AI and cloud briefing: Boston Scientific cyberattack halts global shipments, IBM opens up Granite 4.2 for AI agents, and Moonshot seeks a cloud deal for its 2.8-trillion-parameter Kimi K3

Here is our latest round-up of what’s changing in AI and cloud for UK businesses, with links to the original sources so you can read further.

Lead story: Boston Scientific cyberattack disrupts global operations and shipments

Medical device maker Boston Scientific confirmed that a cyberattack detected on 25 August has disrupted parts of its global operations, including the systems it relies on to process and ship customer orders. The company, which makes implanted devices such as pacemakers and defibrillators, said the incident affected both IT infrastructure and business applications. It has brought in outside cybersecurity specialists and has not given a timetable for full recovery. Boston Scientific has not yet said whether customer or patient data was accessed, or whether the attack involved ransomware. Its stock fell 5.8% in premarket trading the day the disruption was disclosed. Full detail is available from BleepingComputer.

Why this matters: this shows how a cyberattack can stop physical products moving, not just expose data. A medical device maker unable to ship orders is a supply chain problem as much as a security one. If your business depends on order processing or logistics systems, it is worth checking how quickly you could keep shipping if those systems went down, and whether your incident response plan covers operational continuity, not just data protection.

Frontier AI: IBM releases open-weight Granite 4.2 models built for enterprise AI agents

IBM released Granite 4.2, a family of open-weight AI models aimed at businesses that want reasoning and agent capabilities without sending every task to a large cloud-hosted model. The models come in 3 billion, 8 billion and 30 billion parameter sizes, all with a 128,000-token context window, and are released under an Apache 2.0 licence so companies can download, adapt and run them commercially. The larger models received extra training aimed at agent tasks such as using tools, working in a terminal, and following multi-step instructions. Granite 4.2 can run in the cloud, on a company’s own servers, or on smaller edge devices. Full detail is available from IBM Research and MarkTechPost.

Why this matters: this is a useful option for businesses that want AI agents working with sensitive company data but are cautious about sending everything to a large external provider. A smaller, openly licensed model that runs under your own control can be a sensible starting point before committing to a bigger frontier model contract.

Cloud & enterprise AI: China’s Moonshot AI seeks cloud deals with Microsoft, Amazon and Google

Chinese AI startup Moonshot AI is in early talks with Microsoft, Amazon and Google over revenue-sharing deals that would bring its Kimi K3 model to Azure, AWS and Google Cloud. Moonshot is reportedly seeking up to 30% of revenue generated from K3-related cloud services. Kimi K3 is a 2.8-trillion-parameter open-weight model, large enough that most companies cannot practically run it themselves, which is why cloud distribution matters so much. The talks are still at an early stage, with unresolved questions including how revenue would be calculated and how model usage would be audited. Full detail is available from Reuters.

Why this matters: if this deal goes ahead, it would be one of the first major revenue-sharing agreements between a Chinese AI developer and the big three Western cloud providers. It is a sign that demand for competitive, low-cost AI models can cross geopolitical lines even while governments tighten rules on chips and data. Businesses picking AI models through a major cloud provider may soon have more non-US options to weigh up, alongside the usual questions about data handling and compliance.

The takeaway

Today’s stories cover three different pressures businesses are facing with AI and cloud right now. A cyberattack on a medical device maker shows that security incidents increasingly hit physical operations, not just data. IBM’s Granite 4.2 gives companies a way to run capable AI agents on their own infrastructure rather than a public cloud. And Moonshot’s cloud talks suggest the choice of AI model available through your cloud provider is about to get more varied. We can help you review your incident response plans for operational continuity, assess whether a smaller self-hosted model fits your AI use case, and think through the compliance implications of new model options as they arrive. Get in touch.

AI Agent based research and content, AI can do mistake, please help us to improve.