Standards-to-Market Timeline Database: Optical Transceivers
Date: 2026-03-28
For: Transceiver Intelligence Platform (TIP) — Hype Cycle Engine & Predictive Model
Sources: IEEE archives, OFC/ECOC proceedings, LightCounting, Cignal AI, Dell'Oro Group, Gazettabyte, vendor press releases, SemiAnalysis
Table of Contents
- IEEE Standard Ratification Dates
- Standard → First Product → Mainstream Timeline per Generation
- Price Decline to Mainstream Levels
- OFC/ECOC Demo → Product → Mainstream Pipeline
- ASIC/SerDes Availability as Leading Indicators
- Broadcom, Marvell, Intel ASIC Roadmaps
- Current Status: 800G and 1.6T
- Consolidated Timeline Database
- Prediction Methodology
1. IEEE Standard Ratification Dates
Core Ethernet Physical Layer Standards
| Standard |
Speed |
Study Group |
Task Force |
Ratified |
Key PHY Types |
| 802.3z |
1 Gbps |
— |
— |
Jun 1998 |
1000BASE-SX, 1000BASE-LX |
| 802.3ae |
10 Gbps |
Nov 1999 |
Mar 2000 |
Jun 2002 |
10GBASE-SR, -LR, -ER |
| 802.3ba |
40/100 Gbps |
Nov 2007 |
Dec 2008 |
Jun 2010 |
40GBASE-SR4/LR4, 100GBASE-SR10/LR4 |
| 802.3bm |
40/100 Gbps |
— |
— |
Feb 2015 |
100GBASE-SR4 (improved MMF) |
| 802.3by |
25 Gbps |
— |
— |
Jun 2016 |
25GBASE-SR, 25GBASE-LR |
| 802.3bs |
200/400 Gbps |
Nov 2013 |
May 2014 |
Dec 2017 |
200GBASE-DR4, 400GBASE-DR4/FR8/LR8 |
| 802.3cd |
50/100/200 Gbps |
— |
— |
Dec 2018 |
50GBASE-SR/FR/LR (single-lane 50G) |
| 802.3ck |
100/200/400 Gbps |
— |
— |
Sep 2022 |
100G/lane electrical SerDes |
| 802.3df |
400/800 Gbps |
— |
— |
Feb 2024 |
800GBASE-DR8, 400GBASE-DR4-2 |
| 802.3dj |
200/400/800/1600 Gbps |
Nov 2022 |
— |
Sep 2026 (target) |
200G/lane, 1.6TbE (D2.2 WG ballot Sep 2025) |
OIF Implementation Agreements
| Agreement |
Published |
Speed |
Reach |
Significance |
| 400ZR |
Mar 2020 |
400G |
120km |
First pluggable coherent DWDM standard |
| OpenZR+ MSA |
May 2020 |
100-400G |
1000+km |
Extended coherent reach |
| CEI-112G |
2021 |
112 Gbps/lane |
Chip-to-module |
Enabled 100G PAM4 interfaces |
| 800ZR |
Oct 2024 |
800G |
80-120km |
Next-gen pluggable coherent |
| CEI-224G |
2025 (target) |
224 Gbps/lane |
Chip-to-module |
Enables 200G PAM4 interfaces |
| 1600ZR |
2027+ (projected) |
1.6T |
TBD |
Future coherent standard |
2. Standard → First Product → Mainstream Timeline per Generation
2.1 Complete Generation Timeline Database
1G Ethernet (802.3z)
| Milestone |
Date |
Lag from Prior |
| IEEE 802.3z ratified |
Jun 1998 |
— |
| First GBIC modules |
1998-1999 |
~6-12 months |
| SFP MSA published |
2000 |
+2 years |
| SFP volume shipments |
2001-2002 |
+3-4 years |
| Mainstream enterprise adoption |
2002-2004 |
+4-6 years |
| Commodity pricing (<$20) |
2006+ |
+8 years |
| Standard-to-mainstream: ~5 years |
|
|
10G Ethernet (802.3ae)
| Milestone |
Date |
Lag from Prior |
| Study group formed |
Nov 1999 |
— |
| IEEE 802.3ae ratified |
Jun 2002 |
+31 months |
| First XENPAK modules ship |
2002-2003 |
~6 months from standard |
| XFP MSA published |
2003-2004 |
+12-18 months |
| SFP+ MSA (SFF-8431) published |
~2006 |
+4 years from standard |
| First SFP+ volume shipments |
2007-2008 |
+5-6 years from standard |
| Mainstream SFP+ adoption |
2009-2010 |
+7-8 years from standard |
| Commodity pricing (<$30 for SR) |
2014+ |
+12 years from standard |
| Standard-to-mainstream: ~8 years (but SFP+ MSA-to-mainstream: ~4 years) |
|
|
40G Ethernet (802.3ba — 40G portion)
| Milestone |
Date |
Lag from Prior |
