Former Intel CEO Gelsinger transforms into a venture capital investor, investing heavily in next-generation lithography and AI inference chips

Kuai Technology News on July 26, according to media reports, more than a year after leaving Intel, Pat Gelsinger finally clarified his new identity as a venture capitalist. The former CEO has joined deep technology investment institution Playground Global as a partner, focusing on early-stage investments in semiconductors, energy and next-generation computing. He set a radical goal: "Let AI improve 10,000 times instead of 10 times." In December 2024, Kissinger resigned as Intel CEO. In the 100 days after his departure, he held about 100 intensive meetings and finally decided to join Playground Global, which manages about US$1.2 billion in assets, rather than continue to run a listed company or move to private equity. At present, Kissinger's most deeply invested project is xLight, a next-generation lithography light source startup, and he personally serves as executive chairman. xLight uses free electron laser technology to provide a higher-power light source for extreme ultraviolet lithography equipment, aiming to improve the efficiency of ASML's existing equipment and explore shorter wavelength lithography paths. In June 2026, the company has received US$150 million in funding from the U.S. Department of Commerce's "Chip and Science Act" to build a prototype in New York, and is expected to start testing in 2028. According to Kissinger, AI has greatly accelerated the growth of the semiconductor market - global sales will reach US$791.7 billion in 2025, and the trillion-dollar node has been significantly advanced. He also pointed out that the focus of AI competition in the next stage will shift from training to inference chips and energy supply: dedicated inference chips are expected to bring 10 to 100 times efficiency improvements, while the sharp increase in data center power consumption (the International Energy Agency predicts that it will reach 950 terawatt hours in 2030) will make energy a key bottleneck for computing power expansion. He believes that the competitive boundary of the AI ​​industry is extending from models and GPUs to lithography, storage, power supply and even the entire energy system.