Editorial Team
6 min read
Centrica says its roughly 1,300 proposed role reductions reflect lower customer contact and a wider transformation programme. The record shows AI is one tool in that programme, while the headline total reaches beyond the 500 contact-based roles it has specifically identified.
Centrica, the FTSE 100 owner of British Gas, is shrinking parts of its customer-service operation while moving more interactions to digital channels. Its headline figure—about 1,300 role reductions either delivered or under implementation—is broader than a count of telephone jobs. That distinction matters when assessing both the company’s claim that changing customer behaviour is driving the plan and the union’s claim that people are being replaced by AI.
Company-reported role totals: Centrica proposed around 500 contact-based roles and said its announcements represented around 1,300 reductions overall. Source: Centrica interim results.
Chris O’Shea, Centrica’s group chief executive, has led the company since 2020 after joining its board as group chief financial officer in 2018. His prior finance role matters because the changes are part of a company-wide efficiency programme as well as a customer-service redesign; Centrica’s 2020 appointment statement set out “structural simplification” as a strategic objective.
In its interim-results filing, Centrica says it has proposed around 500 contact-based reductions in customer operations, with additional reductions in offshore outsourced support. It also says a 2025 structural reorganisation produced proposed cuts in group support functions. Those announcements together make up the approximately 1,300 reductions, either completed or currently being implemented.
The filing describes the customer-operations element as an approximately 14% workforce reduction. It does not establish that all 1,300 posts are British Gas call-centre jobs, nor that every one is a compulsory redundancy: it says staffing has also been managed through natural attrition.
The earlier 500-role proposal affected British Gas Energy and Services call-centre teams in Glasgow, Edinburgh, Cardiff, Leicester, Stockport and Leeds. Energy handles accounts, bills, complaints and payment difficulties; Services handles repairs and breakdowns covered by HomeCare policies. A report based on material shown to staff says the company was consulting on those roles and offering voluntary redundancy where possible.
Centrica says average customer contact fell 20% year on year and about 90% of customers now use digital self-service channels, two percentage points more than a year earlier. Its filing attributes the change to efforts to remove root causes of contact, simplify journeys and re-platform technology, supported by the “targeted deployment of AI tools.” It reports lower complaints per customer and higher customer-satisfaction measures, but those are company-reported operating metrics rather than a measure of whether AI removed particular contacts.
O’Shea said in an interview about the plan that “AI isn’t driving these particular job reductions”; he pointed to changing customer behaviour and said the company expected more work around its digital interface and fewer people on the phones. The company’s contemporaneous filing supports the more limited conclusion: AI is named as one enabling tool, alongside changes that cannot be separated from it on the published evidence.
The GMB disputes that interpretation. Its national secretary, Charlotte Brumpton-Childs, said staff were being replaced by chatbots, while British Gas said the suggestion was “simply wrong” and that inbound contact was down around 31% since 2023. The figures use different comparison periods and measures: the company’s filing gives a 20% year-on-year fall in average contact per customer, while the staff material cited in the report says Energy and Services contacts fell from about 20 million in 2023 to 16 million in 2025 and were forecast to reach 14 million in 2026.
Neither channel use nor lower contact proves a general preference for an AI chatbot. In an Office for National Statistics survey conducted from 3 to 28 June, 27% of Great Britain adults said they would trust AI for customer service and support, while 32% said they would not trust AI with any task. The weighted survey had 3,510 responding people aged 16 and over; it is not a British Gas customer survey, and respondents could select more than one task, so it cannot determine the supplier’s channel demand.
Centrica recorded £92 million of technology-led transformation investment in the first half of 2026—£75 million in adjusted EBITDA and £17 million in capital investment. It aims to keep operating costs, excluding bad debt and depreciation, broadly flat in nominal terms from 2025 to the end of the decade. The company says that requires a £500 million underlying cost reduction to absorb inflation and growth costs; it estimates total programme costs of about £600 million, including about £400 million of operating costs and £200 million of capital investment.
This is not a picture of a retail business simply retreating. Retail adjusted EBITDA rose to £346 million in the first half from £338 million a year earlier, although UK energy customers fell to 7.45 million from 7.50 million at the end of 2025. Centrica says average fixed-price margin per UK energy customer was 10% above the end-2025 level. Group adjusted EBITDA, however, fell to £737 million from £900 million, reflecting infrastructure disposals, outages and lower realised nuclear prices. The staffing plan therefore sits within a stated effort to improve efficiency and protect value while the group invests elsewhere.
British Gas says contact-centre support will remain for customers who need additional help and that its transformation strategy reserves it for people who “fall out of the digital journey.” That is a consequential service boundary: it determines whether a difficult bill, payment problem or complaint becomes resolvable before a person is reached.
The company’s recent record raises the importance of that boundary without proving that the present programme will fail. In May, Ofgem said its investigation found that British Gas failed required standards and breached licence conditions designed to protect vulnerable customers during prepayment-meter installations. The settlement covers compensation for affected customers from 2018 to 2021, a £20 million payment to Ofgem’s Voluntary Redress Fund and write-offs of up to £70 million in energy debt for vulnerable customers. Ofgem also says British Gas has since strengthened its procedures and safeguards; the investigation concerns historic practices, not the current digital programme.
The next question is not whether British Gas will use more digital service—it already is—but whether the reduced human operation works for customers whose problems are not routine. Centrica has published contact, channel-use and satisfaction figures. To assess the consequence of the redesign, it would need to show results for cases escalated out of digital channels: time to a human adviser, complaint resolution, repeat contact, debt-support outcomes and outcomes for customers identified as vulnerable.
