On 16 September 2026 the Bundestag published Drucksache 21/8037, a Kleine Anfrage from the AfD parliamentary group. It is titled 'Nutzung von Abwärme aus Rechenzentren – Wirtschaftlichkeit, Verhältnismäßigkeit staatlicher Eingriffe und Auswirkungen auf den Rechenzentrumsstandort Deutschland'. It asks the federal government what CO2 savings the data-centre provisions of the Energy Efficiency Act (EnEfG) have delivered since the law took effect in November 2023. It also asks about cost per tonne avoided, operator compliance costs and effects on Germany as a location. The parliamentary service's summary says it covers §§ 11-17 EnEfG.
The AfD is not an obvious ally for a pro-innovation publication, and most of its energy politics deserves scepticism. But a parliamentary question is not wrong because of who files it. This one asks what any regulator should be able to answer.
The strongest case for the rules
The case for mandatory waste-heat reuse is real. Data centres turn nearly every kilowatt-hour they draw into heat. Germany's heat supply is one of the harder parts of its decarbonisation, and district heating networks need low-carbon sources. A statutory duty forces operators and municipalities to talk to each other, which voluntary schemes have often failed to do. Standing rules also give investors predictability.
What the law requires today
Under § 11 EnEfG, data centres starting operation from 1 July 2026 must reuse at least 10% of their waste heat, rising to 15% from 1 July 2027 and 20% from 1 July 2028. New facilities must reach a power usage effectiveness (PUE) of 1.2 or better. Existing sites must reach 1.5 by July 2027 and 1.3 by July 2030. Operators must also cover their electricity demand with 100% renewable power from 1 January 2027, up from 50% since 2024.
The technical problem is that data-centre heat is low-temperature. Using it usually needs heat pumps, extra electricity and a nearby network willing to take it. The AfD's question about additional technical and energetic effort is fair. Whether a percentage quota reduces emissions net of that effort depends on the site, and a uniform national target cannot capture it.
The cabinet has already conceded the point
The government appears to accept part of this. On 24 June 2026 the cabinet adopted a draft to speed up implementation of the EU Energy Efficiency Directive. According to the law firm CMS, the draft:
- extends the time to meet waste-heat requirements from two years to four;
- recognises internal heat use, such as heating offices, as reuse;
- exempts operators where no technically and economically reasonable connection to a heat network exists, verified by a cost-benefit analysis;
- moves the 100% renewable-electricity deadline from 2027 to 1 January 2030, with 50% remaining until then.
The draft is still before the Bundestag, and the Verband kommunaler Unternehmen reportedly expects it to take effect around the end of November. So the AfD filed its questions about the old rules while the coalition was already rewriting them. The inquiry is partly political theatre. It is also a request for the government to show its evidence.
Why the evidence question matters
The Kleine Anfrage has 26 questions, and the most useful ones are simple. What did a tonne of CO2 avoided through waste heat cost? How many operators applied for exemptions or hardship relief? How do the rules compare with those in France, Poland, Czechia, Slovakia and Hungary, which compete with Germany for hyperscale investment? The government drafted a quota in 2023 without, as far as the public record shows, publishing per-site economics. It is now replacing part of that quota with a case-by-case cost-benefit test. That is a tacit admission that the economics were not settled beforehand.
The inquiry also asks about projected power demand from AI data centres. That is where the tension is sharpest. Germany wants AI compute on its own soil, and its own data-centre strategy says so. Compute is built where power is available, connections are fast and the rules are stable. A regime that is tightened, delayed and litigated over will lose out to a neighbour with a simpler one.
A proportionate way forward
Proportionate regulation here would have four features:
- Keep the cost-benefit test, and publish the results. If operators must model heat-network connections, aggregated outcomes should be public. That would show whether the exemption becomes the rule.
- Tie renewable-power rules to measurable outcomes. Moving from 2027 to 2030 lowers the near-term burden. But an annual balance of purchased power says little about hourly carbon intensity. Better matching would be a stronger climate tool than a fast deadline.
- Put heat-network planning on the municipal side. Operators cannot reuse heat that no network can take. Municipal heat planning is the binding constraint, and the operator duty cannot fix it.
- Answer the parliamentary questions in full. If the savings are large and the costs modest, the answer helps the law's defenders. If not, better to learn it now.
The risk in this debate is that each side hears only the other's motives. Climate advocates may read any question about cost as sabotage. Sceptics may treat any duty as overreach. A duty that cannot state its cost per tonne avoided is hard to defend either way. Germany can keep an ambitious efficiency law and still be honest about what it costs. The government's reply to Drucksache 21/8037 will be an early test of whether it is willing to be.