Sivers Semiconductors AB (publ) - SIVE.ST

Deep Dive Research Report | May 10, 2026

Concall dates used: Q1 2025 (May 8, 2025) | Q2 2025 (July 21, 2025) | Q3 2025 (October 24, 2025) | Q4 2025 (February 19, 2026)

1. What the Company Does

Sivers Semiconductors is a Swedish semiconductor company that makes two fundamentally different things in two separate subsidiaries that happen to share a parent company: radio frequency chips for millimeter-wave wireless communications, and indium phosphide laser devices for optical networks and sensing. The wireless chips go inside satellite ground terminals, 5G base stations, and military communications systems. The photonics chips go inside datacenter transceivers, autonomous vehicle LiDAR systems, and wearable health monitors. These two businesses are joined at the hip historically and financially, but they serve almost entirely different customer bases, use different manufacturing processes, and compete in distinct markets.

The company that exists today bears little resemblance to what it was in 1951, when engineer Carl von Sivers founded Sivers Lab AB in Stockholm to make passive microwave components - think cavity resonators and slotted-line VSWR detectors for laboratory use. Philips acquired the company in 1960, added active circuits, and eventually merged it with IMA AB (another Stockholm microwave house, founded 1975) to create Sivers IMA. In 1990, management bought the company out of Philips in a leveraged buyout. Through the 1990s, the company progressively shed legacy products - high-power microwave operations were divested in 1997 - and reinvested in active silicon designs for commercial wireless. The 2000s brought MMIC technology and ultra-wideband voltage-controlled oscillators. By 2017, the company had demonstrated world-class performance at 57-71 GHz WiGig frequencies, and the millimeter-wave thesis was becoming real.

The 2017 acquisition of CST Global, a Glasgow-based III-V compound semiconductor manufacturer, was the event that defined the modern Sivers. CST Global - renamed Sivers Photonics - brought a working 700-square-metre cleanroom, proprietary InP fabrication processes, and a team of compound semiconductor specialists who had spent decades growing indium phosphide wafers. The acquisition gave Sivers a manufacturing anchor it never had in the wireless business and opened an entirely new market: optical communications, where the race to feed AI compute clusters is currently one of the hottest technology supply chains in the world.

In 2020 the group rebranded as Sivers Semiconductors. In 2021 it graduated from Nasdaq First North Growth Market to the main Nasdaq Stockholm exchange. In 2022 it acquired MixComm, a US startup that commercialized RFSOI (radio frequency silicon-on-insulator) technology out of Columbia University, deepening its mmWave IP portfolio and providing initial US operations in New Jersey (now relocated to Chatham).

The company employs approximately 130 people across sites in Kista (Stockholm headquarters and wireless design), Gothenburg (wireless design), Glasgow (photonics manufacturing and design), Chatham (US wireless), and newer presences in San Jose and Bangalore opened in late 2025.

The core value proposition is different for each segment. In wireless, Sivers sells to system integrators who need millimeter-wave chips that can steer beams electronically - so a satellite dish can track an orbiting satellite without physically rotating, or a 5G base station can serve multiple users simultaneously. These beamforming integrated circuits are technically difficult to design, require access to advanced silicon foundry processes, and must pass stringent qualification cycles. In photonics, Sivers sells custom indium phosphide laser chips to manufacturers who need high-power, single-mode, temperature-stable optical sources - the kind that can pump light into a silicon photonics chip to feed a 400G or 800G datacenter transceiver, or maintain sub-megahertz linewidth for coherent LiDAR ranging.

What makes both businesses genuinely hard to replicate is time. The Glasgow photonics cleanroom took decades to optimise. The SATCOM beamformer RFIC design process requires multiple silicon spins at a premium foundry, each lasting six months, before a chip is ready for qualification. A new entrant would need both the process knowledge and the customer trust that comes from surviving those qualification cycles without shipping bad parts.

2. Business Segments

2.1 Sivers Wireless AB

Sivers Wireless designs millimeter-wave radio frequency integrated circuits (RFICs) and beamforming integrated circuits (BFICs). It is a fabless business: the chips are designed in Kista and Gothenburg, then fabricated at GlobalFoundries using an RFSOI (radio frequency silicon-on-insulator) process, then sold as bare die or packaged into antenna modules that Sivers assembles. There is no fab on the wireless side.

The core technical capability is beamforming. When a wireless signal travels at millimeter-wave frequencies (24-71 GHz), it behaves more like light than like a radio wave - it travels in nearly straight lines, bounces off surfaces, and attenuates quickly over distance. To make this useful, systems combine many antenna elements (16, 32, or 64 in typical SATCOM terminals) and shift the phase of each element slightly so that the signals constructively interfere in one direction and cancel everywhere else. This beam can be steered electronically at microsecond timescales. The RFIC or BFIC chip controls this phase-shifting function for all antenna elements simultaneously, while also handling frequency conversion, amplification, and filtering. Designing a chip that does all of this at millimeter-wave frequencies with low power consumption and competitive noise figures is the hard part.

The primary end market is satellite communications (SATCOM). Modern flat-panel SATCOM terminals for LEO (low Earth orbit) satellite constellations like SpaceX Starlink or OneWeb need phased-array antennas that can track fast-moving satellites overhead. Each terminal requires tens to hundreds of Sivers chips. The company's flagship SATCOM customer is ALL.SPACE (formerly Isotropic Systems), which builds multi-orbit, multi-beam terminals and has been working with Sivers since the SATCOM antenna module business was a concept. As of Q4 2025, ALL.SPACE achieved Technology Readiness Level 6 (TRL6) with the US Army, meaning the terminal was validated in a realistic environment for military use. Sivers estimated approximately 1,800 chips per ALL.SPACE terminal with content of around $9,000 per terminal.

The secondary market is 5G Fixed Wireless Access and next-generation cellular infrastructure. Sivers' Summit series RFICs target 28 GHz and 39 GHz bands used in 5G densification. In January 2025 the company signed a major chip development program valued at SEK 60M with a Tier-1 telecom vendor - disclosed only by tier, not by name - for what is likely 5G or 6G mmWave infrastructure.

The third market is defense, which has become increasingly important. In Q4 2025 Sivers won an $800K development contract from a leading US defense contractor, an initial validation of the technology for electronic warfare or communications applications. The Tachyon Networks relationship - a $3M production order for 28 GHz antenna modules for fixed wireless access - also overlaps into defense-related FWA deployments.

The US CHIPS Act relationship matters here. Through the Northeast Microelectronics Coalition (NEMC) Hub, Sivers received $6M in year-one CHIPS Act funding for FR3 (7-15 GHz) beamformer research, with up to approximately $30M expected over three years if programs are renewed. Partners include Ericsson, Raytheon, MIT, Boston University, and Northeastern University. This gives Sivers funded access to the US defense-industrial ecosystem that most European semiconductor companies lack.

