AI news, 20 July: Nationwide protests hit AI data centers
US protests against AI data centers go national. Opponents staged 142 rallies across 42 states on Saturday in the first coordinated nationwide pushback against the rapid buildout of facilities powering artificial intelligence, according to Reuters. Organized by the group HumansFirst, demonstrators voiced concerns over water use, energy strain, local disruption and lack of community benefits. A recent Reuters/Ipsos poll found only 14% of Americans would support an AI data center in their own community, underscoring how the issue cuts across political lines. Industry group the Data Center Coalition said operators work with communities to lessen impacts, while organizers framed the actions as a nonpartisan call for transparency and accountability ahead of midterm elections.
TSMC sees multi-year AI chip demand, expands Arizona plans. Taiwan Semiconductor Manufacturing Company, the world’s leading maker of advanced chips used in AI systems and a key Nvidia supplier, reported strong, structural multi-year customer demand, Reuters reported. After record quarterly results, CFO Wendell Huang said the firm is ramping its Arizona investment to a total of about $265 billion, calling progress there satisfactory even as it faces shortages of construction workers and infrastructure limits. The first Arizona plant is achieving yields matching Taiwan sites; additional fabs and packaging facilities are underway. The move aligns with US efforts to boost domestic chipmaking, though geopolitical export controls remain a challenge.
China’s Moonshot unveils massive open Kimi K3 model. Beijing-based startup Moonshot AI launched Kimi K3, a 2.8-trillion-parameter open-weight model it describes as the world’s largest of its kind and competitive with leading US systems, according to Reuters and BBC reporting. The company says the model, with a 1-million-token context window, excels at long coding tasks, reasoning and knowledge work; some independent leaderboards have ranked it highly in coding and multi-step benchmarks. Open-weight release means others can download and adapt it, though its sheer size limits who can run it locally. The debut highlights how Chinese firms are rapidly narrowing performance and cost gaps with American frontier models.
Apple briefly tops Nvidia as world’s most valuable company. Apple overtook Nvidia to reclaim the title of world’s most valuable public company, with valuations around $4.88 trillion versus roughly $4.86 trillion, Reuters reported on 17 July. The shift came as investors reassess the sustainability and concentration of AI-related bets that had propelled the chipmaker. Nvidia shares slipped while Apple held steadier; analysts noted broader re-evaluation of hyperscaler spending and AI infrastructure returns after a prolonged boom.
UK finance chiefs grow more optimistic about AI. A Deloitte survey of chief financial officers at major British companies found 73% now optimistic that AI will improve business performance, up from 59% late last year and 39% two years ago, according to Reuters. Geopolitical risk concerns eased somewhat in the latest readings, though worries about UK productivity and competitiveness lingered. The results, drawn from responses in early July, suggest corporate leaders are moving from caution toward greater confidence in practical AI gains even amid global uncertainties.