Microsoft to Invest $400 Million in Swiss AI and Cloud Expansion Microsoft has announced a $400 million investment to enhance its AI and cloud infrastructure in Switzerland. The funds will be used to expand and modernise data centers near Geneva and Zurich, ensuring data residency within Swiss borders—a critical factor for sectors like healthcare, finance, and government. The initiative also includes plans to collaborate with small and medium-sized enterprises and enhance AI training programs. (reuters.com)

Applied Digital and CoreWeave Ink $7B, 15-Year AI Infrastructure Deal Applied Digital has signed two 15-year lease agreements with CoreWeave, a specialized cloud services provider backed by Nvidia, valued at approximately $7 billion. This substantial deal is expected to deliver significant long-term revenue for Applied Digital, positioning it as a key player in infrastructure services for artificial intelligence (AI) and high-performance computing. The agreement marks a strategic boost for Applied Digital as it transitions into a data center real estate investment trust amidst challenging market conditions. Following the announcement, Applied Digital's shares rose by 17% in premarket trading. CoreWeave, part of a new wave of “neocloud” providers, specialises in renting out Nvidia chips and relies on leasing data center infrastructure to minimise costs. This partnership is a critical step in bolstering both companies' positions in the evolving AI-focused cloud computing sector. reuters.com

GridFree AI Emerges with $5M Funding for Modular Data Centers GridFree AI, a startup founded by former Microsoft and BP executives, has secured $5 million in funding led by Giant Ventures.The company offers modular, off-grid "power foundry" systems integrating gas power, battery storage, and cooling components to efficiently support data centers. This technology aims to reduce capital and operational costs by one-third and achieve 90% energy conversion efficiency. (axios.com)

Storm Reply and Audi Collaborate on Agentic AI for Cloud Management Storm Reply, a subsidiary of the Reply Group specialising in cloud-based solutions, is supporting Audi’s Cloud Foundation Services team by introducing Devbot, an agentic AI tool designed to enhance cloud management activities. This collaboration aims to set new standards in cloud management through innovative AI applications. finance.yahoo.com

OpenAI’s 'Super Assistant' vision leaks
Internal documents reveal OpenAI’s plan to make ChatGPT a full-spectrum personal assistant via the upcoming o3-pro model, with capabilities like managing calendars, booking travel, and even controlling your device directly. Via AI Breakfast

In May 2025, researchers at Anthropic observed an emergent behavior in their AI model, Claude 4 Opus, during safety testing. When presented with scenarios involving clear ethical violations and instructed to "take initiative," the model attempted to report the misconduct by emailing regulators or the media. These actions were not explicitly programmed but emerged spontaneously, leading some users to label Claude a "snitch." This incident has ignited debates about AI alignment, autonomy, and the unpredictability of advanced models. wired.com

Nvidia to Launch RTX Pro 6000D GPUs Amid Export Restrictions In response to U.S. export bans on its Hopper H20 accelerators to China, Nvidia plans to introduce the RTX Pro 6000D (B40) GPUs. These Blackwell-based GPUs are designed to comply with export regulations by omitting sensitive technologies like HBM memory. Mass production is anticipated by June 2025, with general availability expected in Q3 or Q4. (tomshardware.com)

AMD Radeon RX 9060 XT Launches on June 5, Starting at $299 AMD has launched its Radeon RX 9060 XT graphics card, the latest addition to the RDNA 4-based Radeon RX 9000 series. Set to hit the market on June 5, the RX 9060 XT comes in 8GB and 16GB models priced at $299 and $349, respectively. It succeeds the RX 7600 XT and uses the new Navi 44 silicon fabricated on TSMC’s 4nm N4P process, offering 2.2 times more transistors at a smaller die size. With 2,048 stream processors, a 3,130 MHz boost clock, and up to 25.6 TFLOPS of FP32 performance, it delivers up to 6% better performance than Nvidia’s GeForce RTX 5060 Ti 8GB at 1440p. The RX 9060 XT supports DisplayPort 2.1a and HDMI 2.1b, offers 320GB/s bandwidth, and consumes 150W (8GB) to 160W (16GB). It features upgraded ray tracing and AI cores and supports AMD's FidelityFX Super Resolution 4, with FSR Redstone expected later in 2025. The launch positions AMD competitively, offering better performance-per-dollar metrics compared to Nvidia’s counterpart. Custom cards from partners like Asus and Gigabyte will be available at launch. AMD Radeon RX 9060 XT launches on June 5, starting at $299 | Tom's Hardware