Wireless revenues comprised approximately 73-80% of consolidated net sales in 2025, making it the dominant revenue segment. However, the majority of wireless revenue remains Non-Recurring Engineering (NRE) - bespoke development contracts with customers who pay Sivers to design chips for their specific systems. The transition from NRE to product revenue (catalog chips sold at volume) is the most important operational transformation underway.

Competitive position within wireless: Sivers competes with Anokiwave (the most direct comparable in phased-array beamforming ICs for SATCOM), RFcore (Korean specialist), OMMIC (French GaAs-based mmWave), and to a lesser extent with much larger companies like Qualcomm and Analog Devices that have mmWave products but focus on different form factors. Anokiwave is the closest competitor - it has a longer track record in SATCOM beamforming, has been qualified by multiple terminal makers, and has the backing of Renesas. Sivers differentiates on integration (it claims the broadest catalog of SATCOM beamformer products combining Ka-band transmit and receive chips), power efficiency, and multi-orbit capability. In defense, Sivers is at an earlier stage than well-established defense semiconductor suppliers.

2.2 Sivers Photonics Ltd

Sivers Photonics is the operating subsidiary in Glasgow, Scotland. Unlike the wireless business, it has its own cleanroom and manufactures on InP100 - a proprietary 4-inch indium phosphide wafer platform. This is an integrated design-and-fabrication model, sometimes called a fablet or fab-lite operation. In March 2025, Sivers announced a partnership with WIN Semiconductors (a Taiwanese foundry) to scale high-volume DFB laser production externally, beginning a transition toward a hybrid fablet/fabless model similar to what it operates in wireless.

The core product is the DFB (distributed feedback) laser diode - a type of semiconductor laser that emits at a single, stable wavelength with very high spectral purity. When you need to send light down a fiber-optic cable at a specific wavelength, mix it with other wavelengths on the same fiber (wavelength-division multiplexing), or maintain coherence for sensing, DFB lasers are the technology of choice. Sivers makes them from indium phosphide because silicon cannot generate light efficiently; InP is inherently photon-efficient due to its direct bandgap.

The InP100 platform covers wavelengths from 1270 to 1650 nm, which includes the O-band (1260-1360 nm) used in short-reach datacenter interconnects and the C-band (1530-1565 nm) used in long-haul and coherent transmission. The Glasgow cleanroom runs more than 2 million laser tests per month. Yields at this scale in compound semiconductors are notoriously difficult - sub-30% wafer yields are an industry reality, which means the economics of InP manufacturing require process discipline and recipe maturity that takes years to build.

Three application markets currently anchor the photonics business:

AI Datacenter Optical Interconnects. As AI GPU clusters scale to hundreds of thousands of chips, the copper interconnects between them cannot keep pace. Silicon photonics transceivers that move data at 400G, 800G, and eventually 1.6T need external laser sources because silicon itself cannot lase. Sivers' DFB chips provide those light sources. Products include the FR4 CWDM DFB chipset (for CWDM4 400G pluggable transceivers), the DR8 1310nm 70mW DFB laser (targeting 800G and above), and the 8-wavelength O-band CW DFB array for CW-WDM-MSA compliant applications. The partnership with POET Technologies (announced Q3 2025) targets co-packaged optics (CPO) - where the optical components are integrated directly into the switch package rather than plugged in externally, reducing power consumption and board space for hyperscale AI switches. The partnership with O-Net and Enablence also targets AI datacenter external light source modules. In Q4 2024, Sivers signed a SEK 47M MOU with an optical infrastructure company (unnamed) for product qualification and manufacturing readiness targeting 2027 production.

LiDAR. Frequency-modulated continuous-wave (FMCW) LiDAR systems for autonomous vehicles require narrow-linewidth, high-power DFB lasers that can maintain coherence over distances of hundreds of meters. In Q4 2025, Sivers confirmed a LiDAR customer is ramping production in Q4 2026, with cumulative revenue potential of $28M-$53M over 2026-2030. This is the most concrete near-term revenue event in photonics.

Healthcare Wearables. Sivers' laser and photodetector devices enable wearable health monitors - pulse oximeters, continuous glucose monitors, and similar sensing applications. This market is less disclosed publicly but represents a third, more stable revenue stream.

In Q3 2025, photonics segment revenues were SEK 19.4M, up 10% YoY. In Q2 2025, photonics was SEK 22.4M with record product sales of SEK 12.3M. Photonics accounts for roughly 20-27% of consolidated net sales depending on the quarter.

Why photonics exists separately: The Glasgow InP cleanroom is physically separate, operates on a fundamentally different manufacturing process (compound semiconductor epitaxy, wet and dry etching, optical coating deposition), requires different process chemistry and equipment, and serves different customers than the wireless RFIC business. The acquisition of CST Global brought it in as a distinct entity, and it has remained so. The competitive dynamics, margin structure, and technology roadmaps are independent.

Competitive position within photonics: Sivers competes against much larger companies in InP lasers. Coherent (formerly II-VI, which acquired Finisar and Neophotonics), Lumentum, MACOM, Mitsubishi Electric, and Sumitomo Electric are all established InP laser suppliers with higher production volumes and longer customer relationships. What Sivers offers is custom design flexibility and smaller minimum order quantities that make it attractive to emerging applications (LiDAR startups, CPO pioneers, healthcare device makers) that cannot get qualified at a Lumentum or Coherent's minimum volumes. The POET Technologies, Ayar Labs, and O-Net partnerships validate this positioning. However, as volumes scale and production readiness requirements tighten, the question of whether Sivers can compete on cost against vertically integrated giants becomes pressing.

Segment Comparison Summary

3. Products and Business Detail

Wireless Products

Stampede2731 / Stampede2731LP (Ka-band Transmit Beamformer IC) These are Sivers' primary SATCOM transmit chips, operating from 27.5 to 31 GHz. They provide high transmit output power per channel, integrated phase shifting, and support for multi-orbit SATCOM terminal architectures. The LP variant is a low-power version for cost-sensitive deployments. These chips are the heart of the Cloudchaser beamforming chipset launched in 2025 for broad-market SATCOM terminals.

Blueway1721 (Ka-band Receive Beamformer IC) Operating from 17.7 to 21.2 GHz, this is the companion receive chip to the Stampede transmit beamformer. Together they enable full Ka-band duplex SATCOM terminal operation. The Blueway and Stampede chips together form the core of the "Cloudchaser + Maverick" platform - Cloudchaser being the chipset, Maverick being the antenna-array panel.