Without that breakdown, the company has evidence of a changing operating model and a financial incentive to make it work. It does not yet have public evidence that lower contact and smaller teams deliver better outcomes for the customers least able to self-serve.
AMD has introduced ROCm.AI to guide setup, deployment and optimization on its hardware, while HIP and HIPIFY offer a separate source-level route from CUDA for embedded software. The unresolved question is whether those tools lower enough engineering risk to matter for large planned AI deployments.
KTransformers v0.6.4 adds LoRA and hybrid fine-tuning to its CPU-GPU workflow for sparse mixture-of-experts models, expands an Ampere inference path and moves a scheduler endpoint to loopback after a report of unauthenticated pickle deserialization. Its published throughput figures are configuration-specific release claims, not a general comparison of fine-tuning methods.
Microsoft has put MAI-Image-2.5-Pro and MAI-Voice-2-Flash into public preview in Foundry. The launch extends the company’s in-house AI options, but its evidence for lower cost and broader deployment is chiefly tied to MAI-Image-2.5 and company-reported workloads—not an independently comparable test of the new image model.
Microsoft has made its MAI-Code-1-Flash coding model available in paid GitHub Copilot plans and says a related model is live in Excel. Its evidence points to a targeted effort to reduce serving costs, while leaving the scale, task coverage and economics of the broader rollout unresolved.
AMD has detailed the 256-core EPYC 9996, its density-oriented Venice server CPU, alongside a broader 9006 platform with 16 memory channels and PCIe 6.0. But the Helios AI rack uses a separate high-frequency Venice design, and AMD’s largest rack-performance figures remain preliminary company projections rather than independent results.
NVIDIA says it will commit $4 million over three years to let CNCF projects test on real GPUs. With Kubernetes’ core Dynamic Resource Allocation framework already generally available, the consequential question is whether shared validation can make diverse GPU implementations dependable without settling operators’ choices over isolation and control.
Microsoft has made MAI-Image-2.5 the end-to-end default for Bing Image Creator and put it into PowerPoint image-to-image features. The company says the PowerPoint deployment can reduce GPU costs by up to 84% versus GPT-Image-2, while its much larger OpenAI partnership remains in place.
Meta is using a new mass-market film to argue that AI will extend its mission of connection. The company has real distribution, product and infrastructure plans behind the pitch, but the meaningful test is whether it makes access, controls and outcomes legible as those plans reach users.
Google is rolling out an optional selfie-video check for eligible Google Account holders who lose access to their usual devices. The feature adds a biometric reference to recovery, but it excludes several account types and leaves users dependent on Google’s unreported real-world performance and recovery decisions.
Google’s first ATLAS report finds Gemini use across occupations covering 88.4% of U.S. employment, but mostly in a limited share of each job’s tasks. The data offer a detailed view of how Google’s AI is used; they do not show whether it is raising output, eliminating work or changing hiring.
Nokia booked €2.8 billion in AI and cloud order intake in the second quarter, as optical and IP-network sales grew. The orders strengthen its data-centre strategy, but they are not revenue yet, and restructuring, supply constraints and negative cash flow leave execution as the central question.
Super Micro Computer says it received more than $60 billion in fiscal fourth-quarter orders and now expects 15% to 17% gross margins, despite revenue tracking near the low end of its outlook. The preliminary figures point to a much larger AI-server delivery pipeline, but not yet to booked revenue or a durable improvement in profitability.
Nebius says it has brought up its first full NVIDIA Vera Rubin NVL72 rack in Finland. The deployment is an early integration milestone backed by a $2 billion NVIDIA private placement, but the company still has software and production testing to complete before customers can use it.
According to people familiar with the effort, Amazon founder and executive chairman Jeff Bezos has pressed Prime Video to make AI and personalization central to a proposed redesign known internally as Lighthouse. The early test could improve discovery, but Amazon has not said how it would reconcile personalized rankings with the paid placement and subscriptions that make Prime Video a complex storefront.
A report says some Chinese cities are gradually issuing robotaxi permits again after a review prompted by Baidu’s Wuhan outage. Shenzhen’s revised rules show that road testing, demonstrations and unmanned trials remain separately governed—and do not by themselves establish broad commercial service.
Databricks says it will run core operations and analytics on Azure Databricks, deepen Microsoft product integrations and expand use of Azure Cobalt chips. The partnership runs into the 2030s, but the companies have not disclosed its value, workload scale or customer results.
OpenAI is beginning a U.S. rollout of Health in ChatGPT for logged-in adults on web and iOS. The feature can bring optional Apple Health and supported medical-record context into ordinary chats, but connected data may be incomplete and requires user permission before use.
Yelp will supply ChatGPT with reviews, ratings, photos and business details, with branded links back to Yelp and a planned quote-request feature. The non-exclusive deal extends an existing licensing strategy, but neither company has disclosed its financial terms or how ChatGPT referrals will be measured and sold.
U.S. stocks fell as attacks on Saudi tankers added a second threat to Middle East oil flows and earnings from Alphabet and Tesla highlighted the cash demands of the AI buildout. The selloff was concentrated in megacap technology rather than a uniform retreat from equities.
Blackstone’s second-quarter distributable earnings rose 26% as realizations and fee-related earnings increased. Its data-center and other AI-linked investments are becoming a larger part of the firm’s portfolio narrative, but the results do not separately show how much of the gain came from AI while retail private-credit fundraising slowed.