🔥 Today’s Top AI Stories
Published today, sourced from stories within the last 24 hours
🇨🇳 DeepSeek Launches V4 — Optimized for Huawei Chips
Chinese AI startup DeepSeek released preview versions of its next-generation DeepSeek V4 model on Friday, marking a significant shift in the US-China AI race. The V4-Pro packs 1.6 trillion parameters with an ultra-long context of one million words, while the cheaper V4-Flash has 284 billion parameters.
What makes this release truly significant: V4 is the first DeepSeek model optimized for domestic Huawei Ascend AI chips, rather than relying solely on Nvidia hardware. Huawei announced “full support” from its Ascend chip lineup for V4 inference. The move is being seen as a test of whether China’s homegrown AI ecosystem can loosen its dependence on US chip giant Nvidia.
DeepSeek claims V4-Pro “significantly leads other open-source models” and is “only slightly outperformed by the top-tier closed-source model, Google’s Gemini-Pro-3.1.” The company also promises rock-bottom inference pricing, continuing its reputation for strong cost-performance.
Shares in Semiconductor Manufacturing International Corp (SMIC), which makes Huawei’s Ascend processors, jumped 10% in Hong Kong trading on the news.
Context: The White House on Thursday alleged Chinese entities were engaging in “industrial-scale distillation campaigns to steal American AI.” Beijing rejected the allegations.
🤝 Cohere Merges with Aleph Alpha — $20B Transatlantic AI Powerhouse
Canadian AI firm Cohere announced Friday it is acquiring Germany’s Aleph Alpha, creating a transatlantic “sovereign AI” powerhouse valued at roughly $20 billion. The deal, backed by both the Canadian and German governments, aims to build an alternative to US hyperscalers for European and international customers who want greater control over their AI infrastructure.
As part of the deal, the Schwarz Group (a major Aleph Alpha investor) committed $600 million in additional investment. The merged entity targets business and government customers in highly regulated markets who are uncomfortable relying on US-based tech giants.
The move reflects growing global demand for “sovereign AI” — AI systems that governments and enterprises can control within their own jurisdictions — and could reshape the competitive landscape outside Silicon Valley.
🚀 SpaceX Secures Option to Acquire Cursor for $60 Billion
Elon Musk’s SpaceX announced a major AI partnership with Cursor, the AI code-generation startup co-founded by Sualeh Asif. The deal comes with the right to acquire Cursor for $60 billion later this year — or walk away and pay roughly $10 billion for the collaboration.
The move signals SpaceX’s aggressive push into AI capabilities, though it raises questions about the financial sustainability of Musk’s AI spending. According to Reuters, “AI is burning the cash that Starlink earns.” Bloomberg notes that Microsoft had also considered acquiring Cursor but chose not to make a formal bid.
Reports also indicate SpaceX/Cursor have explored a three-way partnership with French AI startup Mistral to better compete against Anthropic and OpenAI.
☁️ Google Unveils Gemini Enterprise Agent Platform
Google Cloud rolled out the new Gemini Enterprise Agent Platform, a unified platform for building, scaling, governing, and optimizing enterprise-grade AI agents. Google reports that nearly 75% of Google Cloud customers now use its AI products, and CEO Sundar Pichai is targeting $175B–$185B in capital expenditure to fuel the buildout.
The platform offers a no-code interface where employees describe what they want automated in plain language, and the system builds the agent. Google’s positioning: rivals are “handing you the pieces, not the platform” — Google’s bet is a unified stack from chip to inbox.
The platform integrates with Google Drive, Microsoft OneDrive, SharePoint, HubSpot, Jira, and more, and includes an Agent Marketplace for partner-built agents.
🇬🇧 OpenAI Pauses UK Data Centre Project
OpenAI is pausing its main data centre project in Britain, citing an unfavourable regulatory environment and high energy costs. The decision deals a blow to the UK government’s push to position the country as a global AI hub.
The move comes as the UK quietly increased its AI emissions forecast 100-fold, with new estimates suggesting the planned AI buildout could produce 34m–123m tonnes of CO₂ between 2025 and 2035. Meanwhile, OpenAI is also pulling back from the Stargate Norway data center deal, with Microsoft reportedly taking over that project.
Anthropic is reportedly planning its own European expansion, eyeing the $600B+ AI infrastructure budgets of American technology companies — but faces the same energy and regulatory headwinds that caused OpenAI to hit pause.
💼 Meta Cuts 10% of Workforce, Microsoft Offers Buyouts Amid AI Spending Surge
Two of tech’s biggest companies are restructuring to fund their massive AI investments:
- Meta plans to cut about 10% of its workforce (~8,000 employees) starting May 20, plus leave ~6,000 open roles unfilled, according to an internal memo dated April 23.
- Microsoft announced its first-ever voluntary buyout programme for US employees, potentially affecting up to 9,000 workers (≈7%). The company has already reduced headcount in targeted engineering groups this year.
Combined, the cuts and buyouts could affect up to 23,000 jobs. Both companies are pouring billions into AI infrastructure while streamlining operations. Microsoft’s buyout comes just ahead of earnings as investors scrutinize its elevated AI capex trajectory.
Meanwhile: Intel stock soared 25% after reporting earnings with raised AI-related revenue guidance, and SAP beat cloud-growth estimates on AI-agent integration — signs that the AI infrastructure demand remains voracious.
⚖️ White House Accuses China of “Industrial-Scale” AI Theft
The White House on Thursday alleged that Chinese entities are engaging in “industrial-scale distillation campaigns to steal American AI,” escalating the rhetoric in the US-China AI rivalry. Beijing rejected what it called “baseless allegations,” adding that China “attaches great importance to the protection of intellectual property rights.”
The accusations come on the same day DeepSeek released its V4 model, and add a geopolitical dimension to what is already an intense technology competition.
🏛️ Idaho Signs Conversational AI Safety Act
Idaho Governor Little signed SB 1297, the Conversational AI Safety Act, on April 2, set to take effect July 1, 2026. The bill adds to a growing patchwork of state-level AI regulations as federal legislation continues to stall. The act establishes safety requirements for conversational AI systems — part of a broader wave of AI governance legislation advancing across US states.
📊 Quick Hits
- OpenAI GPT-5.5 was announced earlier this week, positioning it as a step toward a unified AI “super app” combining ChatGPT, coding tools, and browser capabilities. Better at coding, computer use, and deeper research.
- OpenAI shared Cyber AI model info with the US government and allies, per Axios.
- Oracle’s $300B OpenAI data-center deal is reportedly running into financing challenges, with banks struggling to spread the loans.
- UK AI emissions could reach 34m–123m tonnes of CO₂ by 2035, officials now admit — a 100x increase from earlier estimates.
Published April 25, 2026. All stories sourced from breaking news within the last 24 hours. Links to original sources available from the respective publishers.
Leave a Reply