Editorial Article

A Camera Has No Regulator. It Has a Customs Line.

Updated November 20, 2026

Canon’s additional tariff cost in 2025 was ¥45.4 billion. It recovered ¥41.2 billion of that in price. A 90.75 per cent pass-through, on a cost line the company did not choose and could not engineer out of the product.

In the same twelve months, the average camera got 3.93 per cent cheaper.

Both facts belong to one company and one fiscal year, and together they explain the oddest structural feature of this market. Nothing in a camera is compulsory except the line at the border — so the line at the border is where the price actually gets set.

The measurement first. The Camera & Imaging Products Association’s full-year 2025 series records 9,438,876 bodies for ¥880,574.519 million: units up 11.2 per cent, value up 6.8 per cent. Those two figures cannot both hold unless the average camera got cheaper, and it did — average value per body fell from ¥97,113 to ¥93,292, down 3.93 per cent. The class detail, and the full perimeter the market is sized on, are set out in the Global Digital Camera Market Report, Size, Share and Forecast 2026–2033.

What a camera is not required to do

There is no performance regulator for consumer imaging. No agency sets a minimum resolution, a low-light threshold, a service life, a failure-reporting duty or a price of any kind. Conformity regimes do exist — electrical safety, radio equipment, electromagnetic emissions, battery chemistry — but every one of them is paperwork about the product not hurting anybody. None of it is about the product working.

That distinction has a price consequence, and the contrast with a regulated category is the cleanest way to see it. In healthcare security, the obligations are documents: a workplace violence programme, a training record, a worksite analysis, a reporting line. A hospital cannot elect to skip them, which is precisely why its security spend behaves differently from a consumer’s camera spend — the argument in our healthcare security systems report rests on exactly that. A compliance floor is a price floor. Consumer imaging has neither.

Which is why 2025 could happen at all. Mirrorless average value fell 8.05 per cent to ¥110,713 while mirrorless units rose 12.5 per cent to 6,311,054. Fixed-lens average value rose 14.90 per cent to ¥62,152 on units up 29.6 per cent. DSLR shipped 690,911 bodies against 997,608, at ¥43,995, down 7.16 per cent. Nothing broke in any of those three directions, because nothing was obliged to hold. Price in this market is a mix decision, not a compliance decision — and a mix decision can be reversed in a single product cycle.

The one line that cannot be engineered around

The tariff is different. It attaches to the shipment, not the specification, and Canon disclosed precisely what it did with it. In FY2025: ¥45.4 billion of additional tariff cost, ¥41.2 billion recovered through pricing, a net negative impact of ¥4.2 billion. For FY2026 the company plans on ¥96.1 billion of tariff cost, ¥84.6 billion of price recovery — 88.03 per cent — and a net negative of ¥11.5 billion. The arithmetic is in Canon’s own FY2025 analyst meeting Q&A.

Two things follow, and the second one is easy to miss. First, a discretionary product can pass a duty through at roughly nine-tenths, which is why the tariff arrived in this market as a price event rather than a volume event — units grew 11.2 per cent in the year the cost landed. Second, the residual loss more than doubles in 2026 even though the pass-through ratio slips only 2.72 points, because the cost base it applies to doubles. A pass-through ratio is not a defence. It is a ratio, and it operates on whatever number it is given.

Why the ratio is so high is the part that matters for anyone planning against this market, and it is not about tariffs at all. Ten point six million interchangeable lenses shipped in 2025 — 1.514 per body. The body is not the transaction; it is the key to one. A duty on a body is therefore a duty on entry to a decade-long annuity, and an annuity is priced where the buyer is least elastic. That is why a camera body can be discounted 8.05 per cent without damage and still carry a tariff increase — the lens does the earning, and the mount is what makes the lens a captive purchase.

The same mechanism explains the strategy of the businesses that do not have a mount. Insta360 spent RMB 762 million in 2025 — roughly 80 per cent of net profit — on two drones, a gimbal camera, a wireless microphone system and three custom chips, on revenue of RMB 9.74 billion, up 74.76 per cent. With no mount to sell into, the category has to be bought. GoPro took the other route and converted hardware into a subscription base: USD 106 million of subscription and service revenue, 2.36 million subscribers, on total revenue of USD 652 million, down 19 per cent. Neither is a camera-maker in the sense the shipment data measures.

Then the ground moved, and the cost did not

On 20 February 2026 the US Supreme Court held, in Learning Resources, Inc. v. Trump and Trump v. V.O.S. Selections, Inc., that the International Emergency Economic Powers Act does not authorise tariffs — affirming the invalidation of both sets of IEEPA duties. The Congressional Research Service’s legal sidebar of 23 February 2026 sets out the holding and the six concurring and dissenting opinions that came with it.

That same day, a temporary 10 per cent surcharge was imposed under Section 122 of the Trade Act of 1974. It was the first use of that authority since Congress enacted it in 1975, and CRS’s April 2026 note on Section 122 is blunt about why the authority is unusual: it permits duties of up to 15 per cent for up to 150 days, and only against balance-of-payments deficits. Its model is President Nixon’s 1971 surcharge, which ran 127 days. Tariff authorities under Section 232 and Section 301 were unaffected by the ruling.

So the legal basis under Canon’s cost line changed on a single day in February 2026 — and the cost line did not. Canon’s 2026 plan still carries ¥96.1 billion. That is the operative fact for anyone building a forecast on this market: the authority is contestable, temporary and statutorily expiring, while the cost in the plan is none of those things.

