The Ethernet transceiver segment is growing at 57% year-over-year as of April 2026, according to TrendForce. That growth is not evenly distributed. AI and machine learning workloads are the primary force behind 400G and 800G demand, and those same workloads are already saturating 800G switch fabrics in the largest GPU clusters.
The math is straightforward. A 128-port 800G switch delivers 51.2 Tbps of total switching capacity. As AI training jobs scale to hundreds or thousands of GPUs, inter-rack and spine-layer bandwidth requirements roughly double every 18 to 24 months. 1.6T per port is the answer to that doubling.
The optical transceiver market reached $23 billion in 2025, per LightCounting. The 1.6T segment is not yet a meaningful slice of that figure, but the design wins being made today will determine which suppliers and which form factors dominate when volume shipments begin.
OSFP (Octal Small Form-factor Pluggable) was introduced to support 400G and 800G by housing eight electrical lanes in a wider, higher-power module than QSFP-DD. It is already in production for 800G applications and sits alongside QSFP-DD as one of the two dominant form factors at that speed.
OSFP-XD (Extended Density) is the next iteration. The key differences:
For teams currently choosing between OSFP and QSFP-DD for 800G deployments, the OSFP-XD trajectory is worth factoring in. OSFP's mechanical roadmap extends more naturally to OSFP-XD than QSFP-DD does — which matters if you expect to carry chassis investments across the 800G-to-1.6T transition.
Reaching 1.6T per port requires either more lanes, faster signaling per lane, or both. The technology that makes 1.6T achievable without an impractical lane count is 224G PAM4.
PAM4 (Pulse Amplitude Modulation, 4-level) is already the dominant electrical interface for 100G-per-lane modules used in 400G and 800G transceivers. At 100G PAM4, each lane carries 100 Gbps. At 224G PAM4, each lane carries 224 Gbps. Eight lanes at 224G PAM4 equals 1.792 Tbps of raw electrical bandwidth — rounded to 1.6T after protocol overhead.
The jump from 100G PAM4 to 224G PAM4 is not incremental. Signal integrity challenges grow significantly:
Broadcom's Tomahawk 5 and Marvell's Teralynx 10 are among the switch ASICs being designed to support 224G PAM4 electrical interfaces. Volume silicon availability for switch platforms supporting 1.6T is currently projected for 2027 at the earliest for most vendors.
Here is how the development arc looks from mid-2026:
This timeline assumes no major disruption to silicon supply chains and continued investment in 224G PAM4 DSP development. Both are reasonable given current AI infrastructure spending, but the history of optical standards suggests six to twelve months of slippage is common.
This question surfaces in every 1.6T conversation. Co-packaged optics (CPO) places optical engines directly on the switch package, eliminating the electrical SerDes channel between the switch ASIC and the module — and with it, the biggest signal integrity obstacle at 224G PAM4.
The case for CPO at 1.6T is real. At very high lane speeds, the electrical channel from ASIC to front-panel cage becomes a hard constraint. CPO sidesteps it entirely.
The case against CPO for most operators is equally real:
The most likely outcome before 2028 is that CPO gains traction in hyperscale spine layers where density and power efficiency outweigh serviceability concerns, while pluggable OSFP-XD modules dominate everywhere else. For enterprise teams and regional ISPs, pluggable is the safe planning assumption.
If you are running 100G today and planning a 400G or 800G upgrade, the 1.6T roadmap should inform your chassis and cabling decisions — without driving them. A few specific considerations:
Choose OSFP-capable platforms where possible. If you are buying new spine switches for 400G or 800G, selecting platforms with OSFP cages rather than QSFP-DD-only cages gives you a more natural migration path. OSFP-XD is a mechanical evolution of OSFP; QSFP-DD does not have a direct 1.6T successor in the pluggable roadmap.
Do not overbuy fiber plant for 1.6T yet. Single-mode fiber installed for 400G and 800G will generally support 1.6T optics. Re-fibering for 1.6T is not expected to be necessary in most scenarios. Verify reach requirements once specifications finalize.
Budget for 1.6T as a 2028 to 2030 line item. Engineering teams that include 1.6T in 2027 capital budgets are likely to be disappointed by availability and pricing. Build it into your three-to-five year refresh cycle, not your next fiscal year.