Summit Series (5G Beamformer ICs) The Summit 2427e, 2629e, and 3741 are integrated beamforming front-ends for 5G applications in the 24-27 GHz, 26-29 GHz, and 37-41 GHz bands respectively. These target 5G base station and fixed wireless access systems. In India, Astra Microwave Products ordered Sivers' 5G mmWave Dual Quad Beamformer IC - the first Indian order for this product family.

TRB Series (Transceivers) The TRB02801 and TRB03901 are advanced transceiver modules combining beamforming, frequency synthesis, and down-conversion. The TRB02801 specifically covers 24.25-29.5 GHz, supports up to 5 Gbit/s data rates, and integrates 32-channel beamforming and beam-steering.

BFM Series (Antenna Modules) Pre-integrated antenna array modules that combine beamformer ICs with antenna elements at 28 GHz, 39 GHz, and 60 GHz. These reduce design burden for customers who need RF performance without full chip-level integration work.

Evaluation Kits The EVK02004 and EVK06005 enable rapid prototyping. These are meaningful for engagement pipeline development - a customer ordering an evaluation kit is the first step in a technical engagement that may convert to a volume order in 12-24 months.

Manufacturing process: All wireless chips are fabricated at GlobalFoundries using an RFSOI process - silicon-on-insulator technology with RF-optimized transistors that can operate efficiently at millimeter-wave frequencies. The RFSOI process offers advantages in integration density and power efficiency compared to alternative technologies like GaAs pHEMT. After fab, chips go through wafer-level testing, dicing, packaging, and final test. The FR3 (7-15 GHz) research program uses a different process optimized for that frequency range.

Photonics Products

FR4 CWDM DFB Chipset A set of four DFB laser chips for coarse wavelength-division multiplexing (CWDM4) applications, targeting 400G optical transceivers. Each chip emits at a different O-band wavelength with external modulation. This is the product that goes into pluggable 400G modules connecting servers within and between datacenter racks.

DR8 1310nm 70mW DFB Laser Diode Designed for the 800G DR8 transceiver standard, this high-power laser operates over a wide temperature range without active cooling. At 70mW continuous wave output, it provides the optical power budget needed for 800G and above datarate transceivers. This is one of the products most relevant to AI cluster interconnect scaling.

8-wavelength O-band CW DFB Laser Chipset An eight-channel laser array compliant with the CW-WDM-MSA (Continuous Wave Wavelength Division Multiplexing Multi-Source Agreement) specification. This is designed for next-generation co-packaged optics where multiple wavelengths are needed simultaneously. The POET Technologies partnership specifically targets this product for CPO light engine integration.

LiDAR DFB Lasers Narrow-linewidth, high-power DFB lasers designed for FMCW LiDAR. The narrow linewidth (sub-megahertz) is critical for coherent ranging because the signal needs to maintain phase coherence over the round-trip time to a distant object. These products are what the unnamed LiDAR customer (ramping production Q4 2026) requires.

Healthcare Lasers and Detectors Emitters and photodetectors for wearable health monitors. Specific products are not publicly described in detail, but this includes near-infrared sources for pulse oximetry and related physiological measurement applications.

InP100 Manufacturing Platform The Glasgow cleanroom operates on 4-inch indium phosphide wafers - smaller than the 6-inch or 8-inch silicon processes used at major foundries, but standard for compound semiconductor production. The platform supports FP/DFB lasers, reflective SOAs (semiconductor optical amplifiers), and booster SOAs. The etched-facet process with integrated optical coatings is the core IP - the cleanliness and repeatability of the facets determines laser output efficiency and failure modes. More than 2 million laser tests are run per month, providing statistical process control data that feeds back into recipe optimization.

WIN Semiconductors Partnership Announced March 2025, WIN Semiconductor (a Taiwanese foundry primarily known for GaAs and GaN wireless chips) is being qualified as an outsourced manufacturing partner for Sivers' DFB laser production. This is counterintuitive - WIN is known for wireless, not photonics - but the underlying compound semiconductor processes share some process chemistry. The goal is to access WIN's higher-volume manufacturing infrastructure to reduce costs and increase capacity beyond what the Glasgow cleanroom can support alone. This transition, if successful, would move Sivers from a small-scale fablet toward a fabless model in photonics, dramatically improving scalability.

Geography of sales: The US is the primary market. Europe is a significant secondary market. Asia is emerging through distribution partnerships - specifically the Nisko partnership in Israel and the O-Net relationship (a Hong Kong/China-based optical module maker with global customer reach including hyperscalers). The new San Jose office targets Silicon Valley hyperscalers and deep-tech startups directly. The Bangalore office supports engineering and qualification work for Asian customers.

4. Customers

Who Buys

SATCOM Terminal Makers: The primary wireless customer type. Companies like ALL.SPACE design complete flat-panel satellite terminals and source the underlying beamformer chips from Sivers. ALL.SPACE is the named flagship customer - the relationship goes back several years through qualification and development cycles. The terminal maker's engineering team makes the design choice during what Sivers calls "technical engagements" - a process that begins with an evaluation kit, progresses through chip integration into a terminal prototype, survives EMC, thermal, and environmental qualification cycles, and eventually results in a production design that specifies Sivers chips for the lifetime of that terminal platform. At approximately 1,800 Sivers chips per ALL.SPACE terminal, a production ramp of even a few thousand terminals creates meaningful revenue.

Tachyon Networks and Doosan represent newer SATCOM module customers operating at smaller scale.

Defense Contractors: The unnamed $800K contract winner and the Intelsat digitizer relationship suggest Sivers is in early-stage qualification cycles with US defense primes. In defense, the qualification cycle is even longer than commercial SATCOM - typically 2-4 years from first engagement to production order. The CHIPS Act partnership with Raytheon and BAE Systems is a funded development collaboration, not yet a production relationship.

Datacenter Optical Module Makers: POET Technologies, O-Net, Enablence, and Jabil represent companies that build the physical transceivers, light engines, or CPO modules that go into AI switches and servers. They buy Sivers laser chips as components to integrate into their products. The buying decision is made by the optical engineer who designs the transceiver module, but the ultimate customer acceptance depends on the hyperscaler (Meta, Google, Amazon, Microsoft) validating the module. This creates an indirect relationship with hyperscalers that is high-leverage if it works - one hyperscaler design win typically means hundreds of millions of units over the product lifetime.

LiDAR System Makers: The unnamed LiDAR customer (production ramp Q4 2026) buys Sivers lasers and amplifiers for inclusion in an FMCW LiDAR sensor. LiDAR customers validate at the complete sensor level - the laser chip is buried inside a system that Sivers never ships directly to the automotive OEM. Revenue predictability comes from the LiDAR maker's vehicle program ramp schedule, which Sivers tracks but cannot control.

