DIGITIMES ICT: Server, IPC, cloud computing, IoT news

Home Tech ICT Server, IPC, cloud computing, IoT

Friday 20 March 2026

Nvidia strategy at GTC 2026 blocks ASIC rivals with Groq deal

At GTC 2026, Nvidia unveiled a strategic US$20 billion partnership with AI chip startup Groq, licensing its LPU technology and hiring key team members to integrate advanced inference capabilities into the Vera Rubin server stack. This move positions Nvidia to define foundational rules for agentic AI while effectively blocking ASIC competitors like Google from encroaching on its market.

Friday 20 March 2026

Nvidia to supply 1 million GPUs to Amazon through 2027 in landmark AI cloud deal

Nvidia will supply 1 million graphics processing units (GPUs) to Amazon.com's cloud computing division by 2027 in one of the largest artificial intelligence infrastructure agreements to date, a company executive told Reuters.

Friday 20 March 2026

Nvidia Vera Rubin servers to drive liquid cooling demand

Nvidia has revealed more details about its next-generation Vera Rubin (VR) servers at Nvidia GTC 2026, confirming a full transition to liquid cooling architecture. Thermal module makers estimate that compared with the GB-series design, where each compute tray is paired with two liquid cooling modules, the VR generation will require up to seven modules per tray, more than doubling demand.

Friday 20 March 2026

Supermicro co-founder charged in US$2.5 billion AI chip smuggling scheme to China

US Federal prosecutors have charged a co-founder of Super Micro Computer Inc. (Supermicro) and two associates in a massive conspiracy to illegally divert billions of dollars worth of restricted AI technology to China.

Friday 20 March 2026

Analysis: Alibaba and Baidu raise AI pricing as token-based model reshapes cloud economics

Alibaba and Baidu have raised prices for AI computing services, reflecting a broader shift in how cloud providers monetize AI as demand for tokenized workloads accelerates.

Friday 20 March 2026

Alibaba outlines US$100 billion AI roadmap as cloud growth accelerates

Alibaba Group Holding set out an expansive artificial intelligence strategy targeting more than US$100 billion in combined cloud and AI revenue within five years, as strong demand for AI services continued to drive growth in its cloud business.

Friday 20 March 2026

Alibaba's AI and cloud growth cushions slowing earnings but raises profit questions

Alibaba Group reported modest overall revenue growth while disclosing a sharp fall in operating income, underlining a tension between investment-led expansion in AI and near-term profitability.

Friday 20 March 2026

Microsoft considers legal action over Amazon-OpenAI cloud deal

Microsoft is considering legal action against Amazon and OpenAI over a multibillion-dollar cloud arrangement that could breach its exclusive partnership with the ChatGPT creator, according to the Financial Times.

Friday 20 March 2026

The 'golden triangle': Inside Taiwan's industrial engine to win the AI robotics race

The next front in the global AI race may not be a data center. It could be a robot. Taiwan has quietly staked its claim — and it's building the industrial machinery to back it up.

Thursday 19 March 2026

Taiwan turns to Stargate playbook: private capital powers AI infrastructure

Foxconn (Hon Hai) plans to build an AI data center in Taiwan by 2027 with over 10,000 GPUs. That figure dwarfs the Ministry of Digital Affairs' (MODA) target of just 140 GPUs by the end of 2026. MODA minister Yi-Jing Lin has acknowledged the gap is real. Taiwan's answer, she says, is to follow the US Stargate playbook — mobilizing private investment through government tax incentives rather than direct public spending.

Thursday 19 March 2026

Nvidia elevates AI storage, putting NVMe SSD makers at the center

Nvidia is recasting storage as a core pillar of AI infrastructure, shifting it from a passive repository to an active extension of memory — a realignment now reshaping data center design and the broader supply chain.

Thursday 19 March 2026

Intel Xeon 6 wins CPU slot in Nvidia DGX Rubin, stakes claim in AI inference stack

Intel's Xeon 6 processors have been selected as the host CPU for Nvidia's DGX Rubin NVL8 system — a move announced at GTC 2026 that gives concrete form to the two companies' strategic alliance. The partnership's significance can be gauged through its product rollout timing and the revenue opportunities it unlocks.

Thursday 19 March 2026

Nvidia standardizes Vera Rubin liquid cooling, names four cold plate suppliers

Nvidia plans to launch its next-generation AI server architecture, Vera Rubin, in the second half of 2026, with liquid cooling set to become standard. The company will centralize procurement of cold plates and, at GTC, named four suppliers: Asia Vital Components (AVC), Cooler Master, Jentech, and Delta Electronics.

Thursday 19 March 2026

Taiwan's AI supply chain scales at GTC 2026: Foxconn, Wiwynn, Advantech, BizLink deepen Nvidia ties

Taiwanese hardware vendors expanded their presence at Nvidia GTC 2026, underscoring a coordinated push into AI servers, edge computing, robotics, and data center infrastructure. Foxconn, Wiwynn, Wistron, Advantech, and BizLink showcased capabilities spanning AI factories, rack-scale systems, and power interconnects, reflecting Taiwan's deeper integration into Nvidia's next-generation computing ecosystem.

