Goldman Sachs has significantly revised its 2027/2028 optical module demand forecasts upward, citing supply chain constraints as the biggest bottleneck for shipments next year.

Goldman Sachs has significantly revised its 2027/2028 optical module demand forecasts upward, citing supply chain constraints as the biggest bottleneck for shipments next year.

The real constraints in the optical module industry have shifted from the demand side to the supply side. The scarcity of raw materials such as DSPs and lasers, as well as the capacity lock-in strategies of leading manufacturers, will become key variables in determining the profit distribution pattern in 2027.

After visiting 15 optical communication companies in China, Goldman Sachs significantly raised its demand forecast for high-speed optical modules and judged that pricing discipline would be better than market expectations. The core logic revealed in this research report goes far beyond order scheduling—the bottleneck of raw materials on the supply side and the capacity-locking strategies of leading companies are quietly reshaping the profit distribution pattern of this round of AI infrastructure investment.

Goldman Sachs has revised its demand forecasts for 800G and above optical modules upwards by 39% and 36% in 2027 and 2028, respectively, to 144 million and 171 million units. Meanwhile, the report predicts that the price of 1.6T optical modules will remain above $700, with price reductions for 800G products in 2027 expected to be only single digits. This pricing judgment is supported by the continued shortage of key components such as DSPs, lasers, and PCBs, as well as the supply chain management strategies employed by leading manufacturers to proactively lock in production capacity through prepayments and long-term agreements.

Regarding the commercialization path of CPO (Co-packaged Optics), Goldman Sachs points out that test and coupling equipment vendors are the first to realize revenue. Robotec's Q2 revenue grew 172% quarter-over-quarter, exceeding Goldman Sachs' expectations by 141%, confirming this assessment. The report also projects that CPO switch shipments will jump from 10,000 units to 131,000 units between 2026 and 2028, with equipment order visibility extending to 2028.

Upward revision of demand: driven by both scale expansion and specification upgrades

Goldman Sachs' upward revision of demand is based on a comprehensive assessment of global demand for 800G and 1.6T optical modules by companies it interviewed—the interviewed companies gave a range of 130 million to 200 million units, and Goldman Sachs ultimately settled on 144 million and 171 million units for 2027 and 2028, respectively.

The driving forces come from three levels: the continuous expansion of AI server racks, the extension of optical module application scenarios from horizontal scaling out to vertical scaling up and cross-layer interconnection, and the continuous upgrading of product specifications. This means that this round of upward revision in demand is not driven by a single factor, but rather by the combined effect of simultaneous cluster expansion and specification iteration.

In the Chinese market, surveyed companies expect overall demand to grow by double digits, exceeding 50 million units, primarily 800G and 400G, with 1.6T still in its early stages of scaling up. Notably, the surveyed management teams generally emphasized that every optical module shipped is already deployed in data centers, reflecting real end-user demand rather than channel stockpiling. This mitigates the risk of repeating the "double-ordering" mistakes of the 2000s dot-com bubble.

Shortages of lasers and other components will limit actual shipments in 2027.

While demand is being revised upwards, supply-side constraints are becoming the real variable for the industry. Surveyed companies generally pointed out that the continued shortage of DSPs, lasers, and PCBs may suppress actual shipments in 2027, meaning that the demand forecast of 144 million units may not be fully realized on the supply side.

In the laser segment, Chinese suppliers are accelerating the improvement of their 200G EML production capacity and technological capabilities to support the mass production of 1.6T optical modules in 2027. Of the laser production capacity Everbright (688048.SS) has prepared for 2027, EML accounts for slightly more than CW lasers, with 100G lasers dominating within the EML segment, while 200G still accounts for a small proportion. Dongshan Precision's 200G EML has entered mass production, but shipments are currently limited due to DSP supply constraints; the company has secured a supply of 10 million DSPs for 2027 to support laser and optical module shipments.

