Clouds, GPU clusters, inference stacks and orchestration — the picks-and-shovels layer between chips and applications.
308 articles
Microsoft has established a new dedicated AI deployment group backed by a $2.5 billion commitment. This move aligns the company with other major AI players like Amazon, OpenAI, and Anthropic in scaling AI infrastructure and implementation.
TechCrunch AI · 79d ago
SK Hynix plans to build a massive new NAND flash memory factory, M17, in Cheongju, South Korea. The facility is designed to meet the surging demand for high-capacity memory required by AI applications, with production slated for 2029.
The Next Web (TNW) · 79d ago
The New York Times has escalated its legal stance against Microsoft, alleging that the company built a supercomputer specifically to facilitate copyright-infringing AI training for OpenAI. This move follows recent legal shifts regarding copyright protections in the wake of Supreme Court rulings.
Ars Technica · 85d ago
Apple has officially raised prices on iPads, Macs, and smart speakers. The company cites the surging costs of memory and storage driven by AI companies as a primary factor.
The Verge · 86d ago
Cloud startup Runpod has raised $100M, reaching a $1 billion valuation in under two years. The company provides GPU rental services, capitalizing on the massive demand for AI computing power.
The Next Web (TNW) · 87d ago
As AI workloads shift toward persistent agentic systems, context management has become a critical bottleneck exceeding GPU compute limits. Industry leaders are proposing a dedicated 'context tier' of high-performance flash storage to handle KV cache and inference state.
VentureBeat · 89d ago
Databricks introduced Lakehouse//RT and LTAP to unify transactional and analytical data storage, removing the need for complex ETL pipelines. This infrastructure update aims to provide AI agents with real-time access to live data without performance degradation.
VentureBeat · 95d ago
NVIDIA's Blackwell platform achieved a clean sweep in the latest MLPerf Training v6.0 benchmarks. It demonstrated industry-leading scale and performance, securing the fastest training times and highest per-accelerator efficiency across all tests.
NVIDIA Developer Blog · 95d ago
China's official average daily token consumption reached 500 quadrillion by the end of June, but the article warns that comparisons with the US remain uncertain because providers use incompatible scopes and may double-count or omit workloads. Token volume also does not directly translate into AI revenue, as model quality, latency, workload value, and infrastructure economics matter.
虎嗅 · 11d ago
Alibaba Cloud, Cambricon and Ant Group joined the PyTorch Foundation, with Alibaba Cloud and Cambricon becoming Platinum members. The companies gathered with Huawei at PyTorch Conference China in Shanghai to advance the open-source AI stack.
PyTorch Blog · 11d ago
Zhongke Naono has completed a B+ financing round led by a centrally administered state-owned enterprise. The company plans to use the opportunity to develop a “Token Factory,” positioning the financing within a broader shift in AI infrastructure and production capabilities.
钛媒体 · 12d ago
Prime Minister Mark Carney’s effort to attract investment in Canadian data centers is encountering political and public backlash. The resistance mirrors opposition to data center expansion in the United States and could complicate AI infrastructure growth in Canada.
Bloomberg Technology · 15d ago
Nazhen Technology passed its Hong Kong Stock Exchange listing hearing and is expected to become Hisense’s sixth listed company. Although it holds only 4% of the global optical-module market and remains a second-tier vendor, it is Hisense’s largest and most AI-relevant technology business outside consumer appliances.
虎嗅 · 15d ago
China State Construction reported first-half revenue of 975.8 billion yuan and net profit of 23.0 billion yuan, both down year over year as housing and infrastructure demand weakened. To offset the downturn, the company is shifting toward industrial facilities, data centers, compute centers, and overseas projects, with compute-center contracts up 328.3%.
虎嗅 · 16d ago
Industry executives at a Wuxi conference projected that global new data-center infrastructure investment could exceed $4 trillion by 2028. Semiconductor spending is expected to account for more than half of the total, while China showcased advanced-packaging platforms.
Pandaily · 17d ago
Nvidia is investing $3.5 billion in MediaTek to deepen their chip collaboration. The move highlights Nvidia’s strategy for competing with Big Tech companies developing custom AI silicon.
Bloomberg Technology · 19d ago
MiniMax increased the spending ceiling in its three-year Alibaba Cloud agreement by 220% to $1.2 billion. Annual limits now reportedly reach $300 million in 2026, $400 million in 2027, and $500 million in 2028.
TechNode · 19d ago
China’s four domestic GPU makers are entering a differentiation phase after going public. Their focus is shifting from simply producing chips to building complete commercial ecosystems and customer deployments.
Pandaily · 19d ago
Lambda has reportedly raised about $1 billion in short-dated private debt arranged by JPMorgan to purchase Nvidia GPUs that Microsoft will lease. The financing structure resembles Nebius’s $775 million chip-backed borrowing tied to a five-year, $19.4 billion Microsoft contract.
The Next Web (TNW) · 22d ago
光本位科技完成数十亿元融资，投后估值接近200亿元。公司被称为全球融资规模最大、产品研发和商业化进度最快的光计算企业之一。
钛媒体 · 22d ago
全球招聘服務公司在經歷兩年下跌後，2025年股價大幅反彈，部分公司年內接近翻倍。AI 一方面自動化篩選、面試與排程，另一方面提高了經過真實招聘結果驗證的人才數據價值。
虎嗅 · 12d ago
Yotta Data Services is discussing a public offering very soon to raise capital for rapidly growing AI service demand. The move highlights how AI infrastructure providers are seeking funding to expand capacity.
Bloomberg Technology · 22d ago
The UK energy regulator is using various measures to prevent speculative data center projects from connecting to the power grid. The restrictions could affect the country’s plans to expand its AI infrastructure.
Wired AI · 23d ago
The article argues that major technology companies are racing to control chips, data centers, models, agents, applications, and devices, but this breadth may create coordination and resource-allocation problems. OpenAI's platform-oriented stance and Google's AI reorganization are presented as signs that focused control points may be more sustainable than doing everything in-house.
虎嗅 · 24d ago
The article examines OpenAI’s potential 2027 IPO as a test of technology commercialization and trillion-dollar valuation ambitions. It frames the listing around capital-duration matching and an estimated 600 billion yuan in off-balance-sheet compute liabilities.
钛媒体 · 25d ago
OpenAI CFO Sarah Friar outlines how progress in chips, compute, models, and products compounds to deliver more useful intelligence. The strategy aims to increase scale while reducing costs.
OpenAI News · 25d ago
IBM announced a next-generation mainframe processor whose 11 cores can dynamically execute both IBM Z and Arm instruction sets. The chip is designed to run Arm-native Linux software, including AI frameworks, alongside z/OS workloads on the same system.
VentureBeat · 26d ago
Malaysia is introducing data centre rules intended to make the rapidly expanding sector more sustainable rather than restrict investment. Strong demand continues, with global occupancy at 97% and most capacity under construction already committed.
SCMP Technology · 26d ago
DeepSeek、智譜與月之暗面（Kimi）正從追趕式競爭轉向前沿模型競爭，但更大的參數規模也帶來更高的算力、資本與商業化壓力。文章比較三家公司不同的擴展路徑：Kimi押注預訓練、智譜加碼後訓練，DeepSeek則將基礎設施效率納入模型擴展。
虎嗅 · 27d ago
China Mobile, China Telecom and China Unicom are making billable AI tokens a central growth metric as AI adoption and computing demand accelerate. Their strategies increasingly emphasize AI infrastructure and computing operations as new revenue engines.
SCMP Technology · 29d ago