Track DSP and ASIC announcements. Switch silicon is the gating factor. When Broadcom, Marvell, or Intel Tofino successors announce production-ready 224G PAM4 switch ASICs, that is the real signal that 1.6T module timelines are on track.
For teams evaluating the current generation of 800G and 400G optics, HYTOPTODEVICE covers the catalog from 1.25G through 800G across OSFP, QSFP-DD, QSFP28, and other major form factors, with OEM/ODM and white-label options for operators building at scale. The 1.6T roadmap is worth watching, but the 800G procurement decisions you make today are the ones that matter right now.
The optical transceiver market is not waiting for 1.6T to arrive. At $23 billion in 2025 and expanding at roughly 50% year-over-year, it is already being reshaped by AI infrastructure spending at a pace that has surprised most analysts.
400G is the current volume leader for enterprise and data center deployments. 800G is in active hyperscale deployment and beginning to appear in regional data center builds. 1.6T is the horizon technology that hyperscalers are designing toward but have not yet deployed at scale.
For most network teams, the practical implication is a two-phase reality: execute the 400G or 800G upgrade that is already overdue, and keep 1.6T in your architectural awareness without letting it delay decisions you need to make today.
Q1:What is a 1.6T transceiver?
A:A 1.6T transceiver is an optical module capable of transmitting and receiving data at 1.6 terabits per second. It achieves this through a combination of higher lane counts — typically 8 to 16 lanes — and faster electrical signaling per lane, most commonly 224G PAM4, compared to the 100G PAM4 used in current 400G and 800G modules.
Q2:When will 1.6T transceivers be commercially available?
A:Volume commercial availability is projected to begin in 2028, initially for hyperscale customers. Broad market availability at competitive pricing is more realistically a 2029 to 2030 timeframe for enterprise and mid-market buyers, based on current silicon and standards development timelines.
Q3:What is OSFP-XD?
A:OSFP-XD (Extended Density) is a next-generation pluggable module form factor being developed under the OSFP Multi-Source Agreement. It supports up to 16 electrical lanes and is designed to carry 1.6T aggregate throughput. It uses a wider mechanical footprint than standard OSFP and requires new cage hardware in switch platforms.
Q4:What is 224G PAM4 and why does it matter for 1.6T?
A:224G PAM4 is an electrical signaling standard that carries 224 Gbps per lane using four-level pulse amplitude modulation. It is the key enabling technology for 1.6T transceivers because it allows eight lanes to carry approximately 1.6 Tbps of usable throughput. The main challenges are signal integrity at high frequencies, DSP complexity, and power consumption.
Q5:Should I wait for 1.6T before upgrading my network?
A:No. If your network needs a 100G-to-400G or 400G-to-800G upgrade now, waiting for 1.6T will delay a capacity improvement you already need. 1.6T is a 2028 to 2030 technology for most operators. Plan your current upgrade using available 400G and 800G optics, and factor 1.6T into your next three-to-five year refresh cycle.
Q6:Will current fiber infrastructure support 1.6T optics?
A:In most cases, yes. Single-mode fiber installed for 400G and 800G applications will generally support 1.6T optics. Specific reach requirements will depend on the wavelength plan and module specifications once standards are finalized. Re-fibering for 1.6T is not expected to be necessary for most existing deployments.
Q7:How does co-packaged optics (CPO) relate to 1.6T?
A:CPO integrates optical engines directly onto the switch package, eliminating the high-speed electrical channel between the ASIC and a front-panel module — which addresses one of the core signal integrity challenges at 224G PAM4. That said, CPO reduces serviceability and limits configuration flexibility. For most enterprise and ISP deployments, pluggable OSFP-XD modules remain the more practical path to 1.6T.
The 1.6T transceiver roadmap is real, but the timeline is measured in years, not quarters. OSFP-XD and 224G PAM4 are the two technologies you need to understand to follow that roadmap intelligently. Standards work is ongoing, silicon is in development, and volume production is a 2028 story at the earliest.
The most useful thing you can do today is make 800G decisions that do not close off your 1.6T options, stay current on ASIC announcements from Broadcom and Marvell, and treat 1.6T as a planning input rather than a purchasing decision. When specifications finalize and engineering samples ship, you will be ready to move quickly rather than starting from scratch.
