Business Insider: SK hynix could reach $1 trillion valuation as AI‑chip demand surges

SK hynix is close to a $1 trillion valuation because the AI buildout is no longer centered only on the graphics processor. The South Korean memory maker has become one of the clearest public-market ways to own the part of the stack that sits next to Nvidia’s accelerators: high-bandwidth memory, or HBM, the dense and power-hungry memory used in AI servers. Reuters reported on May 14 that SK hynix was on the verge of topping a $1 trillion market value, weeks after Samsung Electronics crossed that threshold. SK hynix’s own numbers show why investors have focused on memory. The company said on April 22 that first-quarter 2026 revenue rose to 52.5763 trillion won, with operating profit of 37.6103 trillion won and net profit of 40.3459 trillion won, all record quarterly results, driven by sales of higher-value products tied to AI demand. (money.usnews.com) ### Why is a memory company suddenly being discussed with trillion-dollar stocks? May 14 is the date Reuters tied to the latest move in SK hynix shares, but the underlying driver has been building for several quarters. AI systems need HBM chips placed close to GPUs to move data faster than conventional server memory can handle, and SK hynix has become a key supplier into that market. (news.skhynix.com) Reuters said strong artificial-intelligence demand had put South Korea at the center of Asia’s AI boom. January 28 is the date SK hynix posted its 2025 annual results, and those figures showed how sharply the business had changed. The company reported 97.1467 trillion won in 2025 revenue and 47.2063 trillion won in operating profit, citing demand for AI memory and other high-value products including HBM. (money.usnews.com) ### What exactly does SK hynix sell into AI servers? HBM is the product that matters most here. SK hynix said it completed the world’s first HBM4 development and had finished preparations for mass production, and it has also said 12-layer HBM4 samples were shipped to customers with mass-production preparations targeted for the second half of the year. (news.skhynix.com) March 2025 is when SK hynix said it became the first company in the industry to supply 12-layer HBM4 samples to major customers. At Computex Taipei 2025, the company said the product was slated for mass production in the second half of that year and outlined a 16-layer HBM4 roadmap for 2026. (news.skhynix.com) ### Why does this story extend beyond memory chips? Nvidia has been expanding the list of hardware bottlenecks around AI clusters. On March 18, 2025, Nvidia introduced Spectrum-X and Quantum-X silicon-photonics networking switches, saying they were designed to connect millions of GPUs across sites while cutting energy use and operating costs. Nvidia said the systems were developed with partners including TSMC, Coherent, Corning, Foxconn, Lumentum and SENKO. (news.skhynix.com) May 5 is when Nvidia and Corning announced a multiyear partnership to expand U.S.-based optical-connectivity manufacturing for AI infrastructure. Corning said it would increase U.S. optical-connectivity manufacturing capacity by 10 times, expand U.S. fiber production capacity by more than 50%, build three new manufacturing plants and create more than 3,000 jobs. (nvidianews.nvidia.com) ### Where do the constraints show up if GPUs are not the only bottleneck? Nvidia’s own revenue mix points to how much infrastructure now sits around the accelerator. The company said on February 25 that fourth-quarter fiscal 2026 data-center revenue reached $62.3 billion, up 75% from a year earlier, as customers continued building AI systems at scale. (nvidianews.nvidia.com) Those deployments pull on several categories at once: HBM for memory bandwidth, optical components for longer-reach links, switches and NICs for networking, and power and cooling equipment for dense racks. The direct evidence in current company disclosures is clearest in memory and optical networking, where SK hynix is posting record profits and Nvidia and Corning are adding U.S. manufacturing capacity. (investor.nvidia.com) ### What should investors watch next? The second half of 2026 is the next milestone in SK hynix’s own roadmap. The company has said it is preparing mass production for 12-layer HBM4, while continuing to position HBM4 as its next major AI-memory product. May 28, 2026, is Nvidia’s scheduled date for reporting first-quarter fiscal 2027 results, according to its investor relations calendar, and that report will give the next official read on data-center demand. (news.skhynix.com) SK hynix’s next quarterly update will be the next company filing to watch for pricing, margins and supply commentary tied to AI memory. (investor.nvidia.com) (news.skhynix.com)