Workplace reviews and discussion: 全智科技股份有限公司

This is the most magical aspect of the capital markets. Intuitively, one might think, “To claim that it’s a smart factory, wouldn’t the information systems alone have to be worth at least several million, or even tens of millions?” But for semiconductor assembly and testing plants of this kind, they use “million-dollar hardware to package hundred-thousand-dollar software.” In traditional manufacturing, this is actually a very common “arbitrage leverage” tactic.

They can use a lousy system costing a few hundred thousand to borrow several billion. The key is that evaluators and banks simply “don’t understand the technology”; they only look at “physical assets”:

• Hardware is the protagonist; the system is merely a supporting actor (there for decoration):

When the company applies for government-subsidized-interest loans or seeks bank financing, the main assets are those high-end semiconductor testing machines, often costing tens of millions of New Taiwan dollars each and imported from the United States or Japan (such as Teradyne and Advantest). Ninety-five percent of a budget worth several billion is tied up in this hardware.

• “Minimum-standard automation” that gets evaluators to nod:

When the government reviews “smart manufacturing subsidies” or loans, its standards are extremely superficial. As long as this system costing a few hundred thousand can achieve “connecting the machines to a network” and “periodically sending data back to some database” (even if it is merely dumped there and everything afterward relies on people manually pulling data into Excel), the proposal can check the box for “smart manufacturing and automation capabilities already in place.” Most evaluation committee members are university professors or government officials. During an on-site inspection, the company only needs to open a very crude dashboard that periodically displays changing numbers for the committee, and it passes.

• Banks recognize collateral, not source code:

For a bank, the risk in lending several billion is whether the company will go under. If the company buys testing machines, those machines are hard assets; if the company actually collapses, the bank can still auction them overseas and recover cash. But if the company spends tens of millions hiring a top-tier IT team to build an ultra-modern cloud architecture, then if the company goes under, that code is just a pile of worthless garbage to the bank, with no liquidation value whatsoever.

This creates an extremely distorted ecosystem: a company would rather spend 5 billion buying the world’s top sports car—the hardware equipment—but, to save money and satisfy inspections, it is willing to spend only 200,000 on a cheap plastic toy steering wheel bought at the roadside—the IT system—and bolt it on.

As long as the car can move (the equipment can perform tests), the government provides subsidies, and the bank is willing to lend money, senior management simply does not care how the IT inside is manually operated every day, or how employees use Excel to somehow make this lousy steering wheel flush the toilet. After all, from their perspective, the cost of the “human API” has long since been shifted onto frontline employees in the form of overtime pay.

o