Thursday 19 March 2026

Groq anchors Nvidia's inference strategy; CPU redefines architecture for AI agents

As AI evolves from generating information to executing tasks, inference scenarios characterized by coding agents and requiring low latency and high throughput are ushering in the next phase of AI infrastructure commercialization. However, even powerful systems like Vera Rubin face challenges when confronted with extreme generation demands. While Nvidia remains the undisputed leader in throughput, traditional GPU architectures appear too heavy for ultra-low-latency token generation. This is why Jensen Huang moved quickly to secure IP licensing and talent from Groq ahead of Christmas 2025.

Thursday 19 March 2026

AMD deepens ties with Naver in bid to expand AI infrastructure

Advanced Micro Devices (AMD) is set to deepen its collaboration with Naver Corporation after CEO Lisa Su visits South Korea, with the companies signing an MOU to jointly develop GPU and infrastructure technologies for large-scale AI models.

Wednesday 18 March 2026

Nexcom launches humanoid robot development kit built on Nvidia IGX Thor

Nexcom Group has launched its first humanoid robot development kit based on the Nvidia IGX Thor platform, unveiling the MARS400 T20 at the Nvidia GTC 2026. The kit integrates real-time motion control, functional safety modules, high-speed communication, and thermal management technologies. It is aimed at collaborative robot system integrators to shorten development and production cycles.

Wednesday 18 March 2026

Nvidia partners with chipmakers to advance industrial robotics

At Nvidia's GTC 2026 conference, major chipmakers — including Texas Instruments (TI), Infineon Technologies AG, NXP Semiconductors, Analog Devices (ADI), and Synopsys — announced new collaborations centered on robotics built around Nvidia's platform. Robotics is now the defining focus for IDM firms and the broader industrial control chip ecosystem.

Wednesday 18 March 2026

Asus, Foxconn take Taiwan smart city model global

Asus is teaming up with Foxconn and industry partners to develop a replicable "city-as-a-solution" model, positioning itself to capture demand as smart cities enter a fourth-generation AI City phase. The approach focuses on integrating group resources into scalable deployments that can be exported as complete city solutions.

Wednesday 18 March 2026

IBM completes Confluent acquisition to power real-time AI data

On March 17, IBM completed its acquisition of Confluent, Inc., the data streaming platform used by more than 6,500 enterprises — including 40% of the Fortune 500 — to power real-time operations. Under the agreement, IBM acquired all issued and outstanding common shares of Confluent for US$31 per share in cash, representing an enterprise value of approximately US$11 billion.

Wednesday 18 March 2026

OpenAI teams with AWS to target US govt AI contracts

OpenAI has signed an agreement with Amazon Web Services (AWS) to provide its artificial intelligence tools to US government agencies for both classified and unclassified work, according to The Information and Reuters, as the company steps up efforts to secure defense and public-sector contracts.

Wednesday 18 March 2026

Nvidia GTC 2026: Nvidia charts optical-copper path

At GTC 2026, Nvidia outlined a dual-track approach combining copper cables and silicon photonics (SiPh) to meet surging AI-driven demand for data center interconnects. CEO Jensen Huang emphasized that as AI compute needs escalate, future data centers will require expanded capacity across copper cabling, optical communication, and co-packaged optics (CPO) — establishing optical-copper parallelism as the industry's defining development path.

Wednesday 18 March 2026

Foxconn and SAP forge strategic alliance to drive enterprise AI in APAC

On March 18, Foxconn (Hon Hai) announced a strategic partnership with SAP to accelerate the adoption of next-generation enterprise artificial intelligence across the Asia-Pacific region. The collaboration is built on the AI Factory initiative, which Foxconn says seeks to reshape future manufacturing processes and supply chain management while opening pathways for global deployment.

Wednesday 18 March 2026

Exclusive: AI server supply chain hits hidden stalemate as CSPs and manufacturers favor price hikes over expansion

Geopolitical tariff conflicts combined with the outbreak of Middle East hostilities have pushed even the booming AI server and data center industry—and its supply chain—into a subtle deadlock. Supply chain insiders reveal that despite ample funding, major cloud service providers (CSPs) are hesitating to back suppliers' capacity expansions amid shortages.

Wednesday 18 March 2026

Chinese server maker Sugon unveils 400G AI fabric to rival Nvidia InfiniBand

Chinese server maker Dawning Information Industry Co. (Sugon) has introduced scaleFabric, its first fully self-developed 400G native remote direct memory access (RDMA) data center networking architecture, aimed at hyperscale AI clusters. The platform integrates 112G SerDes IP, switching chips, network interface cards and management software, all developed domestically, reflecting the company's push to build an independent high-speed AI interconnect stack.

More Server, IPC, cloud computing, IoT coverage