Brief

Apple is emphasizing privacy with its new AI features, dubbed Apple Intelligence, which integrate across its devices. While aiming for on-device processing, queries requiring more power will use a secure Private Cloud Compute system. This system, however, now relies on Google Cloud infrastructure and Nvidia GPUs, a departure from its initial in-house Apple silicon focus. Apple asserts that its data handling remains private and secure, even with third-party hardware involvement, differentiating itself from competitors by prioritizing user privacy. AI

SpaceX has unveiled its Gigasat factory in Texas, a massive 11-million-square-foot facility dedicated to producing AI satellites for orbital data centers. The company, led by Elon Musk, aims to achieve an annual production rate of 1 gigawatt (GW) of space-based AI compute capacity by the end of 2027, with ambitious plans to scale to 100 GW per year by 2030. This initiative involves vertically integrating the manufacturing of satellite components, including solar arrays and compute payloads, to support its goal of providing vast AI computing power from space. AI

Elon Musk is planning a massive orbital constellation of one million satellites to support AI, with the first 150 kW units potentially launching in 2027, though experts raise concerns about radiation and cost. Meanwhile, Google is rolling out Gemini 3.5 Live Translate, enabling seamless audio translation in over 70 languages without manual setting changes. Additionally, a new analytics platform called Mora has emerged from its closed phase, aiming to simplify data integration and reporting by combining natural language processing with SQL code access. AI

CATL, a major battery manufacturer, has invested in DeepSeek's initial funding round, signaling its strategic expansion into the artificial intelligence sector. Chairman Zeng Yuqun is directing approximately $1 billion towards AI data center energy infrastructure, including power transmission and data center operators. This move strategically positions CATL at the nexus of energy storage and AI infrastructure. AI

Amazon employees are urging Seattle to implement a one-year moratorium on new data centers, citing concerns about the environmental and resource costs associated with the rapid expansion of AI development. These employees, part of the "Amazon Employees for Climate Justice" group, testified before the Seattle City Council, highlighting issues like water consumption and electricity prices. They advocate for stricter standards and responsible energy sourcing for data centers, arguing that the pursuit of AI should not come at the expense of environmental goals. AI

Xiaomi's MiMo team, in collaboration with TileRT, has released MiMo-V2.5-Pro-UltraSpeed, a 1-trillion-parameter AI model capable of generating over 1000 tokens per second on standard GPUs. This significant speedup is achieved through a combination of FP4 quantization, DFlash speculative decoding, and the TileRT serving system, an approach they term extreme model-system codesign. The model's enhanced speed is particularly beneficial for latency-sensitive applications like coding agents and real-time decision-making systems. AI

A new research paper explores using rainfall time series and functional regression for predicting regional landslides, with potential applications in early warning systems. Separately, an analysis of the AI boom highlights massive expenditures on infrastructure like datacenters, which are significantly boosting GDP. While AI adoption is accelerating, the article raises concerns about the sustainability of this spending and the increasing cost of using AI models. AI

Xiaomi has released MiMo-V2.5-Pro-FP4-DFlash, a new model optimized for efficient inference. It features expert-only FP4 quantization to reduce memory footprint and bandwidth pressure while maintaining quality. The model also incorporates a BF16 DFlash drafter for speculative decoding, enabling faster token generation by proposing blocks of tokens per forward pass. AI

An AI agent startup has switched its primary model provider from Anthropic to DeepSeek, citing significant cost savings on inference. The company claims this move addresses the growing challenge of sustainable AI deployment, which is increasingly hampered by high inference expenses. This decision highlights a trend where cost-effectiveness is becoming a major factor in selecting AI infrastructure. AI

ChatGPT is reportedly set for its most significant upgrade, with rumors suggesting a major overhaul beyond simple chat capabilities. Separately, Alphabet's Google has placed a substantial order for Intel's TPUs, indicating a large-scale investment in AI hardware. Additionally, AMD has announced plans to invest up to £2 billion in the UK over the next five years to bolster AI innovation and infrastructure. AI

Chinese startup Prinano has announced a breakthrough in photonic chip production, utilizing a novel nanoimprint lithography (NIL) process. This method reportedly bypasses the need for expensive deep-ultraviolet (DUV) lithography equipment, cutting manufacturing costs by approximately 90%. The company has successfully produced 8-inch photonic chip wafers, a significant development that could help circumvent US export restrictions on advanced semiconductor technology. AI

The demand for data center CPUs has significantly increased, driven by the rise of AI agents. While GPUs have been the primary focus for AI workloads, CPUs are now recognized as crucial for handling the "everything else," including operating systems, workload scheduling, and continuous, sustained operations required by agentic AI. This shift is altering the traditional CPU-to-GPU ratio in data center infrastructure, with AMD reporting a doubling of its CPU market growth forecast to 35% annually, projecting a $120 billion market by the end of the decade. AI

A significant portion of new AI data centers in the U.S. are being constructed in regions experiencing drought conditions. Approximately two-thirds of the 809 planned data centers are located in areas that have faced water shortages over the past year. While data center cooling accounts for a smaller percentage of AI's water demand, the fabrication of chips and the power generation required to run them contribute substantially to the overall water footprint, particularly in water-scarce locations. AI

Google has reportedly placed a significant order for at least three million chips from Intel, with deliveries expected in 2028. This move comes as TSMC faces challenges in meeting the high demand driven by the AI boom. The order suggests a potential payoff for Intel's foundry services amidst intense competition in the semiconductor market. AI

