Nvidia’s Triple Play: Record Revenue, a $150 Billion Taiwan Campus, and a 20-Year Software Overhaul

Nvidia retires its classic graphics control panel, pledges $150 billion annual investment in Taiwan, announces $80B buyback, and prepares Vera Rubin architecture amid surging AI demand.

After two decades, Nvidia is quietly turning off the lights on its classic control panel. With driver update 610.47 WHQL, released on 26 May, all configuration tasks — from 3D settings to overclocking — now live inside the central Nvidia app. The old interface will linger in the Microsoft Store but receives no more fixes or new features. For a company that made its name selling graphics cards, it is a symbolic retreat from the past into a future dominated by AI infrastructure.

That future is being built in Taiwan. At a groundbreaking ceremony this week, CEO Jensen Huang committed to spending roughly $150 billion annually in the country — up from just $10-15 billion four or five years ago. The pledge came as Nvidia broke ground on “Nvidia Constellation,” its first overseas headquarters, in Taipei’s Beitou-Shilin Science Park. The campus, built under a 50-year lease, covers about 700,000 square metres and carries construction costs exceeding 40 billion New Taiwan dollars. Groundwork begins in June or July, with full operations targeted for 2030. Huang wants to more than triple Nvidia’s local workforce to 4,000 employees, deepening ties with TSMC, Foxconn, Wistron and Quanta Computer. Industry data suggests Nvidia is poised to overtake Apple as TSMC’s largest customer by order volume.

Alongside the Taiwan expansion, Nvidia is simultaneously pouring cash into its own shares. Having generated $48.6 billion in free cash flow last quarter, the board authorised an $80 billion share buyback programme and raised the quarterly dividend from $0.01 to $0.25 per share. The financial firepower comes from a record first quarter: revenue of $81.6 billion, up 85% year over year, with the data centre segment alone contributing $75.2 billion. Management sees that momentum continuing, guiding for $91 billion in sales during the second fiscal quarter of 2027.

The scale of the numbers is reflected on Wall Street, where analysts are scrambling to update their models. Rothschild & Co Redburn lifted its price target to $300, while Tigress Financial went even further to $425, citing “parabolic” demand for AI infrastructure. Rosenblatt, which had the highest target among 62 analysts tracked by the primary source, set its target at $325. The stock currently trades around $213, roughly 10% below its all-time high of $236.54, with recent bond yields and sector rotation weighing on the share price. The consensus remains a strong buy, though the annualised volatility remains high.

On the product front, Nvidia is preparing to ship its next architecture, code-named Vera Rubin, starting in the third quarter of fiscal 2027. The hyperscalers – Alphabet, Microsoft, Meta and Amazon – have collectively raised their combined AI spending this year to over $700 billion, up from a previous estimate of $600 billion. That voracious appetite for Blackwell and Rubin platforms currently outstrips supply, even as export controls cap Nvidia’s ability to serve the Chinese market.

Huang’s dual announcement of a gargantuan Taiwan campus and a record buyback underscores a company that is both investing aggressively in its manufacturing ecosystem and rewarding shareholders handsomely. The $150 billion annual outlay in Taiwan alone dwarfs the $500 billion four-year AI infrastructure plan flagged for the United States. Between the software legacy being retired and the hardware ambitions being built, Nvidia is scripting a new chapter that leaves its graphics-card origins firmly in the rearview mirror.