UK businesses are entering 2026 with a clear, measurable ambition: raise productivity by investing in artificial intelligence, technical infrastructure, and workforce skills—while confronting a widening talent shortage that threatens to slow or skew the benefits of automation.
Lloyds Bank’s latest Business Barometer shows a marked shift in priorities for UK firms entering 2026: productivity is top of the agenda, with many firms planning targeted investments in training, AI tools and IT upgrades to translate that ambition into measurable performance gains. The Lloyds survey of 1,200 firms finds that 42% of businesses rank improving productivity as their leading priority for the year ahead, while 39% cite upskilling and 37% cite strengthening technology infrastructure. Those headline priorities are not abstract: Lloyds reports that prior waves of the Business Barometer show strong correlations between AI adoption and firm-level outcomes—82% of AI-using firms reported higher productivity and 76% reported improved profitability—and many firms plan fresh AI investments in 2026. This article dissects what the numbers mean for IT leaders, HR chiefs and CFOs, compares corroborating evidence from independent sources, and offers a practical roadmap for turning AI intent into durable productivity gains while managing talent, governance and infrastructure risks.
To be competitive, firms must treat 2026 as a year of disciplined execution: narrow, measurable pilots; investments in data plumbing; a clear people strategy to hire, retain and reskill; and governance practices that make outputs auditable and safe. The Lloyds data and independent reporting are consistent on the promise—AI and skills are the levers—but they are also emphatic about the prerequisites: infrastructure, talent and governance. If UK firms can synchronise those levers, 2026 could be the year when pilot enthusiasm becomes measurable productivity growth at scale rather than a patchwork of promising experiments.
Source: Petri IT Knowledgebase UK Firms Bet on AI and Skills to Drive Productivity in 2026
Background
Lloyds Bank’s latest Business Barometer shows a marked shift in priorities for UK firms entering 2026: productivity is top of the agenda, with many firms planning targeted investments in training, AI tools and IT upgrades to translate that ambition into measurable performance gains. The Lloyds survey of 1,200 firms finds that 42% of businesses rank improving productivity as their leading priority for the year ahead, while 39% cite upskilling and 37% cite strengthening technology infrastructure. Those headline priorities are not abstract: Lloyds reports that prior waves of the Business Barometer show strong correlations between AI adoption and firm-level outcomes—82% of AI-using firms reported higher productivity and 76% reported improved profitability—and many firms plan fresh AI investments in 2026. This article dissects what the numbers mean for IT leaders, HR chiefs and CFOs, compares corroborating evidence from independent sources, and offers a practical roadmap for turning AI intent into durable productivity gains while managing talent, governance and infrastructure risks.Where firms will spend: AI, training and infrastructure
The investment trifecta
Across surveys and press coverage, three linked investment themes recur:- AI tools and automation: roughly one-third of firms say they will focus on AI this year, signalling movement from experimentation to expansion.
- Team training and upskilling: about 35% plan to invest in team training in 2026, with upskilling noted as a central priority to capture AI’s potential.
- Technology upgrades: 37% intend to strengthen infrastructure—cloud, data platforms and integration layers needed to run AI at scale.
Which sectors report the biggest gains
Lloyds’ prior findings identify retailers as seeing the biggest productivity effects and manufacturers as realising the strongest profitability uplift from AI—signals that use-cases tied to inventory, demand forecasting and customer communications are delivering tangible returns.The skills bottleneck: scale, money and mismatch
Vacancy volumes and concentration
The UK’s surge in AI interest runs headlong into a skills shortage. Industry analyses show more than 11,000 active vacancies in automation and AI-related roles during the previous summer, with AI positions making up nearly 70% of that demand and high need for data engineers and Python specialists. This skills shortfall shapes hiring, pay and the ability of organisations to scale projects beyond pilots.Escalating pay pressures
Demand has translated to higher compensation. Public-facing salary research and reporting indicate substantial median pay for top AI roles in the UK—AI and machine learning engineers with median pay figures discussed in the mid-five figures to low six figures (£100k+), and engineering managers also commanding strong packages. Independent trade reporting highlights the gap between UK and US compensation but confirms steep domestic rises in AI-related salaries. This creates two immediate consequences: (1) competition for senior hires is expensive, and (2) smaller firms will struggle to attract talent without creative hiring or partnership models.Implications for workforce strategy
The skills story forces firms to be deliberate about how they fill capability gaps:- Build focused training for adjacent roles (data literacy, prompt engineering, verification workflows).
