Michael Dunne spent the segment on numbers, and the number that should stop a room is not the market share. It is the calendar.
By his account, Chinese automakers went from effectively no presence in Europe to exporting over a million cars there in a single year, roughly a tenth of the market. He puts Chinese models at about 20% of sales in the UK and in Mexico, and about 35% in Australia. Volkswagen is planning to close German factories for the first time since the Second World War. Honda, he says, has taken unprecedented losses across key Asian markets under pricing pressure.
Dunne is the CEO of Dunne Insights and the author of a forthcoming book called Car Wars, and he had a Wall Street Journal opinion piece running the same week, so read all of the above as an informed practitioner's characterization rather than as an audit. Weigh it accordingly. The shape of the curve is the part worth arguing with, and the shape is not in dispute.
The lesson is not the one in the headline
His framing of the method is the line people will remember: "the Chinese way is not to attack militarily unless they absolutely have to. But they'd prefer to work industries and countries from the inside out."
We are not a geopolitics shop and this is not a geopolitics post. We take no position here on trade policy, and nothing below depends on you holding one. There is an operating lesson buried in the least quotable thing Dunne said, and it transfers to a business of any size in any industry. It arrived in the last ninety seconds of the interview, and it is the reason the rest of it matters.
How fast it actually moved
Take his European figure at face value for a moment. Near zero to a tenth of a continent's market, inside a year.
Nothing in an incumbent's planning apparatus is built to see that. Annual planning cycles assume the future is the recent past plus a percentage. Category shifts do not arrive as a percentage. They arrive as a curve that looks like noise, then like an anomaly, then like an emergency, and the interval between the second and third of those is much shorter than the interval between planning cycles.
Dunne's warning to anyone expecting a slow ramp is worth taking seriously precisely because the slow ramp did happen once. The Japanese entry into the American market in the 1970s and the Korean entry in the 1990s were both gradual enough to be argued about for years before anyone had to act. That history is now a liability, because it trained an entire industry to recognize a competitive threat by its pace. A threat moving at a different speed does not register as the same category of event.
The transferable version, and it has nothing to do with cars: every incumbent's early-warning system is calibrated to the last thing that nearly killed it. That calibration is the most expensive assumption on the books, and nobody carries it as a line item.
The customer is not the problem
Here is where the argument has to be fair, because this is the part most versions of it get wrong.
Dunne notes that American consumers are looking at an average new vehicle price near $50,000. Against that, a cheap, well-equipped, feature-dense alternative is not a temptation. It is arithmetic. A household choosing the cheaper car that does more is not making an error, and it is not failing a patriotic test. It is doing exactly what every business in the world says it wants its customers to do, which is respond to value.
This matters because of what it rules out. Any strategy that requires your customer to act against their own interest is not a strategy. It is a wish with a budget attached. The moment a company's plan depends on buyers continuing to pay more for less out of loyalty, habit, or obligation, that company has stopped competing and started hoping. Loyalty is real, and it is worth a premium, and the premium is smaller than the people relying on it believe.
Dunne's own argument against the cheap car is not that the customer is wrong. It is that the price does not include everything. He lists three costs that sit outside the sticker: the collapse of domestic manufacturing and the supply-chain employment attached to it; the intelligence exposure created by connected vehicles streaming location, telemetry and personal data out of an open society into a closed one that bans Google, YouTube, Instagram and foreign media within its own borders; and deepening dependence on a strategic rival that already controls battery materials and specialized chips.
Those are real arguments and they belong in a policy conversation. Notice, though, that not one of them is an argument a dealer can make to a customer on a Tuesday afternoon, and none of them makes an expensive product competitive. They are reasons for a country to act. They are not a plan for a company.
What a tariff actually is
Asked whether the United States can simply block Chinese vehicles or level the field with tariffs, Dunne gave the honest answer: trade barriers buy essential time. He floats managed access, a hard quota somewhere in the range of 200,000 to 400,000 vehicles a year, as a way to let competition in without letting it take the floor out.
Then he says the thing the rest of the segment is really about. Protectionism alone will not solve Detroit's internal problems. He names legacy UAW labor agreements and thick layers of executive management overhead as the costs that remain regardless of what happens at the border.
Both halves of that are worth sitting with, including the uncomfortable one. A labor agreement is a contract that was negotiated by two parties and can only be changed by two parties, which makes it a genuinely different kind of cost from an org chart that grew four layers deep because nobody ever removed one. Only one of those two can be fixed unilaterally, and it is not the one that usually gets named first in public.
Set the specifics aside. The structural point is the one every business should copy down.
A tariff is not a moat. It is a runway.
Protection does not make the protected company competitive. It converts a competitive problem into a scheduling problem. It says: you now have an interval before this becomes fatal. What you do inside the interval is the entire question, and the interval is the only thing you actually purchased.
There are two ways to spend it. You can use the time to fix the cost structure, the product, and the operating model that made you vulnerable, and come out the far side able to win without the protection. Or you can use the time to keep doing exactly what you were doing, at exactly the cost you were doing it, and arrive at the end of the runway with the same problem, less market share, and a competitor who spent the same interval compounding.
The second is not a decision anybody makes. It is what happens when nobody makes the first one. Relief feels like resolution, which is the trap. The pressure comes off, the emergency meetings stop, the hard project loses its sponsor, and the calendar keeps running anyway.
The part protection cannot reach
Every business has some version of the tariff, and most of them have never named it.
