Philoptics builds 2mm glass drilling on a bet that AI chip substrates outgrow today's 0.8mm

The hard part of 2mm is the middle of the hole. As glass gets thicker, laser energy is harder to deliver evenly to the interior, and a via can stop short or narrow into a "waist" at its center ^[8]. Philoptics says its optics now drill the same shape from top to bottom in a single pass ^[8]. "With thick glass, the issue is not simply whether the hole goes through. What matters is keeping the difference between the diameter at the center and the diameters at the top and bottom within a set standard," said Kang Sang-ki, president of Philoptics (161580.KQ) ^[9][3]. "We have raised our technology to the level where we can process 2-millimeter thicknesses without defects," he said ^[10].

Kang's case for thickness rests on package size. "As artificial intelligence chips demand higher performance and more power, package sizes grow larger, and glass substrates inevitably have to become thicker and stronger to support them stably," he said ^[4]. If packages grow and customers move to thick glass, Philoptics arrives already verified: "The thickness verified technically so far is around 2 millimeters, but we can handle thicker glass substrates if customers require it," Kang said ^[6]. Should customers settle near the 0.8mm they are developing now, the 2mm work is headroom above the spec they chose ^[2]. A stall before mass production would make thickness irrelevant, because the order book is tied to the 2028 date ^[17].

I think the more consequential remarks were about the steps after drilling. "Looking only at our TGV technology, delivering product performance and securing reliability are no longer major issues," Kang said ^[13]. "The biggest hurdle to commercializing glass substrates is whether customers can achieve the yield levels they want in actual mass production through the other processes," he said ^[14]. A supplier that has solved its own step has its orders timed by the plating, polishing and wiring lines that follow it ^[12]. The counter-case is that the first vendor qualified on thick glass keeps the account whenever those yields arrive. Kang frames the 2mm work that way: "We expect glass substrate demand to take off, so we are securing 2-millimeter TGV technology ahead of the market to take the lead," he said ^[5].

The speed target is a second allocation choice. About 30 minutes for 500,000 to 1 million holes is some 280 to 560 holes a second, and halving it, the goal for next year, moves one tool from 2 substrates an hour to 4 ^[15][2][3]. The company says the point is to ease customers' capital spending, since laser equipment is expensive ^[16]. For a fab with a fixed output target, twice the throughput per tool means half as many tools ^[4]. Unless the price per tool rises, Philoptics is choosing to sell fewer units per line to make its bid cheaper ^[4]. The interview did not cover tool pricing.

Mass-production equipment is usually bought about a year ahead, so a 2028 commercialization date puts the order window in 2027 ^[17][5]. "Next year, we expect semiconductor equipment to surpass display equipment on an order basis," Kang said ^[18]. For a company founded in 2008 on OLED laser tools, display orders still leading next year would be the first sign that the 2028 date has slipped ^[7][17]. "Our goal is to transform from a company centered on displays into a global semiconductor equipment maker centered on glass substrates," he said ^[19].