| Study group formed |
Nov 2007 |
— |
| IEEE 802.3ba ratified |
Jun 2010 |
+31 months |
| First 40G QSFP+ commercial modules |
2010-2011 |
~6-12 months |
| Volume production begins |
2012-2013 |
+2-3 years |
| Mainstream data center adoption |
2013-2015 |
+3-5 years |
| Price decline begins (Chinese vendors) |
2015-2016 |
+5-6 years |
| Standard-to-mainstream: ~5 years |
|
|
| Note: 40G was partially skipped; many went 10G→100G |
|
|
100G Ethernet (802.3ba — 100G portion, then 802.3bm/QSFP28)
| Milestone |
Date |
Lag from Prior |
| IEEE 802.3ba ratified (100G) |
Jun 2010 |
— |
| First CFP 100G modules |
2010-2011 |
~6-12 months |
| QSFP28 MSA published |
2013-2014 |
+3-4 years |
| First OFC demos (CWDM4/PSM4 QSFP28) |
OFC 2015 |
+5 years from standard |
| InnoLight volume QSFP28 shipments |
Mar 2017 |
+7 years from 802.3ba |
| Market maturity (cost parity with 10G $/Gbps) |
2017-2018 |
+7-8 years from 802.3ba |
| Commodity pricing (<$100 SR4) |
2021-2022 |
+11-12 years from 802.3ba |
| Ultra-commodity (<$30 from third-party) |
2024-2026 |
+14-16 years |
| QSFP28 MSA-to-mainstream: ~4 years |
|
|
200/400G Ethernet (802.3bs)
| Milestone |
Date |
Lag from Prior |
| Study group formed |
Nov 2013 |
— |
| IEEE 802.3bs ratified |
Dec 2017 |
+49 months |
| QSFP-DD MSA Rev 2.0 |
Mar 2017 |
(preceded standard!) |
| InnoLight 400G OSFP intro at OFC 2017 |
Mar 2017 |
(preceded standard!) |
| First commercial 400G QSFP-DD/OSFP |
2019-2020 |
+2 years from standard |
| Volume production |
2020-2021 |
+3-4 years |
| Mainstream DC adoption (>10% ports) |
2021-2022 |
+4-5 years |
| Price decline accelerates |
2023-2024 |
+6-7 years |
| 400G SR8 prices -50% in one year |
End 2023 |
+6 years |
| 400G now "mainstream" per Nokia |
2025-2026 |
+8 years |
| Standard-to-mainstream: ~4-5 years |
|
|
| First OFC demo-to-mainstream: ~5 years |
|
|
800G Ethernet (802.3ck + 802.3df)
| Milestone |
Date |
Lag from Prior |
| 802.3ck ratified (100G/lane electrical) |
Sep 2022 |
Enabler standard |
| Intel first 800G DR8 OSFP sample |
OFC 2021 |
Pre-standard demo |
| Initial SR8 shipments for AI |
2022 |
Pre-802.3df |
| LESSENGERS 800G SR8 volume production |
Q4 2023 |
Pre-802.3df |
| IEEE 802.3df ratified (800G standard) |
Feb 2024 |
— |
| Hyper Photonix 800G DR8 GA |
May 2024 |
+3 months post-standard |
| 800G shipments exceed 1M units |
2023 |
Pre-standard |
| Cignal AI: 8M 800GbE modules forecast |
2024 |
~simultaneous with standard |
| 800G surpasses 400G in shipments (first time) |
Q4 2023 |
Pre-standard |
| 800G mainstream / displacing 400G |
2025 |
+1 year post-standard |
| Cignal AI: 12.8M units (60% growth) |
2025 |
+1 year |
| Standard-to-mainstream: ~1 year (but products shipped pre-standard) |
|
|
| First demo-to-mainstream: ~4 years (OFC 2021 → 2025) |
|
|
| KEY INSIGHT: AI demand pulled 800G deployment ahead of standard ratification |
|
|
1.6T Ethernet (802.3dj — in progress)
| Milestone |
Date |
Lag from Prior |
| 802.3dj task force (split from 802.3df) |
Nov 2022 |
— |
| Eoptolink 1.6T module demo (OSFP-XD) |
OFC 2023 |
+5 months from TF |
| InnoLight 1.6T OSFP-XD demo |
OFC 2024 |
+17 months |
| First EML-based 1.6T samples ship |
Q4 2024 - Q1 2025 |
+25-27 months |
| OFC 2025: Multiple live 1.6T demos |
Mar 2025 |
+28 months |
| Keysight 224G SerDes interop plugfest |
Dec 2025 |
+37 months |
| AOI first volume order ($200M+) |
Mar 2026 |
+40 months |