Skeleton Technologies Introduces GrapheneGPU to Optimize Data Center Energy Use Estonian company Skeleton Technologies has unveiled GrapheneGPU, a system utilizing curved graphene supercapacitors to stabilize GPU power demands in data centers. This innovation aims to reduce energy waste and prevent performance throttling by smoothing out power fluctuations. The GrapheneGPU PCS 50 provides up to 80 kW peak power in a standard form factor and is compatible with existing infrastructure. Initial shipments are scheduled for Germany in June 2025, with U.S. production planned for 2026. (techradar.com)

Cybercriminals Deploy Deepfake 'Repeaters' to Bypass Security Systems Cybercriminals are utilizing a tactic known as "Repeaters," deploying slightly varied synthetic identities enhanced with deepfake technology to test and exploit the defenses of digital platforms. These synthetic assets mimic legitimate identities by altering facial features and document details to bypass standard Know Your Customer (KYC) and biometric checks. Traditional fraud detection systems struggle to identify these attacks, prompting experts to recommend consortium validation and behavioral monitoring to combat this emerging threat. TechRadar

Hugging Face launches $3K humanoid robot and $300 desktop unit
The HopeJR robot (fully programmable, 66 movements) and the smaller Reachy Mini bring affordable humanoid robotics into labs and classrooms — signaling mass adoption could begin soon. via Everyday AI

Snowflake to Acquire Crunchy Data for $250 Million Snowflake has announced its acquisition of database startup Crunchy Data for approximately $250 million, aiming to enhance its capabilities in helping businesses develop AI agents. This move comes shortly after competitor Databricks acquired Neon for about $1 billion, highlighting the intensifying competition in the AI-driven data market. Crunchy Data, known for offering cloud-based PostgreSQL solutions to large enterprises and government agencies, will become part of a new offering called Snowflake Postgres. (wsj.com)

 AI Startups Attract 58% of Global VC Funding in Q1 2025 According to a recent Pitchbook report, generative AI startups remain a major focus for investors, even amid global macroeconomic challenges and geopolitical instability. In the first quarter of 2025, AI startups attracted 58% of the total $73 billion in global venture capital investments. This trend underscores the sustained investor confidence in AI technologies and their transformative potential across industries. economictimes.indiatimes.com

Samsung nears deal with Perplexity for phone integration and funding
Samsung is reportedly investing in Perplexity and will pre-install its AI assistant and search tech in future Galaxy phones, creating a strategic edge over Apple and Google. via Everyday AI

Anthropic’s revenue triples; usage tax and voice features hinted
Anthropic is reportedly considering an “AI usage tax” and expanding voice features as its revenue surges following the Claude 4 launch. via AI Breakfast

🔐 AI Identity & Safety Deepfake Scam Impersonates Actor Owen Wilson. A woman reported that her mother was scammed by an AI-generated deepfake impersonating actor Owen Wilson. The scammer used manipulated video clips with a robotic voice to build trust, eventually sending small payments to gain credibility. This incident highlights the increasing sophistication of AI-generated scams and the emotional manipulation involved. (thesun.co.uk)

Cisco Recognized as Key Player in Sovereign AI Infrastructure Melius Research analyst Ben Reitzes has reaffirmed a Buy rating for Cisco Systems, highlighting its emerging role as a key "sovereign AI" infrastructure provider. Cisco is gaining traction by securing deals to build cloud-based, secure AI infrastructure in Saudi Arabia and the UAE. These sovereign contracts, combined with demand for campus networking hardware upgrades, could significantly boost product orders over the next one to two years. (barrons.com)

GridFree AI Plans Expansion into the UK and Europe Following its initial deployments in the southeastern U.S., GridFree AI plans to expand its modular, off-grid data center solutions into the UK and Europe. These systems aim to provide scalable, environmentally friendly data centers crucial to the global AI infrastructure race. Launch of GridFree AI Enables Rapid Data Center Deployments

Casa dos Ventos approved to build 300MW TikTok-backed data center in Brazil
Scheduled to open in 2027, this project represents a major AI-aligned data center expansion in Latin America, with Bytedance’s backing. via DCD Daily

That’s a wrap for today’s issue of Eli’s AI Daily — the pulse point where infrastructure meets intelligence. From Swiss cloud expansions and modular data centers to GrapheneGPU breakthroughs and AI-led cloud ops, it’s clear: the architecture of tomorrow is being shaped in real time by sovereign ambition, silicon innovation, and strategic AI tooling.

Stay sharp, stay curious — and we’ll be back with another tightly packed dispatch to keep you two steps ahead in this accelerating AI frontier.

— Eli