- Prioritise hiring for scarce roles where in-house capabilities unlock ROI (data engineering, MLOps).
- Use partnerships with vendors, consultancies and educational institutions to scale learning and talent pipelines.
- Offer competitive compensation packages and non‑salary incentives (learning budgets, role design, career routes).
What the evidence says: productivity and profitability gains are real — but conditional
Hard outcomes reported by firms
Lloyds’ internal survey and reporting show a strong self-reported link between AI adoption and improved performance: 82% report productivity increases and 76% report higher profitability among AI users. These are compelling signals for boards weighing investments. Independent outlets and analyst commentary amplify the point while adding necessary caveats: the magnitude of impact varies by sector, by the maturity of data infrastructure, and by whether firms have governance and measurement systems in place to convert time savings into measurable output and margin. In short: AI delivers—if you have the plumbing and the people.The J-curve and the cautionary evidence
Several independent analyses document a common adoption pattern: initial implementation often increases near-term operational effort (integration, pilot management, verification) before longer-term productivity benefits materialise. This J‑curve effect is visible in firm-level case studies: early pilots save time on drafting and triage tasks but require human-in-the-loop review and governance, which imposes costs that only dissipate after reliable pipelines and MLOps practices are established.Risks that can derail AI-driven productivity
Data and governance risks
The most immediate operational risks are data leakage, model‑training ambiguity, and insufficient audit trails. Firms using public or ill-specified endpoints risk exposing PII or IP and inadvertently allowing vendor models to be trained on sensitive inputs. Practical mitigation: insist on enterprise contracts with non‑training clauses, use tenant-grounded models, and implement logging and retention policies. These governance steps are repeatedly flagged in sector guidance and industry playbooks.Talent concentration and inequality of access
High salaries and concentrated demand for AI specialists are already producing winner-takes-most dynamics: large firms with deeper pockets can staff full AI teams and absorb integration costs, while smaller firms risk being limited to canned SaaS tools or shadow‑AI experiments. That asymmetry risks widening productivity gaps across the economy unless public and private initiatives expand training capacity and affordable partnerships.Vendor lock-in and architectural fragility
Deep integration with a single vendor’s productivity stack (for example, embedding Copilot-like assistants into core systems) lowers friction but increases switching costs and creates a structural dependency. Vendor lock-in can constrain future innovation and raise data‑egress costs. Decision-makers should weigh convenience against flexibility and insist on exportable artifacts and consistent metadata to reduce exit friction.Infrastructure and environmental costs
Scaling AI requires reliable cloud platforms or on‑prem compute, both of which come with cost and energy implications. Data‑centre capacity and specialised hardware (GPUs/accelerators) drive capital intensity, and firms that underestimate the compute and energy costs risk ballooning TCO and delayed ROI. Industry analyses emphasise careful capacity and cost modelling as essential to responsible scaling.A practical, high‑precision roadmap for 2026
The generic advice to “invest in AI and training” is not enough. The following 10-step plan is designed to help firms convert intent into measurable results within 12–24 months.- Define 2–3 high‑impact use cases tied to measurable KPIs (time saved, error reduction, revenue lift). Focus pilots where automation unlocks clear margins.
- Establish a one‑page AI strategy aligned to business goals and budget thresholds. Make the board accountable for risk and ROI.
- Inventory and prioritise data assets. Clean, accessible data beats raw compute: treat data engineering as the first production problem.
