TÜV SÜD and Accenture are working with NVIDIA technologies to speed up safety certification of robots in factories and warehouses. They unveiled a joint "Physical AI Safety Suite" at NVIDIA GTC Berlin. The suite uses digital twins, AI agents, and simulations to partly replace manual assessments. TÜV SÜD estimates the digital process could bring production online up to 30% sooner. Certification against standards such as ISO 3691-4 is mandatory in many industrial countries before automated facilities can start. At a medium-sized plant, agreeing on safety concepts can take several weeks. The suite checks layouts, robot fleets, and individual robots before deployment. TÜV SÜD provides independent assessment and keeps responsibility for certification.

Accenture has capped employee AI spending at $29 (£21.75) a month after some staff ran up bills of $1,000, according to a report by 404 Media. The consulting firm wants to curb runaway costs as employees use AI for more everyday tasks. Companies are typically billed by usage, like a utility. Last week, some staff were told their access to Anthropic's Claude had been pulled entirely. Others are also reining in costs. Uber capped engineers' Claude spending at $1,500 a month after using up its AI budget in four months, while Microsoft has reportedly cut its Claude spending by about a third. The curbs come as Anthropic prepares for a public listing, targeting a valuation as high as $2tn. Investors are reportedly considering about $1.5tn.

Accenture is shifting AI work from pilots to deeply embedded systems, according to recent deals with UMass Memorial Health, Dell, Microsoft and Workday. The UMass Memorial project is a multi-year AI data platform, while the Dell alliance pitches private AI infrastructure to regulated clients. Analysts expect revenue to grow 5.2% a year over the next three years, with profit margins rising from 11.3% to 11.9%. They forecast earnings of $10.3 billion by 2029, up from $8.4 billion today, on revenue of $86.3 billion. The stock trades at a price-to-earnings ratio of 14.2x, against 21.9x for US IT peers. Risks include slower decision cycles in federal and EMEA work, which could curb consulting demand and margins.