Artificial intelligence is becoming part of everyday business, with tools such as Anthropic’s Claude, ChatGPT, Gemini and Copilot now supporting tasks from research and writing to coding, data analysis and customer service. Claude is one of the strongest AI tools currently available and many organisations are understandably considering it as part of their digital strategy.
As of mid-2026, Anthropic has not published a full corporate sustainability report with audited Scope 1, 2, and 3 emissions, unlike Microsoft and Google/Alphabet. However, as AI adoption grows, businesses with carbon reduction plans, net zero targets or Scope 3 reporting requirements should consider the environmental impact of the tools they use. The impact of individual AI queries is usually very small, but the wider issue is about transparency, reporting and the cumulative effect of AI being used at scale.
| Quick navigation points throughout the blog | |
| 1. The Main Issue With Antrhopic: Limited Public Emissions Data | 2. Per-Query Emissions Are Usually Very Small |
| 3. Why AI's Environmental Impact Still Matters | 4. What Should Businesses Do? |
One of the key challenges with Anthropic is that, as a private company, it does not publicly report its organisational emissions in the same way as larger technology companies such as Google and Microsoft. This does not mean Claude is necessarily more carbon intensive, but it does make emissions reporting less precise for businesses using the platform.
Where supplier specific emissions data is unavailable, organisations often use spend-based emissions factors for Scope 3 reporting. This is a practical method, but it uses industry average data rather than company specific figures.
For example, using a spend-based proxy of 0.1177 kg CO₂e per £ spent, a company spending £10,000 per year on Anthropic or a similar AI/software service would estimate:
1,177 kg CO₂e, or 1.177 tonnes CO₂e, in annual emissions;
an offset cost of around £17.66 per year, assuming £15 per tonne.
For comparison, using organisation emissions intensity from public reporting, the same annual spend could be estimated at around 0.627 tCO₂e for Microsoft Copilot or 0.513 tCO₂e for Google Gemini. At £15 per tonne, this would equate to offset costs of approximately £9.41 and £7.70 respectively. See Table 1 below for a side-by-side comparison.
| AI Provider | Emission Factor (kg CO₂e/£) | Estimated Emissions (tCO₂e) | Offset Cost (£15/t) | Notes |
| Anthropic (Claude) | 0.1177 | 1.177 | £17.66 | DEFRA SIC 62.01 industry average (spend-based proxy) |
| Microsoft Copilot | 0.06271 | 0.627 | £9.41 | Based on Microsoft FY2025 reported intensity |
| Google Gemini | 0.05130 | 0.513 | £7.70 | Based on Alphabet/Google reported intentisty |
The difference is small in financial terms. Anthropic may add roughly £8-10 per year in offset costs compared with Microsoft or Google equivalents. The more important issue is not cost, but the quality and transparency of the data available.
Spend-based reporting is useful for carbon accounting, but it does not always reflect the real impact of individual AI use. On a per-query basis, text-based AI is already relatively efficient.
On a per-query basis, modern AI is already relatively efficient. Indicative 2025–2026 estimates for a typical text query include:
Google Gemini: around 0.03 g CO₂e per text query;
OpenAI ChatGPT GPT-4o: around 0.13–0.19 g CO₂e per query;
Anthropic Claude: around 0.2–0.4 g CO₂e for standard models.
These figures vary depending on model size, query complexity, output length, data centre efficiency and electricity mix. However, for moderate business use, even tens of thousands of text prompts per month may only add up to a few kilograms of CO₂e per year.
This means the carbon cost of using AI in day-to-day business is often very small compared with larger emissions sources such as energy, transport, purchased goods or supply chain activity.
The primary concern regarding AI is not the use of Claude by a single employee to compose a document. Scale is the issue.
The adoption of AI is expanding at a rapid pace in households, public bodies, and businesses. Simultaneously, more sophisticated applications, including advanced AI agents, large-scale automation, complex reasoning, and video and image generation, may necessitate considerably more computing power than straightforward text prompts.
This implies that AI can be efficient at the individual inquiry level while still contributing to the global increase in electricity demand. The overall effect will be contingent upon the rate at which models become more efficient, the manner in which data centers are powered and cooled, the rate at which electricity grids decarbonise and the extent to which AI is employed to facilitate broader environmental advancement.
The good news is that the efficiency of AI is enhancing rapidly. The impact of individual queries is already being mitigated by advancements in model design, specialist processors, cooling systems, and renewable energy matching. The obstacle is to guarantee that the efficiency enhancements remain consistent with the increasing demand.
Businesses do not have to evade AI due to emissions concerns. AI has the potential to enhance efficiency, reduce duplication, and facilitate better decisions based on information in numerous instances. Nevertheless, it is imperative that organisations exercise caution when employing AI and consider its potential consequences when applicable. For a broader look at how organisations can balance AI adoption with sustainability objectives, see our article on AI in sustainable business.
Identify which tools are being used, who is using them and whether usage is limited to text prompts or includes more energy-intensive applications.
Where supplier specific data is unavailable, use a suitable spend-based factor as a transparent reporting proxy.
Request information on emissions, renewable energy use, data centre efficiency, water use and carbon reduction plans.
For most organisations, AI subscriptions will be a small part of the overall footprint. Carbon reduction efforts should still focus on the most material sources of emissions.
If emissions are estimated using spend-based data, be clear about that. If the impact is small and offset within an existing carbon reduction plan, explain it proportionately.
Anthropic’s Claude is a powerful and highly capable AI tool. For most businesses, the direct carbon impact of moderate use is likely to be small and easily accounted for within wider carbon reporting.
The main concern is transparency. Because Anthropic does not currently publish detailed organisational emissions data in the same way as Google or Microsoft, businesses may need to rely on spend-based emissions factors when calculating Scope 3 impacts.
For a £10,000 annual spend, this could equate to around 1.177 tonnes CO₂e, with an example offset cost of just £17.66 per year at £15 per tonne. That figure is not large, but it is still worth measuring because credible sustainability reporting depends on clear boundaries, consistent methodology and honest communication.
The responsible position is not to reject AI. It is to use it thoughtfully, measure its impact where possible, ask providers for better transparency and ensure AI supports wider environmental progress rather than distracting from it.
Tunley Environmental is a science-based environmental consultancy, helping organisations measure, verify and communicate their environmental impact through science-based, impact-driven sustainability solutions. For businesses adopting AI, the right approach is to understand the data, keep the impact in proportion and build credible reporting around the tools they use.