Healthcare Device Makers: Companies making wearable health monitors. Details are not publicly disclosed. Switching costs are high once a device has been FDA-cleared with specific laser specifications built into the 510(k) submission.

Switching Costs

In wireless, once a terminal design is frozen with Sivers beamformer chips, replacing them requires a new chip qualification cycle (typically 6-12 months), PCB redesign, EMC re-certification, and in defense, full program office re-approval. This is a genuine lock-in, not a theoretical one. The exception is early-stage NRE customers who are still in the design phase - they could switch before committing a design, which is why converting NRE customers to production customers is the key milestone.

In photonics, a DFB laser chip is specified by wavelength, output power, linewidth, and package type. Once a transceiver module is qualified with a specific laser chip in a specific package, substituting a different supplier's chip requires repeating the qualification. For an 800G module approved by a hyperscaler, this is a multi-million-dollar re-qualification process. Healthcare applications with FDA clearances are even stickier.

Customer Concentration

ALL.SPACE is the most concentrated revenue risk - one customer, one product family, one technology platform. If ALL.SPACE's defense program stalls, Sivers' SATCOM wireless revenue stalls. The Q2 2025 concall noted the pipeline now includes 30 wireless technical engagements with 30% coming from "strong brands" - a signal that management is deliberately building away from ALL.SPACE dependence, but the diversification is still in the engagement phase, not the revenue phase.

The unnamed LiDAR customer, when it ramps, will similarly create a concentrated photonics dependence. The $28-53M cumulative revenue guidance over 2026-2030 is meaningful relative to the company's total revenue, suggesting this one customer could represent 10-15% of annual photonics revenue at peak.

5. Competitive Landscape

Wireless - mmWave Beamforming

The mmWave beamforming IC market for SATCOM is a niche within a niche. Volumes are orders of magnitude smaller than consumer 5G chips (where Qualcomm dominates), and the technical requirements - high transmit power per channel, low noise figure in receive, strict thermal and radiation tolerance - are demanding enough to keep out general-purpose RF IC vendors.

Anokiwave (subsidiary of Renesas since 2021) is the closest direct competitor to Sivers' SATCOM beamformer business. Anokiwave has been in SATCOM phased arrays since approximately 2015, has been designed into multiple LEO terminal platforms, and has the distribution and support infrastructure of a Renesas subsidiary behind it. The main knock on Anokiwave is integration level: historically its chips required more external components per antenna element than Sivers' more integrated designs. However, Renesas' resources give Anokiwave substantially more development bandwidth.

RFcore is a Korean mmWave IC specialist with SATCOM beamformer products competing directly in Ka-band. Less visible in the Western market, but active in Asian SATCOM terminal makers.

OMMIC is a French compound semiconductor foundry and design house with GaAs-based mmWave ICs. Older technology process but established European defense customer base.

Qualcomm has 5G mmWave RFICs for mobile handsets and small cells, but its technology is optimized for the power-constrained, form-factor-constrained handset market - not for high-power SATCOM terminals where different design tradeoffs apply. Not a direct competitor in SATCOM beamformers.

Analog Devices (ADI) has phased array products including the ADAR series, which are widely used in defense AESA radars. These are S-band and X-band radar products rather than Ka-band SATCOM - different frequency, different application. ADI's size and defense relationships are a potential future threat if it chooses to compete in Ka-band SATCOM, but there is no evidence of that roadmap today.

Barriers to entry in this market are real but not impenetrable. A new entrant needs: access to an advanced mmWave process (GlobalFoundries RFSOI or TSMC's equivalent), engineers with chip design experience at Ka-band frequencies, willingness to absorb 2-3 silicon development cycles before reaching production, and patience to survive the customer qualification cycle. The total investment to reach production qualification is likely $15-25M and 3-5 years. This keeps startups out and limits competition to established RF semiconductor houses, of which there are relatively few with the combination of process access and design competence.

Photonics - InP DFB Lasers

This competitive landscape is harder for Sivers. The established InP laser suppliers are significantly larger:

Coherent Corp (formed from II-VI's acquisition of Finisar and Neophotonics) is the dominant player in InP DFB lasers for telecom and datacenter applications. It has massive wafer capacity, decades of recipe refinement, and designs into most major transceiver platforms. Sivers cannot compete at Coherent's volumes.

Lumentum is the other major vertically integrated InP laser supplier. Similarly dominant in pluggable transceivers and traditional DFB applications.

MACOM Technology Solutions competes in InP lasers with particular strength in long-haul and coherent applications.

Mitsubishi Electric and Sumitomo Electric supply InP lasers from Japan with strong historical relationships in Asian and telecoms markets.

Where Sivers competes is in the customization gap that large suppliers won't fill. A LiDAR startup that needs a specific linewidth, center wavelength, and output power combination in a specific chip geometry, in quantities of thousands rather than millions, cannot get quick response from Coherent or Lumentum's sales teams. Sivers can engage in days and commit to design iteration in weeks. The POET Technologies, Ayar Labs, and healthcare device maker relationships all fit this pattern - innovative customers who need a supplier willing to co-develop.

The risk is that this positioning is temporary. When LiDAR volumes scale to the point where Coherent sees a meaningful revenue opportunity, it will compete aggressively. The same applies to CPO - if co-packaged optics becomes a mass-market product, laser supply will shift toward large-volume contracted suppliers. Sivers must either grow fast enough to be cost-competitive before this happens, or maintain a differentiation in custom design capability that larger suppliers are structurally reluctant to match.

The WIN Semiconductors partnership is Sivers' answer to the volume problem. If WIN can be qualified as a second manufacturing source, Sivers gains capacity that its Glasgow cleanroom alone cannot provide and potentially reduces cost as WIN's utilization spreads fixed overhead.

6. Industry

mmWave Wireless

Millimeter-wave wireless (roughly 24-100 GHz) is being driven by two convergent forces: bandwidth demand in satellite communications and 5G densification. The global satellite communications ground terminal market is growing as LEO constellations (SpaceX Starlink, Amazon Kuiper, OneWeb/Eutelsat) deploy at scale and need flat-panel phased-array terminals that can track moving satellites. Each flat-panel terminal requires multiple beamformer ICs. Estimates from multiple market research firms put the millimeter-wave technology market at roughly $1.9 billion in 2025, growing to $6.1 billion by 2030 - a CAGR of approximately 26%.

In the broader 5G context, mmWave (24-100 GHz, also called FR2) is the high-frequency band that delivers multi-gigabit capacity in dense urban environments. Deployment has been slower than initially forecasted due to infrastructure cost and indoor penetration challenges, but fixed wireless access (FWA) deployments - providing home broadband via mmWave - have been a relative success, particularly in the US.

