AI

Last year, AI server shipments grew 46%, says TrendForce, but this year there is more uncertainty caused, in part, by the DeepSeek effect. DeepSeek’s influence will drive CSPs toward lower-cost proprietary ASIC solutions, shifting focus from AI training to AI inference.   This shift is expected to gradually increase the share of AI inference servers to nearly 50%. DeepSeek may ...

Synopsys has added the HAPS-200 prototyping and ZeBu-200 emulation system to its hardware-assisted verification (HAV) portfolio. Both the HAPs-200 and ZeBu-200 are based on the AMD Versal Premium VP1902 adaptive SoC and built on the company’s emulation and prototyping, or EP-Ready, hardware for use with reconfiguration and optimisation software. ZeBu Server 5 delivers scalability beyond 60bn gates, to target large ...

Datacentre capex is projected to surpass $1 trillion by 2029, says Dell’Oro. AI infrastructure spending will maintain its strong growth momentum despite ongoing sustainability efforts. “We project that data center infrastructure spending could surpass $1 trillion annually within five years,” says Dell’Oro ‘s Baron Fung, “while AI spending has yet to meet desired returns and efficiency improvements, long-term growth remains ...

In 2027, companies will spend more than $30 billion on AI-related infrastructure, platforms, software, and services to support their ability to compete on highly personalised customer experiences, says IDC. “Marketing leaders must remember that technology itself is not a differentiator but an enabler of differentiations. To stand out, they should focus on creating value-based narratives that build authentic connections with ...

Codasip and RED Semiconductor have signed an MoU  to collaborate on developing  AI acceleration technologies.  Under the MoU, RED will leverage the Codasip Studio processor design tools to integrate its VISC technology as an accelerator for Codasip RISC-V cores.  RED’s AI accelerators will also be demonstrated and sold with Codasip’s RISC-V processors. Versatile Intrinsic Structured Computing (VISC) is an innovation ...

DeepSeek, a Chinese AI software programme developer, is causing consternation in the AI community by saying it can train an industry leading programme at a cost of $5.6 million compared to the $100 million to $1 billion cost cited by top US AI developers Last week DeepSeek launched a programme called R1, for complex problem solving, that was trained on ...

A personal supercomputer, Project Digits, has been announced by Nvidia. Based on the new Grace Blackwell superchip (GB10), it is intended for AI research and data scientists to develop and run inference on models using a desktop system. It can run 200bn-parameter models which can be seamlessly deployed on accelerated cloud or data centre infrastructure. This increases to up to ...

Softbank is out to challenge Nvidia in chips to train and run AI software programmes in a project called Izanagi which could see investment of $100 billion. The plan is to have prototype Izanagi processors in summer 2025 and to be shipping in volume in 2026, reports Bloomberg. Izanagi (pictured) is Japan’s God of creation. Central to the scheme are ...

Revenue for accelerators, consisting of GPUs and custom accelerators, increased by 130 percent in 3Q 2024, says Dell’Oro. The US hyperscalers―Amazon, Google, Meta, and Microsoft―are set to deploy over 5 million AI training-capable accelerators in 2024. “Demand for accelerators has been growing at a breakneck pace as the hyperscalers race to deploy infrastructure for the training and inference of large ...

There was an 82% increase in hyperscale data centre capex in 3Q 2024 while the general-purpose server and storage markets have had double-digit revenue growth for the past three quarters, says Dell’Oro. “Accelerated servers, mainly targeted for AI training workloads, have accounted for the majority of the spending among the US and Chinese hyperscalers year-to-date,” says Dell’Oro’s Baron Fung, “while ...