2026 has been defined by the industry as the "year of volume ramp" for 1.6T optical modules. LightCounting predicts that the combined market size for 800G and 1.6T optical modules in 2026 could reach USD 14.6 billion, accounting for approximately 64% of the total optical module market.
In the 1.6T era, heat dissipation is no longer a supporting afterthought – aluminum nitride (AlN) has taken center stage!
The uptake of AlN has been driven step by step by increasing optical module data rates.
In the 400G era, power consumption generally ranged from 8 to 12 W. Alumina ceramics or BT resin substrates were sufficient to meet routine heat dissipation needs. Although AlN offered superior performance, its high cost meant that its thermal management advantages could not translate into commercial value at that time.
In the 800G stage, power consumption rose to typically 15–20 W, with some long-haul and high-power silicon-photonic modules approaching 30 W. To address the thermal challenges of high-density deployment, leading cloud providers such as Google and Meta have introduced AlN in volume for high-end models (especially long-haul and silicon-photonics solutions), driving a significant increase in its penetration in the premium segment.
Entering the 1.6T era, heat dissipation has become a core bottleneck. Module power consumption is expected to jump to over 25–30 W, with heat flux density rising exponentially. In this context, alumina-based solutions are completely out of the picture; given existing mature mass-production processes, AlN – with its extremely high thermal conductivity – has become the industry-standard mainstream solution for meeting 1.6T thermal requirements, and there is currently no cost-effective alternative.
AlN Ceramic Substrates – Soaring Demand!
AlN (standard for 1.6T / 3.2T / CPO): thermal conductivity of 170–230 W/mK, coefficient of thermal expansion well matched to InP and silicon chips, electrically insulating, hermetic, and with high-frequency low-loss characteristics – perfectly addressing two major pain points of high-speed optical devices: high-density heat dissipation and high-frequency signal interference.
In the 400G era, ceramic substrate cost accounted for only 3–5% of the module's total cost.
In the 1.6T era, the value per AlN substrate set rises from USD 17 to USD 22, with the cost share reaching 8–12% – exceeding that of InP optical chips (5–8%).
The ceramic material weight per module is more than 25 times that of the optical chip raw material, so the market growth potential far outpaces that of the optical module equipment sector. The market is expected to expand more than 10-fold in just two years from 2025 to 2027.
Long-term growth ceiling lifted by CPO: this technology abandons traditional discrete optical-device packaging and adopts integrated ceramic substrates that house multiple optical engines in a single unit. The ceramic substrate usage per equipment increases by 3–5 times. In the 3.2T era, CPO solutions are set to become the mainstream route, driving sustained explosive growth in long-term AlN substrate demand.
But with Dominant Players Controlling Supply – What Lies Ahead for AlN?
Overseas supply faces a chokehold due to rare-earth export restrictions. Japanese giant Tokuyama and other players hold about 70% of the global high-end powder market, but the critical sintering aids – rare-earth elements – are now subject to supply cutoffs. Existing inventories are depleting, forcing them to reduce quota allocations to China and stop taking new orders, leading to a tight supply of premium-grade materials.
This supply disruption is pushing leading domestic optical module manufacturers to accelerate supply-chain diversification, creating a real window of opportunity for domestic AlN powders and substrates. Currently, domestically produced powders can already meet the requirements for 1.6T optical modules (with oxygen content at 0.75%), and prices are around RMB 800,000–1,000,000 per ton, with institutions expecting further short-term increases. However, for 3.2T applications, the oxygen content requirement is below 0.75%, which domestic products cannot yet achieve – so high-end applications still rely on imports.