FR3 (7-24 GHz), the frontier that Sivers is researching under the CHIPS Act, combines the coverage advantage of sub-6 GHz with the capacity advantage of mmWave. Industry bodies expect FR3 to be a key 6G band. The fact that Sivers has a funded research program with Ericsson and Raytheon in FR3 positions it for a technology cycle that is 5-7 years from volume deployment.

Regulation shapes the SATCOM market significantly. Ka-band frequency licenses, type approval, and interference coordination requirements create barriers that slow new entrant terminal programs. US ITAR (International Traffic in Arms Regulations) controls apply to SATCOM equipment with defense applications, creating compliance overhead but also limiting foreign competition in US defense contracts.

InP Photonics

The silicon photonics market - which depends on external InP laser sources - was valued at approximately $2.65 billion in 2025 and is projected to reach $9.65 billion by 2030 at a CAGR of approximately 29.5%. The primary demand driver is AI compute infrastructure. Training and inference for large language models requires interconnects that can move data between GPU clusters at hundreds of gigabits per second without the power and heat penalties of copper. Optical interconnects, and the silicon photonics chips that implement them, are the solution - and every silicon photonics chip needs external light.

This creates a supply chain dynamic that NVIDIA has recognized: the company has been pre-allocating InP laser capacity at suppliers well into 2027, creating an industry-wide supply constraint. This is the "InP laser chokepoint" that the SEQH Research analysis identified. For Sivers, this is a tailwind - supply constraints favor existing qualified suppliers and give them pricing power.

Co-packaged optics (CPO) is the emerging next step. Currently most datacenter optics are pluggable - transceivers that slot into a port on a switch. CPO integrates the photonics directly into the switch package, reducing latency and power. The CPO market was estimated at approximately $46 million in 2024 but is projected to reach $8.1 billion by 2030 - a 137% CAGR - as hyperscale AI switches adopt the technology. Sivers' POET Technologies and Ayar Labs partnerships are positioned for this wave.

LiDAR is a smaller but more near-term revenue event for Sivers specifically. FMCW LiDAR using InP lasers requires precisely specified laser performance that commodity DFB suppliers struggle to match. This niche is real and Sivers has a genuine design-in, with production starting Q4 2026.

Healthcare sensing is a slow-growth but stable market with minimal cyclicality.

Compound semiconductor photonics supply chains are inherently constrained by the small number of InP epitaxy suppliers globally - IQE, Wafer Technology, and a handful of others grow the wafers that companies like Sivers then process. Any disruption in the wafer supply chain propagates quickly.

7. Growth Triggers

• LiDAR customer production ramp, Q4 2026. A named (undisclosed to public) LiDAR customer is scheduled to begin production using Sivers lasers and amplifiers in Q4 2026. Management quantified cumulative revenue potential at $28M to $53M over 2026-2030. This is the most specific and dated near-term revenue event in the photonics segment. (Q4 2025 concall, February 19, 2026)

  "A LiDAR customer is expected to ramp production with Sivers lasers and amplifiers in Q4 2026, with potential to deliver cumulative revenues in the range of $28M to $53M over 2026-2030."

• ALL.SPACE defense deployment ramp, H1 2026. ALL.SPACE achieved TRL6 (Technology Readiness Level 6 - system demonstrated in operationally relevant environment) with the US Army in Q4 2025. Management flagged expected deployment of ALL.SPACE terminals into defense agencies in H1 2026 following completion of qualification schedules, which would drive Sivers chip content at approximately 1,800 chips and $9,000 per terminal. This trigger was first mentioned in Q3 2025 and confirmed with TRL6 milestone in Q4 2025. (Q3 2025 concall, October 24, 2025; Q4 2025 concall, February 19, 2026)

• Broad-market SATCOM beamforming chipset (Cloudchaser/Maverick) sampling and production readiness. Sivers launched its first broad-market Ka-band SATCOM chipset in 2025 targeting terminal makers beyond ALL.SPACE. Management stated sampling was planned for Q4 2025, with production readiness typically 6-8 months after first silicon. This implies production-ready qualification by mid-2026 to late-2026, enabling the company to sell to multiple terminal makers from a standard catalog rather than custom NRE. (Q2 2025 concall, July 21, 2025)

• POET Technologies CPO prototypes to customers, H1 2026; production readiness end 2026. The POET-Sivers collaboration on external light sources for co-packaged optics was announced Q3 2025. Management stated prototypes would be demonstrated to customers in H1 2026 with production readiness targeted end of 2026. This positions Sivers to supply into the CPO market, which is projected for rapid acceleration in AI infrastructure buildouts. (Q3 2025 concall, October 24, 2025; Q4 2025 concall, February 19, 2026)

• WIN Semiconductors manufacturing ramp enabling photonics volume. Announced Q1 2025, the WIN Semiconductors collaboration was described as "progressing on schedule" in Q2 2025. The goal is to qualify WIN as a high-volume DFB laser manufacturer, removing the Glasgow cleanroom as the primary bottleneck and enabling Sivers to accept larger production orders without capacity constraints. (Q1 2025 concall, May 8, 2025; Q2 2025 concall, July 21, 2025)

  "WIN Semiconductors partnership progressing on schedule for AI data center production."

• New AI datacenter photonics customer, production target H2 2026 to H1 2027. In Q2 2025, Sivers secured a laser development contract with a new strategic photonics customer targeting AI, HPC, and datacenter applications. Management guided production with this customer for H2 2026 to H1 2027. (Q2 2025 concall, July 21, 2025)

• Tier-1 telecom chip development program (SEK 60M contract). Disclosed at Q4 2024 results, a major chip development program with a Tier-1 telecom vendor for mmWave technology was signed. NRE milestones from this program will feed through 2025-2026 and if the customer adopts Sivers chips in its network infrastructure, it represents a very large volume opportunity. CHIPS Act FR3 research (with Ericsson, Raytheon) follows a similar pathway. (Q4 2024 year-end report referenced in Q1 2025 concall)

• US CHIPS Act milestone funding. First-half year-one CHIPS Act funding was received in Q1 2025, confirming the company was on track for approximately $30M in total program funding over three years. This provides funded R&D for FR3 (6G) beamformers, reducing Sivers' own development cost while maintaining IP ownership. (Q1 2025 concall, May 8, 2025)

• Opportunity pipeline grew 64% to $453M in FY2025. Management reported this figure at Q4 2025, reflecting growth in both wireless and photonics engagements. The pipeline represents projects at various stages of qualification, not contracted revenue, but it is directionally important as a leading indicator. (Q4 2025 concall, February 19, 2026)

• Nasdaq New York dual listing. Announced April 16, 2026 (post period covered by these concalls but a major strategic event). The company is evaluating a secondary listing on Nasdaq NY, which would broaden access to US technology investors and potentially improve liquidity and analyst coverage. Subject to PCAOB audit completion. If completed, this broadens the investor base significantly.