AMD announced a significant investment of up to £2 billion over the next five years in the UK's artificial intelligence sector. This funding aims to boost innovation, research, and computing infrastructure within the UK. As part of this initiative, AMD is also establishing a strategic partnership with Imperial College London to advance cutting-edge research in various scientific and engineering fields. AI

A First Nation in Alberta is taking legal action against the proposed $70 billion "Wonder Valley" AI data center project. This massive industrial park, backed by investor Kevin O'Leary, aims to be the world's largest of its kind. The First Nation's opposition centers on environmental concerns related to the project's scale and impact. AI

The rapid expansion of AI is driving an unprecedented surge in data center power consumption across the United States. A Business Insider analysis indicates that data centers permitted through 2025 could collectively use between 224.3 and 358.8 terawatt-hours of electricity annually, a 50% increase year-over-year. This demand is largely fueled by hyperscale facilities and the immense computing needs of tech giants pursuing AI ambitions. The number of new data center permits issued in 2025 reached a record high, with many of these massive facilities planned for rural areas. AI

SpaceX and Google have finalized a significant cloud computing deal where Google will pay SpaceX $920 million monthly for access to approximately 110,000 NVIDIA GPUs and other computing resources. This agreement, set to run from October 2026 to June 2029, is intended to bridge Google's surging demand for its Gemini Enterprise AI platform. The deal highlights the intense scarcity of AI infrastructure and the growing interdependence of major tech companies in securing necessary computing power, occurring just before SpaceX's anticipated IPO. AI

NVIDIA is expanding its AI infrastructure and agentic AI capabilities through strategic partnerships and new product releases. The company is collaborating with the UK government and various partners to build sovereign AI deployments, including the powerful Isambard-AI supercomputer. In South Korea, NVIDIA is working with LG Group to develop AI factories for robotics and autonomous driving, while also partnering with Doosan Group on similar initiatives. Additionally, NVIDIA is enhancing local AI agent deployment on Windows PCs with new hardware like RTX Spark and DGX Station, and integrating its NemoClaw framework across its Jetson platform for edge AI applications. AI

Google has developed an algorithm called TurboQuant that significantly reduces the memory requirements for large language models. This innovation can decrease memory needs by up to six times. The development has reportedly impacted memory chip manufacturers like Samsung, SK Hynix, and Micron, causing their stock prices to fall. AI

Nvidia's CEO Jensen Huang has highlighted a new trillion-dollar growth opportunity in AI chips, sparking discussions about the company's future valuation and market position. Several reports predict that specific AI semiconductor stocks may outperform Nvidia in the coming years. Meanwhile, companies like LG Group are significantly increasing their adoption of Nvidia GPUs, with LG planning to use 10,000 units, and ASUS is integrating Nvidia's AI Factory Platform to accelerate revenue generation. AI

ChatGPT is reportedly set for its largest upgrade ever, potentially integrating Codex capabilities to create a more powerful AI agent. Separately, Intel is collaborating with Hitachi to enhance manufacturing efficiency using AI, while also exploring an AI technology center in South Korea with Nvidia and Hyundai. Additionally, several AI companies, including Zhipu, Juepai Xingchen, and Alibaba, are investing in the embodied AI firm Yuanli Lingji. AI

Nvidia's reliance on Asian supply chains for components has increased to 90% of its production costs, impacting newer products like the Jetson Thor platform and automotive SoCs. This dependency strains wafer capacity and memory supply, even as the company commits to U.S. manufacturing. Meanwhile, the broader AI market faces scrutiny, with concerns about a potential bubble and the financial health of older startups, while some AI stocks are outperforming Nvidia and others are experiencing dips. AI

Nvidia is reportedly acquiring assets from AI chip startup Groq for approximately $20 billion, marking its largest deal to date. This acquisition aims to integrate Groq's low-latency inference technology into Nvidia's AI factory architecture. While Nvidia is licensing Groq's intellectual property and hiring key personnel, Groq will continue to operate as an independent company, with its cloud business unaffected. AI

Multiple open-source projects and platforms are emerging to standardize AI agent interactions through the Model Context Protocol (MCP). These initiatives aim to enable AI agents to access real-time data, external tools, and complex workflows via a unified interface. Key developments include command-line clients for MCP, frameworks for representing agents as MCP servers, and cloud-hosted solutions for integrating various data sources and services. AI

OpenAI and Google DeepMind are advancing AI agents for software development and security. OpenAI's Codex is being leveraged to write entire codebases with minimal human intervention, as demonstrated by Harness Engineering's internal beta product. Google DeepMind has introduced CodeMender, an AI agent designed to automatically identify and fix software vulnerabilities, and AlphaEvolve, which uses Gemini models to discover and optimize algorithms for applications like data center efficiency and chip design. Meta is also investing heavily in its own AI infrastructure with the development of its MTIA chip family, aiming to power AI experiences for billions of users. AI

OpenAI has announced a significant partnership with SAP to launch 'OpenAI for Germany,' aiming to bring advanced AI capabilities to the German public sector while prioritizing data sovereignty and security on Microsoft Azure. The company also proposed policy recommendations to the U.S. White House for the national AI Action Plan, focusing on innovation freedom, export controls, copyright, infrastructure, and government adoption. Additionally, OpenAI is collaborating with U.S. National Laboratories to leverage its reasoning models for scientific breakthroughs and national security initiatives. AI