mSAP Optical Module PCB: The 1.6T Bottleneck Nobody Plans For

Through September 2026, demand for 800G and 1.6T optical modules has continued to intensify, and PCB manufacturers have begun accelerating mSAP capacity expansion in response. One major supplier has been building optical module mSAP lines simultaneously in three locations, with planned investment in the range of 8 to 10 billion New Taiwan dollars. If the expansion proceeds on schedule, the associated capacity could generate monthly revenue of 2.5 to 3.0 billion New Taiwan dollars by the second half of 2027. What makes this notable is that the factor limiting expansion speed is no longer building space. It is the delivery lead time for laser drilling, LDI exposure and plating equipment.

Why PCB Capacity Becomes the First Constraint

AI data centers are rapidly pushing optical module specifications to 800G and 1.6T. Higher data rates are not simply a matter of faster chips. The loss, impedance behavior and routing density of the electrical channels inside the module all rise together.

In the 400G generation, conventional HDI could still cover a substantial share of requirements. Moving to 800G and especially 1.6T, line density increases further. The PCB must fit more high speed channels into a limited area while shortening the signal path between the driver, the optics and the host interface.

Conventional subtractive etching approaches its capability limit in line width, spacing and conductor profile control at that point, which is why mSAP has been adopted more widely in high end optical modules. Public industry information indicates that 800G and 1.6T modules commonly use 12 to 16 layer PCBs, with a considerable portion of those layers produced using mSAP. Requirements for conductor profile, copper surface roughness, impedance consistency and insertion loss have all tightened correspondingly.

This is one of the clearest differences between optical module boards and ordinary consumer electronics boards. Demand growth does not distribute evenly across all PCB capacity. It concentrates first on factories that already have fine line capability, high order HDI and stable mSAP volume production.

What Is Actually Scarce Is Not 0.075 mm Lines

The market often reduces mSAP to an ability to produce finer lines. Volume production difficulty extends well beyond that.

As lines become finer, the exposure and etch windows narrow, which raises requirements on conductor sidewall geometry, copper thickness uniformity and imaging precision. As microvia density increases, laser drilling, desmearing, via fill plating and layer-to-layer registration emerge as new yield variables. At the same time, a 1.6T module is a high speed signal device, so the PCB must not only produce the geometry but also control dielectric thickness, copper surface roughness and impedance variation, because a small manufacturing deviation translates directly into high frequency loss.

mSAP capacity therefore cannot be replicated quickly by adding ordinary PCB equipment. The laser drilling machines, LDI exposure systems and plating lines required for this class of expansion have seen lead times extend to nine months and in some cases close to a year.

Even when equipment arrives, commissioning, process ramp and customer qualification still have to be completed. Qualification is the longest step and the one least amenable to acceleration, because it requires demonstrating insertion loss and return loss performance on production hardware rather than on samples. This explains why PCB supply responds slowly when optical module demand rises sharply. The real bottleneck is the combination of equipment, process, yield and certification, not factory floor area.

PCB Capital Expenditure as an Early Signal

Compared with optical module manufacturers announcing orders, PCB companies’ capital spending is often more informative to watch.

An mSAP line takes a long time to move from equipment purchase to volume production. When a PCB company is willing to commit several billion in local currency to expansion ahead of confirmed demand, it generally indicates that visibility into future customer requirements is reasonably strong. Sustained increases in capital expenditure focused on optical module mSAP, with lines distributed across multiple regions, itself reflects how high speed optical communication orders are beginning to shape supply chain structure.

That expansion carries a second implication. High end optical module customers are concerned not only with technical capability but increasingly with regional manufacturing presence and stable volume delivery. Some customers, managing supply chain risk, expect PCB suppliers to operate production bases in more than one region. Competition in high end PCB therefore increasingly layers capital strength and global capacity layout on top of process capability.

This suggests that to gauge the health of CPO, 1.6T and eventually 3.2T optical module demand, watching optical chip and module orders alone may be insufficient. mSAP equipment purchasing, production line construction and customer qualification progress can appear earlier in the cycle and are worth tracking alongside order books.

Value Shifts Toward Manufacturing Precision

The more significant effect of high speed optical modules on the PCB industry may not be volume growth. It is a change in product structure.

Historically, PCB value correlated primarily with area, layer count and material grade. Entering the 1.6T generation, value concentrates increasingly on manufacturing precision: how fine the lines are, how uniform the plating is, how tight the impedance window is held, and how consistently those attributes repeat across production lots.

This changes what customers evaluate. A supplier that can produce a demonstration board at 0.075 mm lines is not the same as one that can deliver production volumes at that geometry with controlled insertion loss. The relevant evidence is process capability data across volume, measured impedance on production coupons and yield distribution rather than a best-case sample.

For engineering teams specifying high speed optical boards, that distinction should be reflected in the qualification plan. Measuring insertion loss and return loss on a coupon structure from production panels, rather than only on a first article, is what establishes that the process, not just the design, meets the requirement. Capability in PCB fabrication spanning fine line processing, high order HDI and controlled impedance is what makes telecommunications PCBA programs feasible at volume.

Material and Process Interaction

Fine line capability interacts with material selection in ways that are easy to underestimate during design.

Low loss laminates are generally softer and more sensitive to process conditions than standard FR-4. Their dimensional behavior during lamination differs, which affects registration budget. Copper foil roughness requirements for high frequency performance influence adhesion to the dielectric, which in turn constrains the surface preparation window before plating. These are not independent variables that can be optimized one at a time.

The practical consequence is that changing laminate to chase a lower loss figure can invalidate process parameters that were established for the previous material. Lamination cycles, drilling parameters, desmear chemistry and plating current profiles may all need revalidation. On an mSAP process, where the tolerance budget is already narrow, that revalidation is a significant undertaking.

This is why PCB manufacturing teams treating a material change as a procurement decision rather than a process requalification run into yield problems. A second qualified material is valuable, but only if it has been qualified through the actual process sequence rather than approved on datasheet parameters.

What Optical Module Designers Should Plan For

Several practical measures follow from the current supply environment.

First, engage the fabricator during stackup definition rather than after the layout is complete. On a 12 to 16 layer mSAP board, decisions about layer assignment, dielectric thickness and via structure determine whether the impedance target is achievable with the material set available.

Second, plan the qualification timeline explicitly. If mSAP line lead times reach nine months and qualification adds further months, a program that assumes boards will be available on demand will slip. Programs that treat fabrication capacity as a schedule dependency, rather than a procurement transaction, plan better.

Third, validate alternate materials during prototype rather than during production. Where a specific laminate is a single-source item, having a qualified alternative reduces exposure to allocation decisions made by suppliers serving many industries.

Fourth, verify impedance and loss on production coupons. On high speed boards, the difference between a design that meets insertion loss specifications and one that does not is frequently a manufacturing variation rather than a design error. Measuring it directly is faster than diagnosing it from system level test failures.

The Wider Pattern

The optical module case illustrates a pattern that now appears across multiple segments of the electronics industry. As data rates rise, the manufacturing precision required to deliver them rises faster than the volume of capacity needed. Capital can build factories, but it cannot compress the time required to establish process windows, ramp yields and complete customer qualification.

That is why quality management systems that document process capability at the lot level have become commercially significant rather than merely administrative. A supplier able to demonstrate measured impedance consistency and yield distribution across shipments is offering something that equipment lists cannot convey.

For programs still in development, working with a manufacturing partner capable of rapid PCBA prototyping during validation, and able to carry the validated process into small and medium volume production, shortens the interval between design freeze and deliverable hardware. In a market where the constraint is process maturity rather than factory space, that interval is the competitive variable that matters most.