- Build a governance baseline before enterprise rollout: data policies, human‑in‑the‑loop checkpoints, vendor contracts with non‑training clauses, and logging for audit.
- Run time‑boxed pilots (6–8 weeks) with clear gates for escalation or termination. Require measurable baselines and reproducible test datasets.
- Invest in targeted, role‑specific upskilling (microlearning, on‑the‑job labs, prompt craft, verification training) rather than generic awareness sessions.
- Secure scarce talent strategically: combine selective hiring for core roles, vendor partnerships for productised capabilities, and apprenticeships or sponsorships with universities to build long-term pipelines.
- Protect operational data—use enterprise endpoints, DLP policies, and tenant‑grounded retrieval systems for sensitive outputs.
- Price total cost of ownership correctly, factoring in MLOps, verification overhead and infrastructure; model scenarios where short-term verification costs may temporarily reduce measured productivity.
- Measure and publish governance KPIs internally: error rates, verification time per artefact, model drift incidents and remediation cost. Make safety a visible budget line, not an afterthought.
Practical models for bridging the skills gap
Internal training plus external pipelines
Many firms will combine internal micro‑learning (role-specific workshops and verification drills) with external partnerships. The most effective models observed across sector playbooks include:- Sponsored apprenticeships and rapid retraining programs with local universities.
- Vendor-led microcredentials that map directly to the firm’s operational stack.
- Dedicated AI 'champions' who bridge procurement, IT and frontline teams and act as internal consultants to make adoption repeatable.
Compensation and retention levers
Competitive pay is essential but not sufficient. Firms that successfully attract talent often combine:- Clear career paths (MLOps → Data Platform Lead → AI Product Owner).
- Learning budgets, conference sponsorships and funded certifications.
- Role redesign to make work more interesting (e.g., human verifier, model governance lead).
Governance: the non‑negotiable foundation
Regulation, professional standards and litigation risk are converging to make governance a board-level topic. Across high-risk industries (financial services, legal, healthcare), regulators already emphasise documentation, audit trails and human accountability. Practical minimums for deployers include:- Contractual guarantees on model training, retention and deletion.
- Versioned logs and reproducible testing datasets.
- Human sign-off processes for any externally used AI output.
Strengths, limits and where to be cautious
Strengths to exploit
- Immediate wins: drafting, summarisation and internal query tasks produce fast, measurable time savings.
- Customer impact: retail and customer service applications already show strong improvements in response times and personalisation.
- Compounding returns: when data platforms and governance are in place, gains can scale across functions.
Limits to acknowledge
- Short-term friction: pilots require extra verification and engineering overhead that can temporarily depress measured productivity.
- Uneven distribution: smaller firms and labour-intensive sectors may lag without policy or partnership support.
- Measurement ambiguity: many early claims are self-reported; firms need rigorous KPIs to distinguish novelty effects from durable gains.
Final analysis: turning intent into durable advantage
The confluence of firm-level priorities, rising AI adoption and aggressive upskilling plans shows the UK entering 2026 with realistic ambition to move productivity forward. However, the work ahead is not purely technological—the decisive margin will be human and managerial. Organisations that pair targeted AI pilots with deliberate data engineering, robust governance, and role‑specific training will capture the productivity prize; those that treat AI as a feature set to bolt on existing workflows risk wasted investment, governance blowback, and loss of trust.To be competitive, firms must treat 2026 as a year of disciplined execution: narrow, measurable pilots; investments in data plumbing; a clear people strategy to hire, retain and reskill; and governance practices that make outputs auditable and safe. The Lloyds data and independent reporting are consistent on the promise—AI and skills are the levers—but they are also emphatic about the prerequisites: infrastructure, talent and governance. If UK firms can synchronise those levers, 2026 could be the year when pilot enthusiasm becomes measurable productivity growth at scale rather than a patchwork of promising experiments.
Source: Petri IT Knowledgebase UK Firms Bet on AI and Skills to Drive Productivity in 2026
