Industrial Policy & Trade Monitor· 2667 actions· ✨ 89 recent

Taiwan's Legislative Yuan passed the Artificial Intelligence Basic Act (人工智慧基本法) on third reading on 23 December 2025, and President Lai Ching-te promulgated the 20-article statute on 14 January 2026, bringing it into force immediately. The Act designates the National Science and Technology Council (NSTC) as the central AI-policy competent authority and codifies seven governance principles — sustainability and well-being, human autonomy, privacy protection and data governance, cybersecurity and safety, transparency and explainability, fairness and non-discrimination, and accountability — that apply to all public-sector AI procurement and high-risk sectoral applications. The statute establishes a statutory foundation for the Taiwan AI Action Plan 2.0, mandates an Executive Yuan National AI Strategy Committee, and provides authority for sector-specific implementing regulations by FSC, NCC, MOHW, and MOTC across finance, telecoms, medical, and autonomous-vehicle AI within a two-year window. As the first national AI governance statute in the Greater China region and the third globally after the EU AI Act and South Korea's AI Basic Act, it frames regulatory expectations for the companies at the heart of the global AI hardware supply chain — TSMC, NVIDIA ODM partners, and advanced-packaging incumbents — that are headquartered or operate substantially in Taiwan.

On 10 December 2025 the National Assembly of Vietnam adopted Law No. 134/2025/QH15 on Artificial Intelligence (8 chapters, 35 articles), Vietnam's first dedicated AI statutory framework and one of the first comprehensive horizontal AI laws in Southeast Asia. The law establishes a three-tier risk-based regulatory architecture (high / medium / low) for the research, development, provision, deployment, and use of AI systems; defines the rights and obligations of providers, deployers, importers, distributors, and users; and mandates state oversight via the Ministry of Information & Communications and Ministry of Science & Technology. Prohibited acts include systematic deception, manipulation of human perception, generation of fake content endangering national security, exploitation of vulnerable populations, and obstruction of human-supervision mechanisms. The law applies to Vietnamese agencies, organizations, and individuals as well as foreign organizations and individuals involved in AI-related activities in Vietnam, taking effect 1 March 2026 with 12-18 month transition windows for existing systems depending on sector.

On 20 November 2025 the White House and US Department of Commerce / Bureau of Industry and Security (BIS) authorised Abu Dhabi AI holding company G42 to import advanced computing chips — equivalent to approximately 35,000 Nvidia GB300 Blackwell processors — under the UAE-pioneered Regulated Technology Environment (RTE) compliance framework. The RTE is an Emirati-designed technology governance and audit architecture, developed by G42 and approved under BIS guidelines, with binding UAE-side controls to prevent onward diversion to foreign adversary nations. The authorisation accelerates the Stargate UAE project — a 1 GW AI compute cluster being built by G42 for OpenAI in partnership with Oracle, Cisco, NVIDIA, and SoftBank Group — and represents the first concluded major country-level advanced-compute authorisation following the May 2025 rescission of the Biden-era AI Diffusion Rule.

Italy enacted Legge 23 settembre 2025, n. 132 — "Disposizioni e deleghe al Governo in materia di intelligenza artificiale" — published in Gazzetta Ufficiale Serie Generale n. 223 of 25 September 2025 (atto 25G00143) and entered into force on 10 October 2025. The statute makes Italy the first EU member state to enact a comprehensive national AI law complementing Regulation (EU) 2024/1689 (EU AI Act), designating AgID (Agency for Digital Italy) and ACN (National Cybersecurity Agency) as national oversight authorities under Presidency-of-the-Council-of-Ministers coordination. It sets sectoral rules for healthcare, labour, intellectual professions, public administration and the judiciary; authorises up to €1 billion in state-backed venture capital (via CDP Venture Capital) for AI, cybersecurity and telecoms; creates criminal penalties of up to five years' imprisonment for harmful deepfakes; mandates parental consent for under-14 users; and delegates secondary legislation to the Government across multiple domains.

On 8 September 2025, the UK Ministry of Defence published the Defence Industrial Strategy 2025 — "Making Defence an Engine for Growth" (CP 1388) — the first comprehensive cabinet-level UK defence industrial strategy in over a decade and the sector plan for Defence under the UK Modern Industrial Strategy umbrella (IS-8). The strategy was published alongside the Strategic Defence Review 2025 and operationalises the largest sustained defence- spending increase since the Cold War (rising to 2.6% of GDP by 2027 with ambition to 3% in the next Parliament). It defines six priority outcomes (growth, backing UK businesses, defence innovation, resilient industrial base, procurement transformation, enduring partnerships); establishes UK Defence Innovation (UKDI) within the MOD with a ringfenced £400m budget to accelerate dual-use technology; identifies priority defence capabilities (combat air, complex weapons, directed-energy weapons, next- generation land and maritime systems) plus dual-use sub-sectors (quantum, drones/autonomy, space, AI, cyber, engineering biology, advanced connectivity); and flags resilience priorities in steel, construction, energetic materials, batteries, semiconductors and rare earths.