It might be a franchise agreement, an exclusive territory, a licensing requirement, a compliance barrier that keeps casual entrants out, a long-tenured customer base with high switching costs, a local search position that took eight years to build, a contract with two years left on it, or simply the fact that the obvious competitor has not gotten around to you yet. Every one of those is doing real work. Every one of them is also a clock, and the clock is running whether or not anyone has looked at it.
The thing that makes this hard is that protection is invisible while it holds. A business shielded by any of the above reads its own margins as evidence of performance, because the margins are genuinely there. You cannot tell from inside the building which part of your profitability is skill and which part is shelter. The two look identical on the income statement and behave completely differently when conditions change.
That is the diagnostic worth running, and almost nobody runs it: how much of the current margin survives if the protection goes away? Not a scenario planning exercise. A number.
The crown jewels are an integration problem
Dunne's constructive argument deserves as much attention as his alarming one, and it gets a fraction of it.
His observation is that the founders of the Chinese companies doing the disrupting are not confused about who is best. Asked who they admire and respect, he says the answer is Tesla, and that the founders of Xiaomi, Xpeng and BYD "know Tesla's number one." He argues that America holds what he calls the crown jewels of autonomy and mobility software across Tesla, Waymo, Zoox and Nvidia, and that the future runs through a partnership between West Coast software and Midwest manufacturing, because software alone does not build a vehicle and a factory alone does not build an autonomous one.
Read that carefully, because it is not an innovation argument. Both halves already exist. They are world class. They are in the same country, in many cases in the same tax jurisdiction, and frequently in the same supply chain.
What is missing is that they do not share an operating model. Different clock speeds, different tolerance for failure, different definitions of done, different release cadence, different vocabulary for the same objects. Software ships weekly and treats a rollback as routine. Manufacturing ships annually and treats a recall as a catastrophe. Neither is wrong. They are simply two disciplines that have never been made to run on one process, and the cost of that gap is not a technology cost.
This is the failure mode we see constantly at a much smaller scale, and it looks the same at every size. An organization has the capability. It has bought the tooling. It has hired the people. And the capability sits in two departments that do not share a definition, a handoff, or a number they are both measured on, so what should compound instead leaks at the seam. The integration gap is almost never a gap in what a company can do. It is a gap in whether the parts of it agree on what done means.
Nobody is protected from that one. There is no tariff on organizational friction.
The playbook
DMAIC at the strategic layer, disciplined delivery underneath. Six moves for anyone currently operating on borrowed time, which is very likely everyone reading this.
Name your protection out loud, and put a date on it. Write down every structural thing currently keeping a competitor off your revenue: contracts, licenses, territories, switching costs, incumbency, inertia. For each one, write the year it plausibly stops working. You will not be right about the dates. Being roughly right is enough to change what gets funded, and an unnamed advantage cannot be defended because nobody knows it is there.
Measure cost to serve, not price. Price is what the market lets you charge and it is not yours to control. Cost to serve is yours entirely, and most organizations cannot state theirs per unit, per customer, or per job with any confidence. When a lower-cost competitor arrives, the company that already knows its own number gets to make decisions and the company that does not gets to react.
Separate the structural costs from the negotiated ones. Some costs can be changed by deciding to change them. Others require a counterparty, a contract cycle, or a regulator. Sort every major cost line into those two buckets. It is a dull afternoon and it produces the only honest picture of your actual maneuvering room. Most organizations discover the changeable pile is larger than the story they have been telling themselves, and that the story was more comfortable.
Assume the challenger can lose money longer than you can. Dunne's account of the mechanic is state-backed players running at zero margin at home and using that footing to undercut abroad. Strip out the geopolitics and this is just the funded-competitor problem, and it lands in every industry the moment somebody arrives with cheaper capital and more patience. You cannot win a losses-endurance contest against a player with a larger balance sheet and a longer time horizon. The only exits are a cost structure that does not require the contest, or a position they cannot buy their way into.
Close the integration gap before you buy anything else. If two parts of the business each hold half of a capability and neither can ship it alone, no purchase fixes that. One shared definition of done, one handoff with an owner on both sides, one number they are both measured on. That work is unglamorous, it has no vendor, nobody gets promoted for it, and it is usually worth more than the next tool.
Run the Monday test. Ask what you would do if the protection disappeared on Monday morning. Write the list. Then do the cheapest third of it now, while you still have the interval and the choice. The expensive two thirds will still be there if the clock runs out, and by then you will be doing them under pressure, at worse prices, with fewer options and someone else setting the pace.
Our position
Dunne closed with a sentence that is about industrial policy and reads like a note to management: "We have to come to terms with the reality that competition is global, and we cannot coast on former glories forever."
Coasting is the right word, and it is worth being precise about it, because coasting is not laziness. Coasting is what a well-run organization looks like when the thing that made it successful is no longer the thing keeping it successful, and nobody has noticed because the numbers are still fine. The people are working hard. The meetings happen. The reports render. That is exactly why it is difficult to see from inside, and exactly why it is easy to see from outside about eighteen months too late.
Whatever happens at the border is not a decision available to most of the people reading this. What is available is the interval. Tariffs, quotas, contracts, franchise protections and switching costs are all the same instrument wearing different clothes: they buy you time, they charge you nothing for it up front, and they present the invoice at the end regardless of what you did with it.
Time is the only advantage that gets spent whether or not you use it.
Reporting by CBT News, "Michael Dunne warns U.S. auto industry cannot afford to ignore China", Inside Automotive (August 7, 2026). All figures and quotes are Michael Dunne's, given in that interview. Dunne is CEO of Dunne Insights and author of the forthcoming Car Wars: How China Seized the Auto Industry and How America Can Win It Back. We have not independently verified the market-share figures.