Upstream +17%, Downstream +24%: The Middle Layer Is Being Squeezed

The whole industry grew 5.5 percent while both ends of the chain ran at three to four times that rate. Value did not disappear — it moved. In-vehicle MDVR makers sit exactly where it moved away from.

The official post-show industry report gives us something exhibition coverage rarely does: a financial description of where the value went. It is less exciting than a robot demo and considerably more useful for deciding what to build next.

The Numbers and the Floor Evidence

Total industry output reached about RMB 1.005 trillion in 2025, up 5.5 percent. Upstream core components grew 17 percent and downstream operations and maintenance services grew 24 percent. The report's own framing is that value is concentrating in technology innovation and continuous operations, with competition shifting from manufacturing toward scenario understanding, scaled delivery and long-term service.

The show floor told the same story in physical form. Post-show reporting from a&s noted that several leading video vendors no longer take oversized booths, preferring closed-door technical meetings and vertical industry salons. On the booths of traditional hardware makers, physical devices occupy less space while large screens run closed-loop case studies, model fine-tuning toolchains and operations platforms. The audience mix has shifted too: city operations, emergency management, transport, energy and telecom technical leads now make up a noticeably larger share.

That reporting describes a three-way split. Some hardware manufacturers hold the line on reliability, low power and domestic supply chains and become the ecosystem substrate. Some integrators have upgraded into industry digital-service providers. And a number of AI companies that started with a single algorithm were quickly absorbed or exited. Its conclusion is blunt: the dividend on generic hardware is fading, depth in vertical scenarios now determines who survives, and security is turning from a product industry into a scenario digital-services industry.

There is a neighbouring pool of demand worth noting. At the industry development forum, a researcher from the Ministry of Emergency Management's information institute put China's safety and emergency industry at RMB 2.6 trillion in 2025, up from 1.5 trillion in 2020 — over 11 percent annual growth — with the equipment market above 1.1 trillion. Five trends were named: intelligence, connectivity, lightweighting, platformisation and modularity.

What the Arithmetic Does and Does Not Allow

Seventeen and 24 percent are, respectively, 3.09 and 4.36 times the industry-wide 5.5 percent. Because the report does not disclose the weights of upstream, midstream and downstream, we cannot back out a specific growth rate for the middle. But we can say with certainty that when both ends run well above the average, the middle must run below it. That is a weighted-average constraint, not a guess.

In plain terms: the middle is still growing, just more slowly than the chain around it. Upstream takes the technology premium, downstream takes the continuing customer relationship, and the middle is left with manufacturing and delivery. The typical in-vehicle MDVR business — buying chips and cameras, assembling a host unit, selling to a fleet or integrator, finishing at handover — sits precisely there.

Our Take

Our first report series read the hall plan and concluded that the camera had been demoted to substrate. This is the same conclusion in financial form, and it is sharper: the problem is not that you stopped mattering, it is that growth no longer originates with you. Fewer physical devices and more case studies on a booth are simply value migration made visible in booth materials.

The tempting misreading is that everyone should now go make chips or run services. Upstream demands long R&D cycles and certification; downstream demands long-term operations, compliance and working capital. A mid-chain vendor may have neither. The real risk is a suspended position — no upstream barrier, no downstream relationship, only replaceable assembly capacity.

Four Directions and What Each Costs

• Upstream — imaging, modules, on-device compute, domestic boards: technology premium and reusability, but long cycles and a direct fight with the chip vendors that already supply you.

• Downstream — evidence custody, alert handling, insurance risk linkage: recurring revenue, but requires SLA, data compliance and capital to carry operations.

• Vertical depth — own one scenario such as hazardous goods, cold chain, school buses or construction vehicles: hard to commoditise and hardest to displace on price, but with a lower ceiling on scale.

• Stay in the middle — short-term shipment volume, at the cost of being compressed into contract manufacturing as unit prices fall.

Vertical depth is underrated. A 2.6 trillion yuan adjacent industry growing above 11 percent, plus the plan's task of cultivating emerging tracks, means downstream scenarios are still expanding. For a mid-chain vendor without the means to design silicon or fund a service arm, mastering one or two vertical workflows is the most realistic second curve.

Our Judgment

The danger in 2027 is not technological backwardness. It is a suspended position. Five questions expose it: does gross margin come from one-off price difference or continuing service; will the customer renew next year without being asked; can the unit be swapped for an equivalent spec; do you control the data, interfaces and model versions; do you understand the customer's workflow rather than just their equipment list. If three of the five cannot be answered, it is time to pick a position.

What to Take Away

• Explain price pressure as value migration, not 'fierce competition' — the problem is structural, not your rival.

• Run those five questions as an annual check on every product line.

• Never invert the upstream/downstream growth rates into a midstream figure; the weights were not published.

Sources: the CSPIA / New Silk Road Alliance industry report coverage of 11 November 2026, a&s Media's 10 November 2026 post-show analysis, and the 2026 China Security Industry Development Forum proceedings. The 3.09x and 4.36x ratios are our calculations; no midstream growth rate is inferred.