points are:
1. Making choices and balancing is an art: In the design of circuit boards for high-performance graphics cards or 5G base stations, engineers must simultaneously address many conflicting requirements: power saving, good heat dissipation, stable signals, no interference, low cost, and ease of production. AI can help calculate data and provide suggestions, but the final decision on how to balance these conflicts relies on the engineer’s experience and intuition. For example, when designing a mobile phone, the antenna (RF) components cannot be crowded together, but the structural engineer may say the space is too small and must be squeezed. In this case, an experienced engineer might come up with a clever solution, such as modifying the grounding design, to satisfy both sides. AI typically just raises an alarm saying “not possible” but cannot provide a good solution.
2. Innovation starts from scratch relies on human brains: For example, the circuits in foldable smartphones that can bend or the complex circuit boards packed into small watches have never been done before, and there are no existing templates for AI to learn from. Designing these novel gadgets relies on engineers’ deep understanding of circuit principles and their imaginative creativity. AI is good at optimizing existing things, but starting from scratch to invent entirely new designs is too difficult.
3. Solving “mysterious” faults relies on experience: Circuit boards sometimes have strange problems, such as not working outdoors in the winter in Northeast China, or occasionally experiencing inexplicable interference, making it hard to trace the cause. Resolving these “mysterious” issues relies on engineers’ years of accumulated experience and the knowledge gained from working in factories. For instance, if an industrial device has problems every winter, an experienced engineer might guess that a certain component (capacitor) is sensitive to cold, and replacing it solves the issue. AI, having never encountered such extreme situations, cannot identify the cause.
4. Acting as a “translator,” coordinating various parties is essential: Designing circuit boards is not something engineers do in isolation. They must understand the hardware architect’s ideas, address the structural engineer’s spatial constraints, meet the procurement department’s cost control requirements, and teach the factory how to produce the boards. This ability to communicate and coordinate across different departments, resolve conflicts, and turn ideas into reality is something current AI cannot achieve.
How to proceed in the future? The combination of humans + AI is the way forward!
Don’t worry about AI taking jobs; the real concern should be those engineers who refuse to use new tools. The future will belong to engineers who use AI, while those who do not will be eliminated.
1. Let AI be your efficient little assistant: In the future, let AI handle repetitive tasks like routing and component placement (especially for simple, regular parts), such as over 30% of basic routing. This way, engineers can free up their hands to focus on more valuable tasks—making critical decisions, driving innovative designs, and solving truly challenging problems.
2. New opportunities are emerging in high-end fields: AI servers (often featuring ultra-complex high-speed boards with 28 or even 46 layers), electronics for new energy vehicles (which have very high requirements and cannot fail at all), and high-density flexible boards used in foldable smartphones/AIPC… These new technologies are highly demanding and particularly rely on engineers’ overall design capabilities. In these areas, AI cannot yet compete, making it a great stage for engineers to showcase their value.
You need to evolve and upgrade:
Hard skills need to be deeper: Focus on high-speed circuit design (such as the currently popular DDR5/PCIe 5.0), truly understand why signals degrade, why power supplies become unstable (SI/PI), learn how to design to avoid interference (EMC), and also manage heat dissipation.
Seize new fields: Target those areas with new technologies, high demand, and high barriers, such as designing circuits for millimeter-wave radar, developing automotive-grade electronic systems, and mastering advanced packaging technologies. Focus on areas with high demand, technical difficulty, and good compensation, such as AI hardware, automotive electronics, and advanced packaging.
Learn to be the “boss” of AI: Treat AI as a capable assistant; it works quickly, but you need to know how to give commands, check, and modify! Don’t trust it completely. Familiarize yourself with the AI features in mainstream design software (like Altium Designer, Cadence). Look at how NVIDIA did it: when designing the H100 chip board, they first used AI to handle 30% of the basic routing, saving engineers time to focus on optimizing the critical GPU power supply and heat dissipation, resulting in a 40% reduction in development time and even better performance!
Proactively learn to use AI tools, but remember you are the decision-maker; use the solutions it provides as references and optimize them with your professional insight.
Systematic learning and honing real skills are more important than anything else!
In the face of changes brought by AI, systematic learning and enhancing practical abilities are key to remaining competitive. Many engineers have reported rapid progress by participating in professional PCB training programs like those offered by Fanyi Education, which focus on solving complex problems and cultivating design thinking. Their courses emphasize practical knowledge and systematic understanding, which can indeed help accelerate advancement in the era of electronic design. I recommend Fanyi Education’s PCB training.
Remember: AI is a powerful tool, but the ability to use that tool skillfully to solve problems that others cannot is what will always keep you valuable! True job security comes from transforming tools into wisdom. AI provides a precise “ruler,” but it is the human designer who understands the essence and dances on the edge of physical limits that truly shapes the soul of the circuit.