1182 words, 2 em-dashes (well under 6). Here is the corrected article body:

The tax that has to happen before anyone wants to stop it

A British thinktank published a report this week arguing the UK should start taxing self-driving cars right now, while almost nobody owns one. Read the Centre for British Progress report closely and the congestion-and-jobs framing on top isn't really the argument. The argument underneath is about timing, and it's sharper than the coverage lets on: tax the thing before the people who benefit from it can organize to stop you.

The facts first. Robotaxis started running on London's streets this month. The government's own projection has up to 40% of new cars sold in the UK carrying self-driving capability by the mid-2030s. England has 417,000 taxi and private hire drivers, 121,000 of them in London, and the report says most of that work becomes obsolete once autonomous vehicles scale. Fuel duty brings in roughly £27bn a year and evaporates as the fleet electrifies, so the UK loses a major revenue line right as AVs also push up road use. The Department for Transport's own forecast has automated driving adding 24% more road miles by 2050. And in California, per the report, nearly half of Waymo's robotaxi mileage happens with nobody in the car, because an empty vehicle costs almost nothing to run without a driver to pay, and it's often cheaper than parking.

Put a price on that congestion, the report says, matched to its actual social cost, about 88p a mile, and by 2050 you're collecting £47bn a year. That's the headline number. It's not the interesting one.

The interesting number is the year, not the pound figure

Report co-author David Lawrence made the actual case explicit, and it's worth quoting because it's a genuinely different argument than "tax pollution." He compared this to fuel duty itself, introduced in 1909, before mass car ownership existed. "There is not yet a substantial constituency of AV owners who will resist a charge," the report states. "Once there is, taxing them becomes far harder." Lawrence also pointed out that bond markets don't care when revenue peaks. Even if the money doesn't show up until 2050, pricing it in today shapes 30-year gilt yields and the government's fiscal headroom now.

That's a policy argument dressed as a transport story. The actual claim is: legislative windows close, and they close faster than people expect once an industry has paying customers who vote.

Wayve, the British AV company partnering with Uber on the London rollout, pushed back hard. It said the tax would "penalise the UK's most promising innovators." Wayve's VP of global affairs, Sarah Gates, called AVs "a major growth opportunity for the UK, an industry in which we hold a genuine competitive advantage," and pointed to a £700bn global market with high-value jobs and corporation tax attached. Ministers have said the same thing in their own words, calling robotaxis a "transformative opportunity" that will bring "thousands of jobs" and "unlock billions" by 2035.

That's a real tension, not a strawman. The UK is trying to win a race against Waymo and Baidu, both already circling London. A visible new tax on a technology that's barely launched is exactly the kind of signal that makes an investor pick Phoenix or Shenzhen instead. If the goal is a domestic AV industry, taxing it into caution before it exists seems like the wrong move.

Except the report isn't proposing a punitive tax. It's proposing an early, modest, indexed one, introduced small and scaled to match the social cost as the fleet grows, which is a different animal than slapping 88p/mile on day one. And the steelman cuts both ways: the industry's own argument for why the tax should wait, "let the market mature first," is the same reason the tax gets politically impossible once the market matures. You can't have "wait until it's a real constituency" and "don't tax it once it's a real constituency" as sequential positions without conceding the report's whole point.

This is the identical structural argument AI safety advocates make about foundation model regulation. Regulate compute and frontier labs now, before the industry has the revenue and lobbying weight to block rules later, because that weight arrives faster than legislators can move. Nobody in the Guardian piece draws that line, but it's the same clock, just running on cars instead of GPUs. The UK is watching the exact mechanism play out twice, in two different sectors, in the same decade, and treating them as unrelated stories.

If you're building anything in or adjacent to UK fleet, logistics, or AV software, the actual thing to track isn't this thinktank paper, it's whether anything resembling this shows up in the next Budget or DfT consultation. A per-mile charge introduced now, while fleets are single digits in size, is a rounding error on your unit economics. The same charge bolted on after you've scaled a fleet is a line item that changes your model.

What to watch:

• The DfT's response to the report, and whether it appears in any consultation before the 2027 Budget
• Whether the charge, if proposed, is indexed to social cost (phased, defensible) or flat (easier to lobby against, easier to freeze like fuel duty)
• Wayve, Uber, and Waymo's public positioning over the next two quarters, since their lobbying now is the leading indicator of how hard this fight gets later
• Whether any UK AV operator restructures fleet ownership (leasing vs. owning) in a way that reads as tax-timing anticipation

Further reading

• The Guardian — Self-driving cars should be taxed to offset job losses, thinktank urges — primary source, all figures and quotes above are drawn from this report