AI Daily Brief: 29 October 2026

29 October 2026

AI suppliers are reporting rapid growth, but the practical questions are about control: over infrastructure, spending and the actions software can take. Today's briefing connects the UK sovereignty debate with new enterprise tools and the physical cost of scaling AI.

Bristol expert urges UK to secure its own AI capabilities

Professor Simon McIntosh-Smith, who directs the Bristol Centre for Supercomputing, has called for a sustained UK programme covering computing resources, foundation models, datasets and security. The University of Bristol reported on 28 October that the House of Lords Science and Technology Committee had chosen the future of UK compute as its next inquiry topic following his evidence.

He also identified affordable energy, connectivity and specialist skills as constraints. For UK businesses, sovereignty is about the ability to keep operating and change suppliers, not simply the location of a server.

Our take: Ask which parts of your AI service you can replace, export or run independently. UK hosting alone does not remove dependence on an overseas model provider.

Microsoft reports 48% Azure growth as cloud revenue reaches $63.9bn

Microsoft reported quarterly revenue of $92.3bn, up 19%, in results published on 28 October. Azure revenue rose 48%, while Microsoft Cloud revenue reached $63.9bn, up 30%.

The figures demonstrate strong supplier demand, but do not establish that every customer is getting a return from AI. UK organisations should separate the health of their cloud provider from the performance of their own projects.

Our take: Keep measuring cost per completed task, rework and staff time. A growing vendor is not evidence that an unused licence is worth renewing.

Google Cloud revenue doubles to $30.7bn

Alphabet reported third-quarter revenue of $127.6bn, up 25%, with Google Cloud revenue increasing 103% to $30.7bn. Its 28 October earnings release attributes cloud growth to enterprise AI infrastructure and solutions, core cloud services and sales of TPU systems.

The company said its first-party model APIs were processing nearly 32 billion tokens per minute, more than four times the year-earlier level. These are supplier usage figures, not an independent measure of customer productivity.

Our take: Demand is spreading across both model services and infrastructure. Compare complete deployment costs, including storage, support and migration, before choosing a cloud on headline token prices.

Claude Enterprise adds spending pools and reports on how teams use AI

Anthropic made Smart Reports, in-app surveys and pooled group budgets generally available on 28 October, according to Unite.AI. Reports can examine up to 28 days of activity, and organisations receive up to 10 reports a month without charge.

Reports are off by default and include restrictions on sensitive topics; Anthropic says they are not designed for individual performance assessment. Availability varies with the organisation's security and deployment configuration.

Our take: Start with team-level questions about useful outputs and wasted effort. Before enabling transcript analysis, review employee communications, data protection duties and who can reveal individual attribution.

ServiceNow workflows arrive in ChatGPT through a new plugin

OpenAI and ServiceNow expanded their partnership on 28 October, with a ServiceNow MCP plugin available in the OpenAI marketplace. Customers connect their own instance and choose the capabilities exposed; ServiceNow says its Action Fabric applies existing business rules and permissions.

Other capabilities remain planned: voice is expected in the first quarter of 2027, security features enter early access in January, and Codex integration in AI Workflow Factory is expected during 2027.

Our take: Distinguish what you can deploy today from the roadmap. Start with a narrow workflow and test permissions, logging and rollback before letting a conversation change operational records.

Google expands advertising inside AI-powered Search

Google announced an international expansion of AI Mode advertising alongside its quarterly results, TechCrunch reports. It is also introducing US ads that can appear in follow-up questions during an ongoing AI Mode conversation.

Search revenue grew 12% year on year, compared with 17% in the previous quarter. UK marketing teams should watch how conversational placements change discovery and lead quality rather than assuming existing search-ad performance will carry over.

Our take: Keep a clear baseline for enquiries, conversion and cost per sale. Test any new placement with a capped budget and track qualified leads, not just clicks.

Fragment raises $56m to automate procurement exceptions

Fragment launched from stealth with a $56m Series A led by 8VC, Tech Funding News reported on 28 October. Its software targets operational exceptions such as invoice mismatches and order changes, using business context from existing systems to carry out permitted resolutions.

The proposition is narrower than a general-purpose assistant: connect the relevant records, understand the policy and resolve a defined class of problem. Performance and integration claims remain the supplier's claims, not a guaranteed result for a new customer.

Our take: Invoice exceptions are a sensible pilot only when approval limits and escalation rules are explicit. Measure incorrect resolutions as carefully as time saved.

US nuclear loan commitment shows the scale of AI power demand

Constellation Energy has secured a conditional US government loan commitment of up to $5bn to expand nuclear generation, TechCrunch reported on 28 October. The planned work covers 11 reactors at eight plants and approximately 1.1 gigawatts of additional capacity.

Much of the output is already committed to Google, Amazon or Meta. The financing is not yet finalised, and the reporting does not establish that the additional generation will lower household electricity prices.

Our take: Power supply is now part of cloud capacity planning. For a UK deployment, ask about actual regional availability and delivery dates rather than relying on a provider's global investment total.

Frequently Asked Questions

How often is the AI Daily Brief published?

Every morning at 7:30am UK time, covering the previous 24 hours of AI news from over 30 sources.

How are stories selected?

UK-relevant stories are prioritised first, then by business impact and practical implications for UK organisations adopting AI.

Why should business leaders follow AI news?

AI is moving faster than any technology in history. Staying informed is essential for making smart decisions about AI investment, adoption, and governance.