GTC 2026: the AI race accelerates, and Europe watchesNVIDIA's annual GPU Technology Conference in San Jose gathered tens of thousands of developers, investors and AI leaders in March 2026. Three observations stood out.©360Crossmedia/BJCapital floods into AI infrastructure
The acceleration of investment in AI infrastructure defies precedent. Nscale, the UK-based AI hyperscaler, raised $155 million in December 2024, followed by $1.1 billion in September 2025, then another $433 million just one month later. Data Center Dynamics By early 2026, the company closed a $2 billion Series C, described as the largest in European history. Nscale Globally, investment managers now seek $474 billion for AI data centres, secondaries liquidity and energy mandates The New York Report, a figure that illustrates how capital markets treat compute capacity as the new critical commodity. During his GTC keynote, Jensen Huang disclosed that NVIDIA doubled its demand forecast within the past year, with purchase orders between Blackwell and Vera Rubin expected to reach $1 trillion through 2027. Fortune The sheer velocity of these funding rounds raises a fundamental question for smaller economies: how do you remain relevant when single companies mobilise more capital in two years than some nations allocate to their entire digital strategy?Silicon Valley's meritocratic melting pot
Walking the GTC show floor reveals a striking reality about the global talent pool fuelling artificial intelligence. Attendees from Asia, India, China and dozens of other countries mingled with American engineers, each delivering micro-pitches to potential investors between sessions. WellWells NVIDIA itself drew over 450 sponsors, hosted 1,000 sessions with 2,000 speakers CNBC, turning downtown San Jose into a concentrated hub of international ambition. Chinese daily token consumption has surpassed 140 trillion, more than a thousandfold increase from the 100 billion daily tokens of early 2024. Substack Silicon Valley operates on a straightforward principle: the most capable rise to the top, regardless of origin. For a country like Luxembourg, which thrives on attracting international talent and capital, the lesson resonates. Competing in the AI era demands not only infrastructure but an ecosystem that rewards execution and welcomes global minds.©360Crossmedia/BJThe end of cheap tokens
At GTC, Jensen Huang proposed a novel compensation model: engineers would receive a token budget worth roughly half their base salary on top of regular pay. CNBC Speaking on the All-In Podcast on the final day of GTC, Huang stated that a $500,000 engineer should consume at least $250,000 worth of tokens annually. R&D World VC Tomasz Tunguz noted that startups already add inference costs as a fourth component to engineering compensation, with top-quartile packages reaching $475,000 when token budgets are included. TechCrunch The message from providers aligns: in 2026, cost per token has become a core component of infrastructure design as organisations scale from pilots to production. Silicondata After years of selling access below cost to capture market share, AI providers now face pressure to reach profitability. Token prices will rise, and every company depending on AI must plan accordingly.©Duke26
