MASSTART · Project

Cutting Data Center Transceiver Costs to €1 per Gigabit Through Automated Mass Manufacturing

manufacturingTestedTRL 6

Imagine every video call, cloud file, and streaming session travels as light through tiny chips inside data centers. Right now, building these light-carrying chips is slow, expensive, and mostly done by hand — like assembling watches one by one. MASSTART figured out how to mass-produce them on an automated assembly line, making them 6 times faster to build and 10 times faster to test. The goal: get the price down to just €1 per gigabit per second, which would make the explosion of internet traffic actually affordable for the companies running the infrastructure.

By the numbers

€1/Gb/s

Target production cost for mass-manufactured transceivers

Improvement in assembly throughput via next-gen flip chip bonders

10x

Reduction in characterization time through wafer-level evaluation

1 minute

Target characterization time per device

1.6 Tb/s

Highest aggregate line rate demonstrated (16-channel WDM module)

600 Gb/s

Coherent transceiver capacity at 64Gbaud/s

400G

Pluggable transceiver in QSFP-DD format (4-channel PSM4)

800G

Pluggable transceiver in QSFP-DD format (8-channel WDM)

9 partners

Consortium size across 5 countries

67%

Industry partner ratio in consortium

The business problem

What needed solving

Data centers are the backbone of the internet, and their hunger for bandwidth doubles every few years. The optical transceivers that move data inside these facilities are expensive to produce — manual assembly, slow testing, and limited throughput keep costs high. As operators need to jump from 400G to 800G to 1.6T speeds, the manufacturing bottleneck threatens to make upgrades unaffordable.

The solution

What was built

Four working transceiver prototypes were built and validated: 400G and 800G pluggable modules in QSFP-DD format, a 1.6T on-board module, and a 600G coherent transceiver. Behind these sits an Industry 4.0-compatible automated assembly line with next-gen flip chip bonders and wafer-level testing tools.

Audience

Who needs this

Photonic transceiver manufacturers looking to scale production and cut assembly costsHyperscale data center operators (AWS, Google, Microsoft, Meta) needing affordable next-gen optical interconnectsTelecom equipment vendors building coherent optical networking gear for 5G and beyondSemiconductor packaging companies expanding into photonics integrationOptical component suppliers seeking automated testing and quality control solutions

Business applications

Who can put this to work

Data Center Operations

enterprise

Target: Hyperscale and colocation data center operators

If you are a data center operator struggling with the cost of upgrading optical interconnects to handle growing traffic — this project developed automated assembly techniques for photonic transceivers that target €1/Gb/s production cost. Their 400G, 800G, and 1.6T transceiver prototypes were built and validated for both inter- and intra-data center applications, directly addressing the bandwidth scaling problem.

Photonics & Semiconductor Manufacturing

mid-size

Target: Photonic integrated circuit (PIC) packaging houses and transceiver manufacturers

If you are a photonics manufacturer facing bottlenecks in assembly throughput and testing time — this project built a complete Industry 4.0-compatible assembly line with flip chip bonders delivering x6 improvement in throughput and wafer-level testing tools that cut characterization time by a factor of 10, down to 1 minute per device. These techniques were validated across four transceiver demonstrators from 400G to 1.6T.

Telecommunications Equipment

enterprise

Target: Optical networking and telecom equipment vendors

If you are a telecom equipment maker needing next-generation coherent transceivers for long-haul and metro networks — this project demonstrated a 600G coherent transceiver using DP-64QAM modulation at 64Gbaud/s line rate. The automated packaging approach using glass waveguide coupling and 3D TSV packaging was designed specifically to scale to mass production volumes.

Frequently asked

Quick answers

What cost reduction can we actually expect?

The project targeted a production cost of €1/Gb/s or lower in mass production for photonic transceivers. This was pursued through automated assembly (x6 throughput improvement) and wafer-level testing (10x faster characterization, down to 1 minute per device). Final production cost data would need to be confirmed with the consortium.

Can this scale to industrial production volumes?

Yes — this was the core objective. The project built an Industry 4.0-compatible assembly line with next-generation flip chip bonders delivering x6 throughput improvement. The wafer-level evaluation tools reduce testing to 1 minute per device, which is critical for high-volume manufacturing economics.

What about IP and licensing for these manufacturing techniques?

The consortium includes Fraunhofer (Germany's applied research leader) and 6 industry partners across 5 countries. As an Innovation Action with strong industrial participation, IP is likely distributed among partners. Licensing terms would need to be discussed directly with the consortium members holding relevant patents.

What transceiver speeds were actually demonstrated?

Four prototypes were built and evaluated: a 400G pluggable (4-channel PSM4 in QSFP-DD), an 800G pluggable (8-channel WDM in QSFP-DD), a 1.6T on-board module (16-channel WDM), and a 600G coherent transceiver (DP-64QAM at 64Gbaud/s). All underwent large signal characterization and full link validation.

How mature is this technology for deployment?

All four demonstrators were fabricated, assembled using the automated line, and validated through full link testing. The project was an Innovation Action (closer to market than basic research), and it ended in June 2023. The technology has been demonstrated at prototype level with industry-grade evaluation.

Does this comply with industry packaging standards?

The project specifically worked with international standards bodies to ensure compliance and standardization of developed packaging form factors (including QSFP-DD format). This was built in from the start to enable rapid commercialization.

Consortium

Who built it

The MASSTART consortium is heavily industry-weighted at 67% (6 out of 9 partners), with 3 SMEs included — a strong signal that this project was built for commercial outcomes, not just publications. Coordinated by Fraunhofer (Germany), Europe's largest applied research organization, the 9 partners span 5 countries (Germany, Greece, France, Israel, Netherlands), covering a good slice of Europe's photonics supply chain. The mix of 2 research organizations, 1 university, and 6 industry players means the manufacturing know-how stays close to the companies that would actually use it. For a business looking to adopt this technology, the industrial partners are the most likely route to licensing or collaboration.

ARISTOTELIO PANEPISTIMIO THESSALONIKISparticipant · EL
COMMISSARIAT A L ENERGIE ATOMIQUE ET AUX ENERGIES ALTERNATIVESparticipant · FR
ALMAE TECHNOLOGIESparticipant · FR
TEEM PHOTONICS SAparticipant · FR
MELLANOX TECHNOLOGIES LTD - MLNXparticipant · IL
BRIGHT PHOTONICS BVparticipant · NL
ADTRAN NETWORKS SEparticipant · DE
FICONTEC SERVICE GMBHparticipant · DE

How to reach the team

Fraunhofer IOSB or Fraunhofer HHI in Germany — contact through SciTransfer for a warm introduction to the right division.

Order a report on this consortium

Next steps

Talk to the team behind this work.

Want to explore how MASSTART's automated transceiver manufacturing could cut your production costs? SciTransfer can connect you directly with the consortium team and provide a tailored technology brief.

Order a report on this topic More in Manufacturing & Industry 4.0