Sivers is building the manufacturing capacity and commercial relationships that could make it a substantially larger supplier to AI infrastructure. Its appeal lies in the combination: specialist laser technology, several routes into optical networking, and a business preparing for customer production programs. The investment case depends on how much of that opportunity becomes profitable, recurring revenue.

The latest development makes the opportunity more tangible

On September 3, Sivers announced a $30 million expansion of its indium phosphide manufacturing facility in Glasgow. The company expects the expanded operation to support more than 100 million continuous-wave distributed-feedback lasers annually. Work is scheduled to begin in the second half of 2026, with the expansion expected to become operational in Q4 2027.

My interpretation is that manufacturing readiness is becoming central to winning business. A customer designing an optical system needs confidence that its chosen supplier can deliver qualified parts when production starts. Additional capacity could strengthen that confidence and allow Sivers to compete for larger programs.

The published capacity target is not a shipment forecast. Utilization, yields, product mix and selling prices will determine its financial value. Even so, the expansion gives investors a concrete project against which to assess management’s ambitions.

Why the laser deserves attention

Sivers develops compound-semiconductor photonics, including laser products used in optical communications. A continuous-wave laser supplies light; a separate modulator can encode data onto that light. Laser arrays provide multiple wavelengths for systems that carry several optical channels. The commercial opportunity is to supply this specialized component as customers build faster optical connections.

For investors, the attraction is the possibility of repeat demand across many units of a qualified system. A successful design can generate orders over its production life. That makes reliability, manufacturing consistency and customer qualification commercially important alongside laboratory performance.

Jabil provides a route into pluggable optics

The April 15 collaboration with Jabil targets a 1.6T linear receive optical transceiver using Sivers’ DFB lasers. A pluggable transceiver is the removable module that connects networking equipment to optical fiber. Jabil’s stated aim is to meet customer bandwidth and power requirements at scale.

This is a useful part of the thesis because it gives Sivers exposure to an identifiable product format. Investors do not have to assume that every data center rapidly adopts co-packaged optics for this opportunity to matter.

I would judge this relationship by qualification progress and subsequent production orders. The collaboration establishes a development route; the announcement does not disclose guaranteed lifetime revenue or a named hyperscale end customer. Successful commercialization would provide stronger evidence of Sivers’ ability to serve the broader transceiver market.

GlobalFoundries could broaden customer access

On June 2, Sivers announced that its laser arrays would be integrated into reference designs on GlobalFoundries’ silicon photonics platform and made available through its SCALE platform. The announced scope includes co-packaged optics and linear pluggable optics. Co-packaged optics places optical components close to the main processor or switch package to improve connectivity.

A reference design gives engineers a supported starting point. My view is that this can lower the integration effort for customers evaluating Sivers’ technology and increase the number of programs in which its lasers are considered. Commercial value will depend on which designs enter production and the supplier allocation within them.

This relationship does not establish that every GlobalFoundries customer buys Sivers products. Its significance is access to a semiconductor platform that customers can use to develop optical systems.

Ayar Labs adds exposure to optical connections between chips

The Ayar Labs relationship predates this year’s announcements. In December 2024, Sivers described discussions to expand manufacturing readiness with Ayar Labs, which it identified as a strategic customer. Ayar’s CEO discussed an intended expansion through development work and product pre-purchases. This is established partnership history, rather than new September news.

Ayar’s current platform combines TeraPHY optical engines with its SuperNova remote light source. The architecture illustrates why a specialist laser supplier can remain relevant as optical connectivity becomes more closely integrated with computing systems.

The potential for Sivers is attractive if optical I/O deployments expand and its arrays retain a meaningful position in those systems. Ayar’s progress alone cannot establish Sivers’ future shipment volumes, exclusivity or revenue share.

SemiNex targets more demanding light sources

The August 13 SemiNex announcement adds a program initially valued at approximately $3.4 million. It targets high-power external light sources, arrays with more wavelength channels and semiconductor optical amplifiers. Customer sampling and early production ramps are targeted for the second half of 2027.

The technical priorities explain part of the opportunity: optical systems need sufficient power at each wavelength, efficient conversion of electricity into light and stable performance as temperature changes. Sivers and SemiNex are addressing these requirements together.

My bullish interpretation is that solving more demanding requirements could support differentiated products and deepen customer relationships. Whether that produces attractive margins will depend on pricing, competition and manufacturing costs.

Scaling production must improve the economics

The Glasgow expansion forms part of a hybrid manufacturing strategy that combines internal production with external foundry, packaging and manufacturing partners. Sivers says this is intended to improve scalability and supply flexibility.

The potential benefit is better control of core processes alongside access to additional manufacturing resources. The financial challenge is to bring capacity online at a pace that customer demand can absorb. An underused factory can consume cash even when the long-term market is growing.

The latest results still show a transition

Q2 sales were SEK 53.8 million, down 12% year over year. Product revenue rose 13%, or 18% at constant currency. Adjusted EBITDA was negative SEK 35.5 million and operating cash flow was negative SEK 70 million. Management reported a $1.2 billion opportunity pipeline in July and expects product ramps to become more visible from Q4 2026 and through 2027.

The pipeline indicates possible business, not contracted backlog. I would not turn it into an annual revenue estimate or apply a valuation multiple to it. Investors need evidence of conversion into orders, deliveries and cash receipts.

The Q2 release also reported approximately SEK 825 million of gross equity raised during the quarter and the subsequent conversion of a $12 million loan into equity. These actions supported funding while increasing the share base.

The relevant outcome for shareholders is value per share. A stronger business can still disappoint investors if growth requires excessive dilution or if the purchase price already assumes a highly successful ramp.

Satellite communications adds another source of demand

Sivers’ wireless business supplies RF beamforming technology for applications including satellite communications. Its Q2 update disclosed an $8.2 million production order from ALL.SPACE for Ka-band beamforming ICs, supporting a 2027 ramp.

A production order gives investors a more concrete commercial milestone than an early technical discussion. It also provides a separate source of potential growth while optical programs develop. The timing of deliveries and follow-on demand will determine how valuable that contribution becomes.

What would make the bullish case stronger

I would look for repeat production orders, a sustained increase in product revenue and better cash conversion. For the factory expansion, the useful evidence is progress against the construction and qualification schedule, followed by utilization. For customer programs, it is the transition from sampling to paid production.

The principal risks are customer delays, qualification failures, competition, price pressure and the cost of scaling. Several promising partnerships can still depend on the same underlying AI spending cycle. Supplier diversification by customers could also limit Sivers’ share of any successful program.

My bullish case is that Sivers can turn its laser capabilities and customer relationships into a materially larger product business. There are credible commercial routes to that outcome, and the announced manufacturing investment makes the ambition more tangible. The strongest future evidence would be customers ordering repeatedly at volumes and margins that improve the company’s cash generation.

That is the development I would follow most closely. It would provide a much firmer basis for valuing Sivers than partner names or headline market forecasts alone.