8. Key Risks

1. Cash Runway and Dilution Dependency

The most immediate risk is capital. Sivers ended Q4 2025 with SEK 43.5M in cash, and the FY2025 adjusted EBITDA was negative SEK 10.8M. The company is not cash-flow positive. To fund operations and growth, the board resolved a directed share issue of SEK 125M in April 2026 at SEK 14.50 per share, subject to EGM approval on May 11, 2026. If that issue is completed and DNB, Storebrand, and other committed investors follow through, the immediate liquidity crisis is addressed. If the EGM rejects the issue or committed investors withdraw, the company would face a funding shortfall within months. The history is notable: a SEK 108M raise occurred in September 2024. Two large dilutive rounds in 18 months signals structural cash consumption that product revenue growth has not yet offset.

2. The Insider Trading Investigation

Sweden's Economic Crime Authority (Ekobrottsmyndigheten) opened a preliminary inquiry into whether confidential information about Sivers' planned Nasdaq dual listing leaked before the April 16, 2026 official announcement. The investigation is examining whether people with advance knowledge traded in Sivers shares ahead of the announcement - which could include employees, advisors, or counterparties involved in preparing the listing. The mechanism of harm here is multi-layered: if an employee or director is implicated, Nasdaq Stockholm may impose trading suspensions or sanctions; the Nasdaq US listing process could be complicated or delayed if US regulators view a pending Swedish investigation unfavorably; and the reputational damage could deter the institutional investors needed for the SEK 125M raise. At the time of writing this report, the investigation is at a preliminary stage and no charges have been filed. But "preliminary inquiry" in Sweden is the formal start of a criminal investigation process, not a trivial administrative inquiry.

3. PCAOB Audit and Potential Restatements

To dual-list on Nasdaq New York, Sivers must have its 2024 and 2025 financial statements audited under US PCAOB standards. This involves a different methodology than the Swedish GAAP audit Sivers has historically filed. Specifically, the company disclosed that the review may trigger adjustments in revenue recognition between periods, inventory valuations, and stock option programs. Revenue recognition adjustments could shift revenue from periods already reported, changing the trend picture. Inventory valuations in a compound semiconductor business with sub-30% wafer yields can be material. If restatements are significant, they could alter investor perception of the growth trajectory and complicate the Nasdaq US listing application.

4. NRE-to-Product Transition Risk

The company has repeatedly stated its aspiration to reach 80% product revenue. In FY2025, product revenues were SEK 85.7M, approximately 28% of total net sales. The remaining 72% is NRE - custom development contracts that are by definition non-recurring. A pure NRE business is a services business, not a product company; it cannot scale without proportional headcount increases. The transition to products requires customers to complete qualification cycles and commit to volume purchases. As of the Q4 2025 concall, Sivers has 30 wireless technical engagements active but these are in various stages of the qualification funnel. The conversion rate from engagement to production order in this industry is low. If conversion takes longer than management expects - a routine outcome in semiconductor sales cycles - the NRE revenue must be replaced by new NRE contracts or the top line stalls.

5. Customer Concentration at ALL.SPACE

The SATCOM wireless business depends heavily on ALL.SPACE. ALL.SPACE is building a differentiated multi-orbit terminal, but it faces its own execution risk: deploying terminals to defense agencies requires those agencies to fund purchases, which depends on US defense budget cycles and program priorities. If ALL.SPACE's defense program slows, Sivers' SATCOM wireless revenue could be materially impacted in 2026. There is no disclosed secondary SATCOM customer with comparable near-term revenue visibility.

6. Forex Headwinds

Sivers reports in Swedish kronor (SEK) but earns significant revenue in US dollars. In Q3 2025, FX headwinds reduced reported SEK growth by approximately 10 percentage points relative to constant-currency growth. If the US dollar remains weak against the Swedish krona and euro, reported revenue growth will continue to understate constant-currency performance, compressing the financial metrics that investors observe. For a company burning cash and raising equity, a weaker reported top line makes investor conversations harder.

7. Competition Scaling in Photonics

As CPO and LiDAR scale from niche to mainstream, the economics change. Coherent Corp, Lumentum, and other vertically integrated suppliers will compete aggressively for volume contracts. Sivers' differentiation today is custom design capability and willingness to serve early-stage customers. At scale, customers typically recompete on price. Without the manufacturing cost advantages of large incumbents, Sivers could find margins under pressure precisely when revenue is growing fastest. The WIN Semiconductors partnership is the strategic response to this risk, but it remains unproven.

8. Short Seller Pressure and Valuation

With the stock up approximately 1,200% and trading at a price-to-sales multiple of approximately 31.1x - versus a European semiconductor sector average of 4.1x - short sellers have established positions totaling approximately 6% of shares outstanding. Short pressure alone does not cause business problems, but it creates reflexive risks: negative news (insider probe, audit delay, cash shortfall) gets amplified into larger stock moves, making it harder to raise capital at acceptable dilution levels and creating management distraction.

9. Walk the Talk

Concall dates used: Q1 2025 (May 8, 2025) | Q2 2025 (July 21, 2025) | Q3 2025 (October 24, 2025) | Q4 2025 (February 19, 2026)

Q1 2025 (May 8, 2025) - Setting the Baseline

Coming off a record Q4 2024 (SEK 76.7M), CEO Vickram Vathulya declared that Q4 2024 "establishes a new baseline for quarterly revenues in 2025." That was a specific claim: the company was saying that no quarter in 2025 would fall below the Q4 2024 level. He cited wireless revenue growth of 85% YoY in Q1 as confirmation of strong demand. He also mentioned the WIN Semiconductors partnership and O-Net expansion as enablers of photonics scaling, and said CHIPS Act first-half funding was received.

The specific WIN partnership and CHIPS Act funding claims were verifiable against press releases and were accurate. The "new baseline" claim was directional guidance - not a formal number.

Q2 2025 (July 21, 2025) - Delivering on the Baseline

Q2 2025 revenue came in at SEK 72.4M - the highest Q2 in company history and +38% YoY. This was above Q4 2024's SEK 76.7M on a headline basis... except Q4 2024 had a strong seasonal push. More importantly, the year-to-date trajectory was clearly accelerating. The wireless division grew 71% YoY in Q2, which was ahead of anything the company had previously shown in that segment. Photonics had a slight revenue dip YoY (-3%) but its product sales hit a record SEK 12.3M - consistent with the stated strategy of shifting from NRE to products.

