The history of Taiwan's technology boom has seen no shortage of unexpected heroes, but the case of 85-year-old Lin Tsung-chi, founder of King Slide Works, stands apart. His company, which started out manufacturing furniture hinges and drawer slides, has today become a key beneficiary of the AI race, and Lin himself has topped the island's rich list.
Since the start of the year, King Slide's shares have surged nearly 280%. The secret to its success lies not in microchips, but in seemingly mundane mechanics: rail mechanisms for server racks. These components hold up the heaviest computing equipment and allow engineers to slide it out for maintenance without disrupting the delicate cooling system. By my estimates, based on industry data, the company controls about 80% of the global market for such slides used in high-performance servers.
Margins Nvidia would envy
Demand for this product is growing in tandem with the construction of data centers and the proliferation of increasingly powerful AI systems. When a server rack costs millions of dollars, the price of the slides takes a back seat—flawless reliability comes to the fore. This understanding has transformed King Slide's financials: in the latest quarter, the company's gross margin reached a phenomenal 87%, compared with around 50% a few years ago. For context, Nvidia's figure stands at 75%, while TSMC's is 68%.
King Slide's Executive Vice President, Jay C. Wang, attributes such profitability to two decades of engineering development that enabled the creation of highly specialized mechanisms. The result is evident: according to Forbes estimates, Lin Tsung-chi's wealth has reached $20.3 billion, allowing him to overtake Foxconn founder Terry Gou, whose fortune has also grown thanks to the AI boom.
New architectures, new opportunities
Ahead of King Slide lie even more enticing prospects. Cloud giants are actively transitioning to their own AI accelerators, increasing equipment density, and adopting liquid cooling. This requires a complete redesign of servers and racks. Each new accelerator differs in size and heat output, thus requiring a unique rack and, consequently, unique slides.
The trend is obvious: Google has already secured an option to purchase up to 58.97 million shares of Marvell Technology under an expanded AI chip agreement, and Alibaba plans to raise $10.2 billion to develop its AI infrastructure. Data center density is rising: the declared capacity of gas power plants for their direct power supply in the U.S. has increased from 97 GW to more than 189 GW by mid-2026.
However, competition is intensifying. According to Daiwa Securities, King Slide's share of shipments for Nvidia system kits could decline to roughly 75% next year, as the chipmaker expands its supplier pool. Large data center operators, having alternatives, will be able to bargain more aggressively on prices. Nevertheless, the business's fundamental dependence on AI infrastructure investment remains direct and indisputable.
My view: King Slide's story is a striking example of how "secondary" players in the AI chain can earn margins unavailable even to semiconductor industry leaders. It is a signal for investors to take a closer look at the less obvious but critically important components of the AI ecosystem.