For VPEC, a leading InP epitaxial wafer manufacturer, the supply of InP substrates is improving as export licenses stabilize, more customers bring their own InP substrates, and second to fourth-tier suppliers outside of China gradually grow. VPEC will expand its MOCVD equipment from the current 62 units to 69 units by the second quarter of 2027, and plans to build a new production base in 2027.

Why haven't prices collapsed despite leading companies locking in production capacity?

Given the tight supply, Goldman Sachs predicts that the price reduction of 800G optical modules in 2027 will be limited to single digits, with the price of 1.6T remaining above $700. This prediction is primarily supported by the competitive landscape of the industry and the supply chain management strategies of leading manufacturers.

The interviewed company executives emphasized that optical modules are not simply an assembly business; the product SKUs are numerous, placing extremely high demands on suppliers' R&D capabilities and responsiveness. Compared to the 5-year product cycle in the telecommunications market, the product iteration cycle in the data communication market has shortened to 1 to 2 years. This makes R&D capabilities and mass production execution a competitive barrier, rather than simply price competition.

The approach taken by global leader Innolight (300308.SZ / 3308.HK) is representative: by collaborating with suppliers on R&D, directly investing in suppliers, and signing prepayment agreements and long-term supply contracts, it locks in sustainable production capacity during periods of supply shortage. This strategy of deeply binding the supply chain further strengthens the competitive advantage of industry leaders against the backdrop of customers demanding diversified production bases under geopolitical pressure, and also provides structural support for overall pricing discipline.

CPO commercialization: Equipment manufacturers are the first to realize their promises, with order visibility extending to 2028.

The commercialization of CPO technology still faces challenges. Laser suppliers point out that 300mW/400mW continuous wave (CW) lasers in external light source (ELS) solutions present heat dissipation difficulties, and the current priority is to promote NPO solutions with 100mW or higher lasers and a power amplifier (PA). Optical module manufacturers have also observed strong demand for 3.2T NPO optical engines in both the Chinese and overseas markets.

While large-scale production of CPOs will take time, testing and coupling equipment manufacturers have already benefited. Goldman Sachs points out that expensive final modules mean multiple tests are needed during production; the earlier defects are detected, the lower the cost, which provides structural support for the demand for testing equipment.

Robotec's second-quarter revenue grew 172% sequentially, exceeding Goldman Sachs' expectations by 141%, directly confirming this logic. Both Robotec and FiconTEC stated that the number of tools they plan to ship in the coming year will exceed half of their cumulative shipments over the past 25 years, with order visibility extending to 2028. Goldman Sachs predicts that CPO switch shipments will grow from 10,000 units in 2026 to 92,000 units in 2027, and then to 131,000 units in 2028.

In terms of technological upgrades, Robotec and FiconTEC plan to reduce wafer-level testing time by 50% to 60% next year and achieve a 4x increase in chip-level testing speed; Enlitech's NightJar hyperspectral imaging system can screen known good chips (KGD) in the early stages of wafer testing, further reducing failure costs.

Verification Window: Three Key Indicators Determine Whether a Judgment is Valid

All of the above judgments have clear and verifiable conditions.

On the demand side, if the actual shipments of 800G and above optical modules reach the lower end of the range of 144 million units in 2027, then the upward revision judgment is valid; otherwise, it needs to be revised downward. The verification window is the annual shipment guidance of each manufacturer in the first quarter of 2027.

On the pricing side, if the average price of a 1.6T optical module falls below $700, the pricing discipline judgment will be overturned.

On the equipment side, if the revenue growth rate of equipment manufacturers turns negative for two consecutive quarters, then the judgment on order visibility is invalid; the verification window is the earnings release period in the third quarter of 2026 (October to November).

Risk warning and disclaimerInvesting involves risk; please exercise caution. This article does not constitute personal investment advice and does not take into account the specific investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Any investment decisions made based on this information are at your own risk.