The French state, via the Agence des participations de l'État (APE), signed a contract on 1 August 2025 to acquire Atos's Advanced Computing division — encompassing the High Performance Computing & Quantum and Business Computing & AI units, commercially known as Bull — for an enterprise value of up to EUR 410 million (later adjusted to up to EUR 404 million after scope changes). The division employs over 2,500 people, mostly in France, generates roughly EUR 800 million in expected 2025 revenue, and operates Europe's only supercomputer manufacturing plant (Angers), including delivery of the EuroHPC JUPITER exascale system. The Ministry of Economy and Finance framed the acquisition as keeping "strategic industrial activities and key technologies" in French hands; the deal closed on 31 March 2026 with the French state as Bull's sole shareholder.

On 1 July 2025 India's Union Cabinet approved the Research Development and Innovation (RDI) Scheme, a six-year, ₹1,00,000 crore (≈USD 11.68 bn) fund to catalyse private-sector investment in research, development and innovation, with ₹20,000 crore allocated for FY2025-26. The scheme finances transformative RDI projects (TRL 4 and above) in strategic and sunrise sectors — deep tech, AI, biotechnology, quantum computing, robotics, space, energy transition and the digital economy — through long-tenor, low-or-nil-interest loans and equity, up to 50% of assessed project cost. Grants and short-term loans are explicitly excluded. Funds flow through a Special Purpose Fund under ANRF (first-level) to second-level fund managers — Alternate Investment Funds, Development Finance Institutions, NBFCs, and focused research bodies including the Technology Development Board (TDB) and BIRAC — which began issuing project calls in February 2026.

On 14 June 2025 Vietnam's 15th National Assembly adopted Law No. 71/2025/QH15 on the Digital Technology Industry (DTI Law) at its 9th session. The Law enters into force on 1 January 2026 (with certain provisions phased) and is the world's first standalone primary statute dedicated to the digital technology industry, covering digital-tech production and services, semiconductor manufacturing, artificial-intelligence systems, digital assets (legally recognised as property under the Civil Code), and Concentrated Digital Technology Zones. It codifies sector-specific incentives — multi-year corporate income tax reductions, R&D-cost deductions, preferential public procurement, five-year personal income tax exemption for high-quality digital professionals, five-year visa and work-permit exemptions for foreign experts, and 50% subsidy for SME advanced-technology acquisition — and sets headline targets of 150,000 digital-tech enterprises and USD 74bn digital-economy contribution by 2030/2035 (with USD 43bn / USD 74bn variants in different government summaries).

On 9 April 2025 the European Commission adopted Communication COM(2025)165, the AI Continent Action Plan, setting out a five-pillar strategy to make the EU a global AI leader. The pillars are (1) computing infrastructure, (2) data for AI, (3) strategic AI innovation and adoption, (4) AI skills and talent, and (5) regulatory simplification. Headline commitments include mobilising approximately €200bn of public+private investment via the InvestAI initiative announced at the AI Action Summit in Paris (11 February 2025), deploying 13 AI Factories (HPC-anchored shared compute facilities) plus regional antennas, building 5 AI Gigafactories powered by >100,000 advanced AI processors with €20bn earmarked from InvestAI, launching the Apply AI Strategy and Data Union Strategy, and proposing a Cloud and AI Development Act with a public consultation closing 4 June 2025. The one-year progress report (9 April 2026) confirmed 19 AI Factories deployed across EU supercomputers with 13 Antennas providing regional access, and €1bn in Apply AI funding calls earmarked.

The US Bureau of Industry and Security amended the Export Administration Regulations to add 12 entities to the Entity List under the destinations of China (11) and Taiwan (1) via Final Rule 2025-05427 (90 FR 14046), companion to the larger 70-entity rule (2025-05426) published the same day. Targets fall in three clusters: (i) Beijing Academy of Artificial Intelligence and Beijing Innovation Wisdom Technology — added for acquiring US-origin items in support of China's military modernization, specifically developing large AI models and advanced computing chips for defense; (ii) the Inspur group — Inspur (Beijing) Electronic Information Industry, Inspur Electronic Information Industry, Inspur Electronic Information (Hong Kong), Inspur (HK) Electronics, Inspur Software, and Inspur Taiwan — added as subsidiaries contributing to supercomputers for military end use; and (iii) Henan Dingxin, Nettrix Information Industry, Suma Technology, and Suma-USI Electronics — added for involvement in the development of Chinese exascale supercomputers. License requirements are for all items subject to the EAR; review policy is presumption of denial for the AI cluster and policy of denial for the supercomputer clusters.

President Trump signed Executive Order 14179 on 23 January 2025 (published in the Federal Register on 31 January 2025 as 90 FR 8741, doc 2025-02172). The order revokes Biden-era Executive Order 14110 of 30 October 2023 ("Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence") and directs federal agencies to identify and rescind, revise, or suspend any policies, regulations, memoranda, or guidance documents adopted pursuant to the revoked Biden order. It mandates that the Assistant to the President for Science and Technology, the Assistant to the President for National Security Affairs, the Special Advisor for AI and Crypto, and the Assistant to the President for Economic Policy develop an AI Action Plan within 180 days to "sustain and enhance America's global AI dominance." The plan was released on 23 July 2025. EO 14179 reframes US AI industrial-policy posture from safety-first regulation to deregulation, infrastructure investment, and export-competitiveness.