Connector Industry Roundup — October 2, 2026:
Lead Times Extend Again, a PHY Split for Single-Pair Ethernet & £300M of UK Aerospace Capacity

The first days of October 2026 gave connector buyers two market readings that point the same way, one architecture change that will quietly reduce the number of cables in a machine, and one very large capacity commitment from the aerospace sector. None of them is a connector launch. All four change what a connector specification has to account for over the next two years.

The market readings are the urgent ones. Two independent surveys covering September both describe demand that is holding up better than expected while supply gets tighter — a combination that historically shows up first as extended lead times on the parts nobody re-qualifies.

Two Surveys, One Direction: Demand Up, Availability Tightening

ECIA published the September 2026 Industry Pulse on 4 October, and the headline is an upside surprise: the overall component index average improved from 136.7 in August to 142.0 in September, beating expectations by 6.6 points. The overall end-market results stayed broadly flat between the two months but still beat expectations by 8.1 points. The survey covers six major component categories, six semiconductor subcategories and eight end markets.

ECIA chief analyst Dale Ford characterised the result as "a very pleasant upside, as every component segment and the end-market measure outperformed expectations from the August survey," adding that the numbers are consistent with measured Q2 financial results, with the outlook from leaders of major companies, and with anecdotal evidence from executives.

Four days earlier, the September IP&E market round-up drawing on Edgewater Research described the other half of the picture. Demand is strengthening across Industrial, Mil/Aero and AI/data-centre markets, bookings, point-of-sale and book-to-bill ratios are healthy, and visibility into calendar 2027 is building. But the same note reports that connector lead times are extending, pockets of shortages are emerging, and MLCC and fiber availability are becoming increasingly constrained.

The two documents are not in conflict; they describe the same market from opposite ends. A rising sentiment index with extending lead times is the signature of a supply chain running close to its capacity, where order intake improves faster than the ability to ship. For a buyer, the operative sentence in the Edgewater note is the one about lead times, because it applies to the parts that are hardest to change late: qualified circular connectors, custom cable assemblies, and anything with a plating or material specification tight enough to limit the approved supplier list.

NXP Splits the Ethernet PHY — and Changes the Harness

NXP has put the first industry-standard 10BASE-T1S Physical Medium Dependent transceivers into production (Bisinfotech, 30 September 2026). Two parts split the application: the TJA1410 for automotive, the TJF1410 for industrial and building automation.

The architectural move is the separation itself. Conventional Ethernet PHY ICs combine the digital control logic and the analogue transmission function in one device. NXP's parts implement only the PMD sub-layer, which lets the digital half be integrated directly into a microcontroller or an Ethernet switch. The company states that this dispersed design reduces both hardware cost and size, and enables a multidrop configuration that behaves like a CAN bus but carries standard Ethernet at 10 Mbps.

Why this belongs in a connector publication: multidrop changes the counting. A point-to-point fieldbus architecture needs a home run from every device back to the cabinet, so the connector count scales with the number of nodes and the cabinet fills with terminations. A multidrop single-pair bus replaces most of those runs with a trunk and taps. Two consequences follow for anyone specifying the physical layer.

First, the connector becomes a bus element rather than a cable end. Trunk-and-tap topologies put a premium on controlled impedance at every junction, because a stub that is electrically long relative to the data rate turns into a reflection. At 10 Mbps the margin is generous compared with gigabit links, but industrial installations add long cable runs, VFD-induced noise and imperfect shielding — the three conditions under which a marginal junction stops being marginal.

Second, the magnetics simplification is real but narrow. Reusing standard CAN FD common-mode chokes removes a specialised part from the BOM; it does not remove the need to specify the choke as part of the interface, and it does not change the ingress protection or mating-cycle requirements of the connector housing. In a washdown plant, the reason to use an IP67 M12 assembly is unchanged by what the PHY does. What changes is how many of them you need, and whether the network can tolerate a daisy-chained section instead of a star.

There is also a qualification angle. A part qualified to ASIL B and offered in an industrial variant gives a machine builder a single silicon family across two very different reliability regimes. That is convenient for engineering, and awkward for procurement, because the automotive variant's change-control discipline and the industrial variant's long-availability expectations rarely run on the same calendar.

£300 Million Into UK Aerospace and Defence Capacity

Rolls-Royce is investing £300 million in its UK manufacturing and engineering footprint (Bisinfotech, 30 September 2026), on top of more than £3 billion committed in the country since its transformation programme began in 2023. The allocation is unusually specific:

• Derby: more than £140 million for new manufacturing and engineering facilities, with upgrades expected to complete by 2028.
• Bristol: more than £90 million for the site described as the hub for UK military power and propulsion, with the facility programme completing by 2031.
• Inchinnan, Glasgow: £43 million for specialist machinery for new engine production.
• Rotherham: £19 million at the Advanced Blade Casting Facility, including £2 million from the South Yorkshire Mayoral Combined Authority.
• Employment: more than 3,500 jobs across the programme, including a large number of specialised engineers.

The connector read here is indirect and should be stated as such: this is a capacity announcement, not a procurement notice, and nothing in it specifies a connector standard. What it does signal is the shape of Mil/Aero demand through the end of the decade — new build and aftermarket support, spread across civil and defence programmes, with the defence half anchored at Bristol through 2031. Aerospace capacity expansions of this length run on qualification timelines measured in years, and the interconnection hardware on those platforms is chosen long before the building is finished.

For a connector supplier the practical implication is about documentary readiness rather than volume. Programmes of this duration tend to standardise on qualified product, which means the specification conversation is about test evidence, traceability and cross-reference to an existing qualified part, not about a new design. That is the same conclusion the market surveys point to from the other direction: when lead times extend, the part that ships is the part that is already on an approved list.

The Buyer's Read

• Treat the sentiment index as a lead-time warning, not a demand forecast. A rising index alongside extending lead times means capacity is the constraint. Place volume against schedules early, and put the escalation path in writing.
• Second-source the passives, not only the connector. The September note flags MLCC and fiber availability specifically. A connector with a three-week lead and a capacitor with a twenty-week lead is a twenty-week assembly.
• Re-check the harness topology before re-quoting a connector count. If a single-pair Ethernet migration is on the roadmap, the number and type of field connectors will change before the enclosure does.
• Keep qualification evidence current. Long aerospace programmes buy qualified product. Test reports and cross-references that are already on file are what convert a capacity expansion into an order.
• Do not read a PHY announcement as a connector standard. 10BASE-T1S changes media and topology requirements; the environmental and mating specifications of the connector remain a separate decision.

What connects these four items is that each one is upstream of the connector decision. A sentiment index, a lead-time trend, a PHY architecture and an aerospace capex programme all arrive before anyone draws a shell size on a panel. Buyers who read them at that stage get to influence the specification; buyers who meet them at the RFQ stage get to negotiate against a date that has already been set by someone else.