Management introduced the 30-engagements metric in wireless, with 30% from "strong brands," and set specific milestones: broad-market SATCOM chipset sampling in Q4 2025, and a new AI datacenter photonics customer targeting production in H2 2026 to H1 2027. These were dateable, specific commitments.

Q3 2025 (October 24, 2025) - FX Complications but Substance Progressing

Q3 2025 came in at SEK 72.5M, essentially flat versus Q2 2025 at constant SEK but with significant FX drag reducing what was actually 34% constant-currency YoY growth to 24% reported growth. Management was transparent about the FX impact. They also delivered on the POET Technologies partnership announcement and confirmed the capital raise (reducing warrants by 50% as stated) - both specific actions promised in prior quarters.

What was less clear: the broad-market SATCOM chipset sampling was mentioned as planned for Q4 but not confirmed yet, and the ALL.SPACE defense deployment was described as an H1 2026 event "following completion of qualification schedules" - the timeline that had been the subject of several quarters of anticipation.

Management named specific new hires (CFO Heine Thorsgaard, VP Global Operations Neeraj Chopra) and opened the IRIS 2 SATCOM program engagement. These are concrete operational acts, not just words.

Q4 2025 (February 19, 2026) - Revenue Record but Cash Crunch

Full-year 2025 revenue of SEK 304.1M (+25% YoY) delivered the "continued year-over-year growth" guided across all prior concalls. The pipeline grew to $453M (+64%), the LiDAR customer production ramp was confirmed with dollar-value guidance ($28-53M), ALL.SPACE achieved TRL6 with the US Army, and the broad-market SATCOM chipset was positioned as heading toward production readiness. New executive hires (CRO Raymond Biagan, VP Operations Neeraj Chopra) were completed as previously stated.

However, the cash position of SEK 43.5M at year-end revealed the burn rate problem. And the adjusted EBITDA for Q4 alone came in at SEK 10.8M, down 30% YoY - despite record revenue, profitability in Q4 deteriorated versus the prior year Q4 (SEK 15.5M). Management attributed this to sales mix and investments for growth but did not walk through the mechanism in detail.

"Our full-year results on revenue growth and improved profitability, underscore the real progress Sivers is making." - CEO Vickram Vathulya, Q4 2025 concall

The phrase "improved profitability" refers to the 31% improvement in full-year adjusted EBITDA - true, but the number remains deeply negative (-SEK 10.8M for the year), and Q4 alone deteriorated. This is a case where the headline claim is technically accurate but the framing is optimistic.

Overall Assessment: Sivers management is directionally credible on business trajectory - revenue has grown consistently at 24-38% per year, the strategy of shifting toward AI datacenter and SATCOM is attracting real customer engagements, and the stated partnerships have been announced as promised. Where management has been looser is on the path to profitability - the "new baseline" and "continued growth" framing has consistently been accurate in top-line terms, but the gap between NRE revenue and recurring product revenue remains very wide. Management acknowledges the target of 80% product revenue but has not provided a timeline for reaching it, and the current 28% ratio shows how far the journey is.

The ALL.SPACE defense deployment has been anticipated "H1 2026" since at least Q3 2025 - as of this report's date (May 10, 2026), that deployment has not been publicly confirmed as completed. This is a specific promise that investors will track over the next 4-6 weeks.

10. Shareholder Friendliness Index

Dividends: Sivers has paid no dividend in any of the last three financial years. The board proposed zero dividend for FY2025, and zero dividend was paid for FY2024 and FY2023. This is not a surprise for a pre-profitability growth company burning operating cash, but the record is clear: this is not a capital-return business at this stage of its lifecycle.

Buybacks and Dilution: There have been no share buybacks. The share count has grown substantially. In September 2024, the company raised SEK 108M through a directed share issue. In April 2026, the board resolved a further SEK 125M directed issue (8.62 million new shares at SEK 14.50), subject to EGM approval. Prior to this, in September 2023, another capital raise added shares. The trajectory of the share count is steadily upward - existing shareholders' stakes are being diluted with each raise. The Q3 2025 capital raise also reduced outstanding warrants by 50%, which partially offsets future dilution from warrant exercise, but the net effect of the directed issues is clear: new shareholders are being created and old shareholders' ownership percentages are shrinking.

Verdict: Consumes Capital. The company is in a cash-consumption phase with no dividend, no buybacks, and a series of dilutive equity issuances every 12-18 months.

11. Insider Activities

Source: Finansinspektionen Insider Register (marknadssok.fi.se) - Swedish Financial Supervisory Authority's mandatory PDMR transaction register under EU Market Abuse Regulation Art. 19. All transactions referenced below are filed as "Insyn" (insider) disclosures.

The past 12 months are dominated by a concentrated burst of selling in late March 2026, which coincided with the stock's dramatic 1,200% rally ahead of the Nasdaq dual listing announcement.

(All transactions: Finansinspektionen Insyn Register, publication date 2026-03-25)

On the Todd Thomson / Kairos transactions: Kairos Venture Investment sold approximately 2.41 million shares between March 20-25, 2026, for a total of approximately SEK 23M. Todd Thomson is the COO of Kairos Ventures and a board member of Sivers. Kairos issued an official press release explaining the sale as "periodic fund level rebalancing and payment of fund expenses." Thomson stated: "Both Kairos and I personally have a stronger conviction than ever in the long-term potential of Sivers Semiconductors, and I continue to maintain my sizeable personal shareholdings in Sivers intact." This explanation is plausible for a venture fund that has expenses and rebalancing needs, and Thomson's personal stake remaining intact adds some credibility.

On Tomas Duffy and Harish Krishnaswamy: Both board members sold personal shares (not fund shares) in the same March 20-25 window. Duffy disposed approximately 350,000 shares worth approximately SEK 4.5M. Krishnaswamy disposed approximately 696,197 shares worth approximately SEK 8.8M. No public explanation was given for these sales. The timing - during the stock's steep rally, and approximately three weeks before the Nasdaq dual listing announcement on April 16, 2026 - is what triggered the Economic Crime Authority's preliminary inquiry. No charges have been filed.

Net Assessment: This is the most concerning insider picture an analyst would want to see: three board members selling in the same narrow window, with no stated reason for two of the three sales, at prices that proved to be below the post-announcement rally. Sweden's Economic Crime Authority does not open criminal preliminary inquiries routinely - the threshold is meaningful. While the Kairos sale has a credible business rationale, the simultaneous personal sales by Duffy and Krishnaswamy without disclosed reasons, within weeks of a material unannounced corporate event, is a genuine red flag that warrants monitoring. There have been zero insider buys over the past 12 months in the Finansinspektionen register.

