Manufacturing & energy · Food & beverage · European Union
Emerging
You have started, but value is not yet captured. The priority is moving from pilots to one production use case with a business case behind it.
What AI could be worth to you
$636k a year in the expected case
Payback in 3.6 years; five-year return −45% conservative, 37% expected, 179% upside. Your readiness (Emerging, 36/100) is what sets these numbers, and it can be raised.
Expected annual profit effect at run-rate
0.53% of revenue
$636k a year
7% of today's operating profit
Conservative 0.21% · Upside 1.1%
Of which cost savings · revenue growth
0.38% savings
$456k a year
0.15% margin on new revenue
$180k a year (on $720k of extra revenue)
Investment needed, per year for 3 years
0.37% of revenue
5% of operating profit
≈ $445k a year
About 45% of that continues as running cost after year three.
What the investment could return
| Conservative | Expected | Upside | |
|---|---|---|---|
| Task productivity gain on addressable work | 12% | 20% | 30% |
| Value actually captured | 30% | 40% | 50% |
| Annual cost savings, % of revenue | 0.16%$189k | 0.38%$456k | 0.80%$954k |
| Revenue uplift by year 3 | 0.23%$270k | 0.60%$720k | 1.1%$1.4M |
| Total annual profit effect, % of revenue | 0.21%3% of op. profit | 0.53%7% of op. profit | 1.1%14% of op. profit |
| Annual investment, % of revenue, years 1–3 | 0.37%5% of op. profit | 0.37%5% of op. profit | 0.37%5% of op. profit |
| Payback | beyond 7 years | 3.6 years | 21 months |
| Five-year return on investment | −45% | 37% | 179% |
Cumulative benefit vs cost, % of revenue
Typical use cases for your profile
Where the savings usually come from in food & beverage.
- 01
Procurement offer processingDemo
An AI agent reads supplier offers and drafts the side-by-side comparison for the buyer to approve.
- 02
Predictive maintenance
Starting with data collection and plant mapping, then models that anticipate equipment failures.
- 03
Quality control with computer vision
Cameras on the line that spot defects consistently and record every inspection.
Why these numbers, and how to raise them
Emerging. You have started, but value is not yet captured. The priority is moving from pilots to one production use case with a business case behind it.
You are here · peers by revenue
You are ahead of about 41% of companies of $100–500M revenue and level with another 32%. EU-wide, 20% of enterprises use AI (Eurostat 2025): 17% of small, 30% of medium and 55% of large firms. BCG expects European firms to gain 7% of revenue from AI by 2028 against 8% in North America; Europe trails on agents (41% vs 51% North America, BCG 2025).
McKinsey State of AI 2025, phase of AI use by revenue band.
How far, and how fast
Exploring
Emerging
You are here
Scaling-ready
with a programme: ~12 months
Leading
~27 months with a programme
Reaching Scaling-ready takes about 12 months with a structured programme, against 30 months or more on your own; McKinsey's data show most firms sit two to three years in the pilot phase. The value capture rate rises from 40% to 55% on the way, which is what changes the return, not the tools.
What peers in manufacturing & energy report
| Use AI in at least one function | 86% |
| Have AI agents in production | 56% |
| Invest in AI, % of revenue (2025 → planned 2026) | 0.7% → 0.8% |
| Have scaled agents in IT | 15% |
McKinsey State of AI 2025/2026, BCG AI Radar 2026, Google Cloud ROI of AI 2025.
Where help would change the outcome
- 1
Pilot to production
Take the best pilot into daily use with evaluation, guardrails and an owner. Most firms stall exactly here.
- 2
Executive ownership and a costed roadmap
Name an owner with budget, agree three priority use cases and a target for each. CEO oversight of AI is the leadership practice most correlated with profit impact.
- 3
Data foundation before models
Map where the data for the priority use cases lives, fix access and quality, and set up a minimal integration layer. Poor data is the most common reason AI pilots are abandoned.
- 4
Map and standardise the target processes
Document the two or three processes AI will touch, remove steps that only exist because of old tools, then automate. Without this, AI speeds up a process nobody has agreed on.
What your own team can carry
| Your team | Integrations, data preparation and first internal automations, with our architecture and evaluation support. |
| Aries | Value assessment, production hardening, guardrails and governance, pair-building. |
Get the full report and a demo for your sector
The report adds sector benchmarks, the three use cases for your profile with a business case each, a 90-day plan and the sources behind every number. The demo shows one of those use cases running on data like yours.
Assumptions and how the estimate is built +
Value = labour capacity released + input savings + margin on new revenue − programme cost. Labour capacity is personnel cost × the addressable share of work × the task productivity gain × the share of that gain the organisation actually captures. Input savings apply a small savings rate to addressable non-labour cost. Revenue growth is valued at its contribution margin, not at face value.
| Sector preset used | Manufacturing & energy: personnel cost 15% of revenue, addressable non-labour cost 55%, three-year revenue uplift potential 1.5%, contribution margin on new revenue 25% |
| Addressable share of work | 17% (from your answer on repetitive work) |
| Task productivity gain | 12% conservative · 20% expected · 30% upside |
| Value capture rate by tier | Exploring 25% · Emerging 40% · Scaling-ready 55% · Leading 70% (−10 / +10 points in conservative / upside, minimum 10%) |
| Non-labour savings rate | 0.4% conservative · 0.8% expected · 1.5% upside of addressable non-labour cost |
| Investment sizing | Expected annual benefit × a tier multiplier (0.90 / 0.70 / 0.55 / 0.45), kept between 0.25% and 2.5% of revenue per year for three years, then 45% as running cost. Cross-check: BCG reports firms moving from 0.8% to 1.7% of revenue on AI. |
| Ramp | Benefits build up linearly over 30 months to full run-rate. |
Sources behind the calibration: Brynjolfsson, Li & Raymond; Cui et al.; METR; Dell'Acqua et al.; Noy & Zhang; Koch et al.; Acemoglu et al.; Yotzov et al. 2026; US Census 2026 AI supplement; PwC 2026 CEO Survey; McKinsey State of AI 2025–2026; BCG AI Radar 2026; Eurostat structural business statistics. This is an indicative estimate for a first conversation, not a business case.
Indicative estimate. Answers are stored only if you request the report. Privacy notice