It is worth being precise about the weakness, because it cuts both ways. Section 122 rests on a Bretton Woods vocabulary — fundamental international payments problems, large and serious balance-of-payments deficits — in an economy that has not defined its trade position that way in half a century. A duty resting on a temporary authority whose statutory footing is arguable is not a price anyone can put in a five-year plan. It is a price with a review date.

The second cost push, in the same division

On the same analyst call, Canon guided to roughly ¥6–7 billion of additional DRAM cost in 2026, about one-third of it landing in Printing and two-thirds in Imaging, with a stated procurement policy of buying everything it can obtain, and flagged rare-earth supply as a risk specifically to Imaging.

That matters for how the 2025 price fall should be read. The average camera got 3.93 per cent cheaper in a year when a tariff was being absorbed inside a portfolio-wide ¥41.2 billion price recovery, and when memory had not yet bound. In 2026 both of those conditions are gone. The forecast assumes the price decline stops; two unregulated input costs are pushing in the other direction, both concentrated in the division that builds the cameras, neither of them a demand signal.

What this does to the 4.76 per cent

The market is sized at USD 7,787.8 million in 2025, reaching USD 11,301.4 million by 2033 at 4.76 per cent, with volume from 14.44 million to 18.51 million units at 3.15 per cent. The value number does not require camera prices to recover. It requires them to stop falling.

On the association perimeter, a repeat of the 2025 price decline against the model’s 3.15 per cent unit growth leaves value about 0.9 per cent lower a year — 0.9607 multiplied by 1.0315 is 0.9910. So the 4.76 per cent headline is not carried by bodies at all. It is carried by units, and by the layer outside the association: hardware CIPA does not count rises from 24.62 to 39.61 per cent of the pool, and supplies 72.85 per cent of all incremental value to 2033. That perimeter is set out line by line, alongside the narrower category-split model in our category-split report.

The practical translation: value growth in this market is a category-creation story and a unit story. It is not a price story, and it cannot be, in a product with no compliance floor and a live customs clock.

Three ways this breaks

Stated as falsifiers, not as caveats.

  1. The Section 122 window closes without a durable replacement. The authority lapses within its 150-day limit in the second half of 2026 unless Congress acts. If the tariff line falls away, the ¥84.6 billion of recovered price has to be given back or taken out of margin, prices resume falling, and 2025 stops being an anomaly and becomes the base case.
  2. A permanent basis is legislated, or Section 232 or 301 coverage reaches imaging. The ¥96.1 billion becomes structural, the 88 per cent pass-through becomes the de facto price floor, and value holds up while units soften. The first class to break in that scenario is fixed-lens, up 29.6 per cent on a 14.90 per cent price rise — the fastest-growing class with the thinnest annuity behind it.
  3. Memory normalises faster than guided. Imaging’s share of the DRAM line disappears, margins recover, and firms price more aggressively to defend share. That is the one path on which the price decline accelerates instead of reversing, and it requires nothing to go wrong at the border at all.

There is a fourth, and it is arithmetic rather than commercial: the perimeter is ours, not the association’s. If CIPA ever widens its membership to include the action and 360 vendors, the 72.85 per cent incremental-value claim dissolves without a single commercial fact changing.

What a supplier should plan against

Because nothing in the product is compulsory, no customer is obliged to buy capability. Capability has to be sold as a category, a subscription, or a mount — and the three carry different tariff mechanics.

  • If you own the mount, a tariff is a price event and the ratio is the thing to watch. Canon recovered 90.75 per cent in 2025 and plans 88.03 per cent in 2026; the slip matters more than the level, because it sits on a doubled base.
  • If you do not own the mount, the same tariff is a margin event. GoPro absorbed USD 20 million of tariff expense inside a year of negative USD 29 million adjusted EBITDA, after cutting USD 93 million of operating cost. There is no lens annuity to carry the recovery, and no price increase that a sealed body can defend.
  • If you are building a multi-year model, the July 2026 horizon is not a footnote. It is the scheduled moment at which the only compulsory cost in the product either becomes durable or disappears, and both outcomes move price by more than any product decision in the market.

The regulator this industry actually has is not a safety agency. It is a customs schedule and a lens mount. One of them is temporary by statute. The other is the most durable thing in consumer imaging, which is why the average camera got cheaper in a growth year and the industry did not blink.

Sources

Canon Inc., FY2025 results presentation, 29 January 2026, and FY2025 analyst meeting Q&A summary — tariff cost, price recovery, net impact, DRAM guidance and rare-earth commentary.

Congressional Research Service, “Supreme Court Rules Against Tariffs Imposed Under the International Emergency Economic Powers Act”, LSB11398, 23 February 2026.

Congressional Research Service, “Section 122 of the Trade Act of 1974”, IF13199, 9 April 2026.

Camera & Imaging Products Association, full-year 2025 camera and interchangeable-lens shipment series, reconciled to the yen in the Pheonix Research market report; GoPro, Inc. and Arashi Vision Inc. (Insta360) FY2025 company results, as reconciled in the same report.

Related Pheonix Research coverage: India Video Surveillance Market, where the tariff exposure and the compliance floor sit in a different place, and the Global Road Safety & Enforcement Systems Market, where the camera is procured against a statutory enforcement obligation.