Do chips lead
connectors?

Sales, orders and backlog are different signals

Sales / billings: business shipped or recognized under the reporting series’ definitions. Bookings: new orders accepted during a period, subject to the treatment of cancellations and changes. Backlog: orders still awaiting fulfillment. Book-to-bill compares bookings with billings over the same stated measurement window.

A ratio above 1 means incoming order value exceeded billed value. It does not by itself establish stronger end demand: the ratio could rise because shipments fell. Read it alongside absolute order growth, cancellations, delivery schedules and lead times.

Semiconductor sales are not semiconductor orders. WSTS’s public historical series covers revenue and moving averages. And SEMI’s familiar historical book-to-bill reports concern semiconductor manufacturing equipment, a separate market. A fab-equipment order should not quietly become a proxy for chips ordered by an electronics assembler. [5] [6]

Six hypotheses worth watching

1. Chips may lead connector shipments—but not by a fixed interval

Both components respond to equipment demand, but purchasing schedules and supply constraints can differ. A chip shipment might reflect an expanding production plan before the corresponding connector revenue appears. It might instead replenish inventory. The test is whether semiconductor activity improves forecasts across several cycles, with leads allowed in either direction.

2. Connector orders may be the earlier warning

If shipments are supported by old backlog, current sales can flatter the outlook. Weakening bookings may reveal a change sooner. The 2023 account is consistent with that idea, but does not establish that connector orders always turn before semiconductor sales. We should compare both indicators on equal terms.

3. The right chip category may matter more than the total

In 2023, automotive IC sales grew 23.7% despite the decline in overall semiconductor sales. [2] That divergence invites a better question: would automotive semiconductor activity have been more useful to an automotive connector supplier than the global total? Analog and microcontrollers versus industrial connectors offer another pairing to investigate.

4. A dollar of chip growth may contain very different information

Revenue combines units, selling prices and product mix. SIA’s initial 2024 year-end release reported memory revenue growth of 78.9%. [4] That figure alone cannot tell us how much additional equipment was built or how many connectors it required. We would want to separate price and mix effects from physical activity before treating chip revenue as a connector demand signal.

5. AI may strengthen selected connections while weakening the aggregate comparison

Our hypothesis: systems with higher compute density and power requirements could increase the value of selected high-speed, power and optical interconnects without lifting every connector category. More expensive silicon does not imply a proportional increase in connector count. The useful comparison may be interconnect content per system, rather than total chip dollars.

6. Inventory may change the apparent lead

When purchasers build safety stock, orders can grow faster than consumption. When stock is drawn down, orders can fall while equipment production continues. A relationship that appears strong during a shortage could reverse or weaken during a correction. The timing itself may be a signal about inventory conditions.

When the signals disagree

The following are hypothetical reading scenarios, not descriptions of today’s market. Each suggests a follow-up question, rather than a buy, sell or capacity recommendation.

Three relationships to put under the microscope

We would examine semiconductor sales against future connector orders; semiconductor sales against future connector sales; and connector orders against their own future shipments. The third comparison is essential. If a supplier’s order intake already carries the signal, does semiconductor data add anything useful?

A credible test would use monthly growth rates across multiple cycles, compare matching end markets where possible, and test leads in both directions. It would account for currencies, prices, inventory conditions and revisions. SIA’s monthly releases use three-month moving averages, so series need comparable treatment; smoothing can alter the apparent timing. [2]

The dates of the information matter as much as the dates of the sales. A useful test would use only the releases available when a decision was made, record their publication delays, and keep later revisions separate. SIA’s 2024 year-end release, for example, now notes that the original $627.6 billion total was subsequently revised to $630.5 billion. A revised history may be more accurate without having been available to the person making the original forecast. [4]

There is also a data gap to close. WSTS offers monthly semiconductor revenue and moving averages; the cited public connector interviews offer snapshots, not a matched monthly order-and-billings dataset. We have not assembled that comparable connector series. Until then, the proposed lead remains a hypothesis, and a supplier’s own bookings, cancellations and end-market mix are essential checks. [1] [3] [5]

Most importantly, any apparent advantage needs to survive testing on later periods that were not used to choose the relationship. Two rising revenue lines can look convincing without providing useful advance warning.

Follow the signal back to its source

Our market history provides published connector sales benchmarks. The broader Trends & Analytics page follows news and patent activity, while Industry Outlook explores the technologies behind those records. These are complementary signals: a burst of product announcements is evidence of supplier activity, not a measure of orders or revenue.

This article connects those views through a market-cycle question. It does not introduce a live semiconductor-versus-connector tracker or claim that our current historical coverage can establish a statistical lead.

The next question may be even more revealing: which semiconductor signal would change your mind?