12. Scenarios

Bull Case

The bull case for Sivers requires several things to go right simultaneously, and the narrative of 2026-2027 would look something like this:

The SEK 125M directed share issue closes successfully in May 2026, providing 18 months of operating runway. The PCAOB audit of FY2024 and FY2025 completes without material restatements, clearing the path for a Nasdaq New York dual listing in Q3 or Q4 2026. The dual listing attracts US technology growth funds who have never heard of Sivers Semiconductors and who are desperately looking for ways to play the AI optical interconnect supply chain without paying Nvidia-like multiples. That fresh investor base funds another capital raise at favorable terms.

On the business side, the LiDAR customer begins production in Q4 2026 on schedule, delivering the first significant photonics product revenue event. The POET Technologies CPO light engine is demonstrated to hyperscaler customers in H1 2026 and goes into qualification at a major AI switch maker - a process that takes 12-18 months but creates a contractual revenue stream visible from 2027 onward. ALL.SPACE deploys terminals to US defense agencies in H1 2026 as guided, and the $9,000-per-terminal content creates a revenue line that grows as more agencies procure. The Cloudchaser SATCOM chipset qualifies with three or four terminal makers beyond ALL.SPACE, creating the volume and diversification that turn SATCOM from a concentrated single-customer bet into a category.

The WIN Semiconductors partnership delivers production-quality DFB lasers by end of 2026, removing the Glasgow cleanroom as the capacity constraint and enabling Sivers to accept orders at 3-5x its current photonics production rate. Product revenue rises from 28% to 50%+ of net sales as volume commitments convert from NRE to catalog chips. The company crosses adjusted EBITDA breakeven in 2027.

In this scenario, Sivers' investment in the FR3 research program under the CHIPS Act begins to generate first product samples, and Ericsson becomes a qualified design partner for a 6G infrastructure chip program that extends the wireless revenue stream beyond the current SATCOM focus.

Base Case

The base case assumes management delivers roughly what it has guided, execution is imperfect in some areas, and one or two timelines slip by a quarter.

The SEK 125M raise completes. Cash is stabilized. The Nasdaq listing proceeds more slowly than hoped - the PCAOB process reveals some revenue recognition adjustments that require minor restatements, and the listing is delayed into H1 2027 rather than 2026. The insider investigation is inconclusive or results in no charges.

LiDAR production starts in Q4 2026 but ramps slowly through H1 2027, delivering the lower end of the $28-53M cumulative range by 2030. CPO partnerships progress through customer sampling but do not reach production by end of 2026 - the 2027 timeline mentioned by management turns out to be more realistic. ALL.SPACE defense revenues start in 2026 but at modest terminal volumes - hundreds rather than thousands - because defense procurement is always slower than anticipated.

The Cloudchaser chipset gets designed into two or three terminal programs beyond ALL.SPACE by end of 2026, with production ramps starting in 2027. The Tier-1 telecom vendor program delivers NRE milestones through 2026. Revenue continues growing 20-25% annually. Adjusted EBITDA approaches breakeven by late 2027 or 2028. The share count continues growing as the company raises capital annually. Existing shareholders are progressively diluted.

Bear Case

The bear case does not require every risk to materialize - just two or three of them hitting simultaneously.

The EGM on May 11, 2026 approves the SEK 125M raise but the insider trading investigation worsens - one of the selling board members is formally charged, triggering a crisis of governance confidence. The US Nasdaq listing is put on hold indefinitely because regulators are unwilling to list a company under active investigation. The SEK 125M is not enough to fund operations through to cash-flow breakeven, and a third dilutive raise within 12 months is needed at a much lower stock price after the rally reverses.

The LiDAR customer's production ramp is delayed from Q4 2026 into 2027 because the LiDAR company itself is struggling to win automotive OEM design slots - a common outcome in autonomous vehicle supply chains where every program takes longer than planned. ALL.SPACE's defense deployment is delayed further because US Army procurement timelines stretch, and the new administrations' defense budget priorities do not align with LEO SATCOM terminals for the agencies ALL.SPACE is targeting.

The PCAOB audit reveals that revenue recognized in 2024 and 2025 under Swedish GAAP needs to be shifted between periods - not fraud, but different timing conventions - and the restatement shows two quarters of 2025 revenue lower than originally reported, shaking investor confidence in the business trajectory.

Coherent Corp, recognizing the CPO laser opportunity, signs exclusive long-term supply agreements with POET Technologies and other CPO platform makers, cutting off Sivers' route to hyperscaler optical supply chains. In wireless, the US defense contractor that gave Sivers the $800K development award redirects the program to Anokiwave as part of a broader defense radio procurement consolidation.

By mid-2027, the company is raising dilutive equity for the third time in three years, the stock has given back most of the 2026 rally, and the business is still generating 70%+ NRE revenue with no clear path to the product business that management has been promising.

13. Further Reading

No coverage found from SemiAnalysis, Stratechery, or MBI Deep Dives at the time this report was generated.

• The SemiAnalysis Co-Packaged Optics deep-dive covers the CPO supply chain including external laser sources - Sivers Semiconductors is a direct participant in this supply chain - Source: SemiAnalysis [paid]

• The GlobalFoundries Fotonix silicon photonics foundry analysis covers the CPO and silicon photonics manufacturing landscape; GlobalFoundries is Sivers' wireless chip foundry - Source: SemiAnalysis [paid]

(Note: Neither the SemiAnalysis CPO article nor the GlobalFoundries article mention Sivers Semiconductors by name in the accessible portions. These are included as adjacent reading for the industry context they provide. No qualifying articles mentioning Sivers directly were found from SemiAnalysis, Stratechery, or MBI Deep Dives.)

Sources:

• Sivers Semiconductors Q4 2025 Interim Report Press Release
• Sivers Semiconductors Q3 2025 Earnings Highlights
• Sivers Semiconductors Q2 2025 Earnings Call - Alpha Spread
• Sivers Semiconductors Q1 2025 Report Q&A
• Sivers Semiconductors About Us
• Sivers Semiconductors History
• Sivers Semiconductors Wireless Products
• Sivers Semiconductors Photonics Products
• Sivers Semiconductors Annual Report 2025 Postponement Notice
• Sivers Semiconductors Directed Share Issue SEK 125M
• Kairos Ventures Clarifies Share Sale Reasons
• Sivers Semiconductors Fraud Probe Coverage
• Finansinspektionen Insider Register - Sivers
• SEQH Research - InP Laser Chokepoint Analysis
• POET Technologies - Sivers Collaboration
• Sivers - WIN Semiconductors Partnership
• Sivers CHIPS Act Funding
• Silicon Photonics Market Data - Reports and Reports
• SemiAnalysis Co-Packaged Optics Article