France Tried to Tax the Internet by the Gigabyte. The Internet Won.

On October 30, the French National Assembly passed a budget amendment by 11 votes to 10. Only 35 deputies were present. The government and the rapporteur of the Finance Committee opposed it. The amendment would tax internet data delivered to French users by volume. Content providers could pay €0.055 per GB. Its proposer, Laurent Lhardit, says France has about 154 million internet users, that roughly 200 billion GB passes through French ISPs each year, and that five major companies account for half of that traffic. He estimates €10 billion in annual revenue.

This is a legislative accident. Taxing digital giants has legitimacy. This instrument does not. It fails in engineering, distorts the economy, and rests on weak legal ground. The 11-to-10 vote and the thin attendance pushed it through its first threshold, not a serious policy consensus.

I. What the bill says and where it stands

The amendment does not tax at France’s international traffic entry points. It does not exempt local CDNs. It makes French ISPs tax collectors. They must count GBs by the source of each end user’s downstream traffic, then file and pay quarterly. Article 302 bis ZP defines the taxpayer as a “content provider.” Article 302 bis ZQ requires ISPs to identify the sending enterprise. Article 302 bis ZR uses the broader term “sender.”

The bill is not law. It is an amendment at the first-reading stage of the 2027 budget. It still needs a full National Assembly vote and Senate review. It can be changed or deleted. Calling it a “new tax category already approved by France” is inaccurate. A controversial amendment survived a poorly attended session. The government and the Finance Committee’s rapporteur opposed it.

II. Four technical failures

1. Encryption

Most internet traffic is encrypted. ISPs see packet sizes, directions, and rough destinations. They cannot reliably identify content type, application source, or commercial owner. Take a concrete case. A French user opens a news page. HTML comes from the site’s server. Images come from Cloudflare. Ads come from Google. Scripts come from another vendor. Fonts come from another CDN. The ISP sees an encrypted packet stream. It does not know which packet belongs to which company. It cannot split that stream into clean tax lines for different content providers. The bill tells ISPs to identify the sending enterprise. Internet routing does not carry tax invoices.

2. Shared infrastructure

Many users share an exit IP. Much content comes from local cache nodes. The line between content provider, distributor, and network intermediary is already blurred. The system would likely collapse into rough ASN aggregation. Whoever sends the most packets gets taxed. That attribution is crude. It also pulls many small and mid-sized networks into the tax net.

3. P2P and PCDN arbitrage

This is the weakest point. The internet lets users send data directly to each other. Large platforms can use users’ upstream bandwidth and hard drives to build PCDN networks. To the ISP, that upstream traffic is encrypted. The ISP cannot tell personal use from ordinary sharing from a platform dodging tax through distributed delivery.

The result is a regressive tax and a competition distortion. Giants with huge user bases and PCDN capacity can avoid or dilute the charge. French SMEs that rely on centralized delivery would pay. A tax sold as “make the giants pay” could end up unfair to domestic firms.

If lawmakers wanted to pursue this model, they have two theoretical options. Tax all users’ upstream traffic, which penalizes all internet use. Or force tax-monitoring software onto user devices to decrypt packets and measure traffic, with no access for those who refuse. Both would upend the open internet, privacy, and encryption.

4. The legal text contradicts itself

Article 302 bis ZP calls the taxpayer a “content provider.” Article 302 bis ZQ tells ISPs to identify the sending enterprise. Article 302 bis ZR uses the broader “sender.” The text assumes the sender is a content provider and a corporate taxpayer. The internet also carries person-to-person messages, non-commercial exchanges, public-institution traffic, and unidentifiable edge nodes. The bill does not say how to tax them. It assumes they do not exist.

III. Who actually pays

The proposer says Facebook and Instagram are free, so the tax cannot reach their users. That is a tax-incidence fallacy. Free does not mean costless. Free does not mean the burden stays put.

Netflix and Spotify can raise subscription prices. Meta and Google can lower service quality, load more ads, or charge advertisers more. The French Finance Committee warned that the burden could reach advertisers and subscribers. A Deloitte assessment of France’s digital services tax found about 55% of the burden fell on end consumers and 40% on suppliers.

A tax sold as “make the giants pay” could become a regressive consumption tax on French internet users. Video and gaming drive most traffic. Spending on those services does not fall in step with income. Low-income households can end up bearing a larger share.

IV. Legal and procedural risk

If the amendment survives, it faces legal challenges. It requires ISPs to inspect and classify traffic for tax purposes. That conflicts with EU net neutrality rules. It targets mainly foreign digital service providers. It could be challenged before the Court of Justice of the European Union as discriminatory or as state aid. France’s earlier digital services tax triggered US retaliation threats. A tax on traffic volume is easier to cast as discriminatory than a tax on revenue. Traffic source has no stable link to French revenue, profit, or user value.

The procedure is also weak. It passed 11 to 10 with 35 deputies present. The government and the Finance Committee’s rapporteur opposed it. France has no serious policy consensus behind it. It looks like a political gesture inserted into a budget bill, not a fully considered fiscal policy.

V. The politics are real. The tool is not.

The anxiety behind the bill is real. France and other European countries face digital tax-base erosion. Corporate income tax struggles to reach multinational platforms. Data traffic looks like a visible economic activity, so it becomes an attractive tax base. France has long used “cultural exception” and “digital sovereignty” to restrict and extract revenue from Anglo-American platforms. This bill continues that tradition.

Political intent does not rescue a bad instrument. You can argue that digital giants should pay more. You cannot assume the internet works like broadcasting. A tax that cannot be enforced, distorts competition, and rests on weak law looks like a domestic political gesture. Its main output may not be €10 billion. It may be legal work for multinational platforms and compliance costs for French SMEs.

VI. Better options

Criticizing this tax does not mean digital giants should pay nothing. Better paths include digital tax coordination under the OECD’s Pillar One and global minimum tax; digital services taxes based on profits, users, ad revenue, or subscriptions; stronger transfer pricing and permanent establishment rules; transparency on bandwidth interconnection and infrastructure investment; and public digital infrastructure and local content ecosystems. These are not perfect. They rest on identifiable economic activity and legal entities, not ISP tax inspection of encrypted packets.

If France wants fair network cost-sharing, it can consider neutral interconnection settlement, peering transparency, and public support for edge caching. Tax policy should change the technical architecture as little as possible. It should not punish encryption, force platforms to monitor users, or turn ISPs into tax police. Otherwise a fiscal measure becomes hidden regulation of the open internet.

Conclusion

The “network download tax” asks a fair question. How should large platforms contribute to network infrastructure and public finances? Its answer fails. It cannot be enforced. It distorts the economy. It rests on weak law.

The irony is sharp. A tax sold as fairness could hurt French SMEs, raise costs for French consumers, and give multinational platforms room for tax planning and litigation. It treats the internet as broadcasting, ISPs as tax collectors, encrypted traffic as a readable ledger, and GBs as an income statement.

If it is ever fully implemented, its legacy may not be a new tax model. It may be a case study in regulatory failure. Political will overwhelmed engineering reality. Fiscal anxiety replaced institutional design. A legitimate goal became a costly legislative accident.