Some say that chef training is useless, and some question the relationship between chip engineers and apprentices, Jingxin believes that these precisely expose the same problem—whether the industry is heading towards “hollowing out” in the name of “efficiency”? Just as the controversy over Xibei’s pre-prepared dishes is astonishingly similar to the SoC design model that relies on ARM IP assembly, we should perhaps pause to consider: Is the so-called shortcut consuming our true innovation capabilities?
1. High Dependence on “Standard Ingredient Packages” Reduces Innovation Threshold· Restaurant Pre-preparation: Central kitchens uniformly complete ingredient preprocessing, seasoning, and packaging, forming standard “cooking packages”. The back kitchen of the store only needs to reheat and plate according to the process to serve meals.
· Chip IPization: Manufacturers like ARM provide verified CPU/GPU IP cores, and chip teams mainly integrate and verify existing modules rather than designing every transistor from scratch. It seems to achieve high efficiency, low risk, and rapid output, even with outstanding product “presentation”. But can you truly practice innovative output?
2. Capability Trap: The Invisible Cost of Core Skills Degradation· Chefs Reduced to “Hot Dish Workers”: If overly reliant on pre-prepared dishes, core culinary skills such as knife skills, cooking time, and seasoning will gradually decline. Once the ingredient package is lost, even a simple stir-fried vegetable dish may be impossible to complete.
· Engineers Become “Integrators”: If they only engage in IP integration for a long time without deeply participating in microarchitecture design, design verification, DFT flow, timing optimization, backend implementation, and low-level instruction scheduling, the team will gradually lose its autonomous design capabilities. This is precisely why Jingxin insists on self-developing all IPs—without truly “cooking”, one can never become a “master chef”.
3. Result Dilemma: Disappearance of Differentiation, Falling into Low-level Competition· Restaurant Homogenization: When all restaurants use the same pre-prepared dish packages, the only differences ultimately lie in decoration, marketing, and service, making it difficult to establish a truly recognizable brand taste imprint.
· Chip Performance Convergence: When SoC manufacturers generally adopt the same generation of ARM IP and process technology, products can only compete on parameters like frequency and core count, making it difficult to achieve architectural breakthroughs.
4. In the Moment of “Supply Disruption”, Can We Survive?· Restaurant Supply Disruption: If the central kitchen stops supplying, does the back kitchen still have the capability to start from scratch to prepare a meal?
· Chip Supply Chain Disruption: If ARM’s licensing is terminated, TSMC’s processes become unavailable, or EDA tools are blocked, can the team still come up with alternative solutions or even achieve tape-out independently? What is the choice of truly top players? The reason why Apple’s A-series chips are powerful is that they have never stayed at the “integration” level. They only use the ARM instruction set architecture as a “recipe outline”, with microarchitecture design and core IP development all self-developed. This is akin to a chef not only seasoning but also brewing their own soy sauce, thus achieving absolute control over flavor and experience.
Jingxin firmly believes that whether it is chefs or chip engineers, “skills” cannot be lost, and “foundation” must be practiced. Training and the apprenticeship system may seem “slow”, but they are the fundamental path to avoid hollowing out.