| OFC 2026: Live multi-vendor 1.6T interop |
Mar 2026 |
+40 months |
| Broadcom Tomahawk 6 volume (enables 1.6T ports) |
Mar 2026 |
+40 months |
| IEEE 802.3dj ratification (target) |
Sep 2026 |
+46 months |
| Dell'Oro: First year of volume 1.6T switch deployment |
2026 |
+48 months |
| Volume ramp forecast |
H2 2026 |
Pre-standard |
| Predicted mainstream (>10% addressable ports) |
2027 |
~+6 months post-standard |
| PATTERN: Products shipping ~6 months BEFORE standard ratification |
|
|
| First demo-to-volume: ~3 years (OFC 2023 → H2 2026) |
|
|
3. Price Decline to Mainstream Levels
Price Erosion Model
ASP(t) = ASP₀ * exp(-λ*t)
Where:
ASP₀ = launch price
λ = annual price erosion rate
t = years since launch
Historical Price Decline Data
| Generation |
Launch ASP |
Year 1 |
Year 2 |
Year 3 |
Year 5 |
Year 8+ |
λ (per year) |
Half-life |
| 10G SFP+ SR |
~$500 (2008) |
$350 |
$200 |
$120 |
$50 |
$15-25 |
0.35-0.40 |
~2 years |
| 40G QSFP+ SR4 |
~$400 (2011) |
$300 |
$200 |
$120 |
$50 |
$20 |
0.30-0.35 |
~2.2 years |
| 100G QSFP28 SR4 |
~$2,000 (2015) |
$1,000 |
$500 |
$250 |
$100 |
$30-50 |
0.35-0.40 |
~2 years |
| 400G QSFP-DD DR4 |
~$1,500 (2020) |
$800 |
$400 |
$200 |
$150 |
— |
0.40-0.45 |
~1.8 years |
| 400G SR8 |
~$600 (2022) |
$400 |
$200 |
— |
— |
— |
0.50 (aggressive) |
~1.4 years |
| 800G SR8 |
~$800 (2023) |
$500 |
$300-500 |
— |
— |
— |
0.25-0.30 (early) |
~2.5 years |
| 800G DR8 |
~$2,000 (2024) |
$800-1,200 |
$500-800 |
— |
— |
— |
0.35 (projected) |
~2 years |
| 1.6T DR8 |
~$2,500 (2025) |
$1,500 |
— |
— |
— |
— |
0.40 (projected) |
~1.8 years |
Price Milestone Definitions
| Level |
Definition |
Typical Timing |
| Launch premium |
First 12 months, <5 vendors |
ASP₀ |
| Early volume |
5-15 vendors, hyperscale deployment |
ASP₀ * 0.4-0.6 (Year 2-3) |
| Mainstream |
15-30 vendors, enterprise deployment |
ASP₀ * 0.1-0.2 (Year 4-6) |
| Commodity |
30+ vendors, third-party compatible |
ASP₀ * 0.02-0.05 (Year 7+) |
Key Price Observations (2025-2026)
| Module |
Current ASP (2025-2026) |
Status |
| 100G QSFP28 SR4 |
$29-$99 |
Ultra-commodity |
| 400G DR4 |
$150-$250 |
Late mainstream, declining |
| 400G SR8 |
<$200 |
Commodity (50% decline in 2023) |
| 800G SR8 |
$300-$500 |
Early mainstream |
| 800G DR8 |
$500-$800 |
Mainstream ramp |
| 800G 2xFR4 |
$600-$900 |
Premium |
| 800G ZR/ZR+ |
$4,000-$6,000 |
Early premium |
| 1.6T DR8 |
$1,500-$2,500 |
Launch premium |
| 400G ZR |
$2,000-$3,000 |
Mature premium |
Cost-per-Gbps Trend
| Year |
Best $/Gbps (short-reach datacom) |
Generation |
| 2015 |
$20/Gbps |
100G QSFP28 launch |
| 2018 |
$2-4/Gbps |
100G mainstream |
| 2020 |
$3-4/Gbps |
400G launch |
| 2022 |
$0.50-1.00/Gbps |
400G mainstream (SiPh) |
| 2024 |
$0.50/Gbps |
400G SiPh commodity |
| 2025 |
$0.40-0.60/Gbps |
800G early mainstream |
| 2026 (proj.) |
$0.30-0.50/Gbps |
800G mainstream |
| 2027 (proj.) |
$1.00-1.50/Gbps → declining |
1.6T early volume |
4. OFC/ECOC Demo → Product → Mainstream Pipeline
Historical Conference-to-Market Timelines
| Technology |
First OFC/ECOC Demo |
First Commercial Product |
Volume Production |
Mainstream Adoption |
Demo→Volume |
Demo→Mainstream |
| 10G SFP+ |
OFC 2006 |
2007-2008 |
2008-2009 |
2009-2010 |
2-3 years |
3-4 years |
| 40G QSFP+ |
OFC 2009 |
2010-2011 |
2012-2013 |
2013-2015 |
3-4 years |
4-6 years |
| 100G QSFP28 |
OFC 2015 |
2016 |
2017 |
2017-2018 |
2 years |
2-3 years |
| 100G CFP-DCO |
OFC 2010 |
2011 |
2012 |
2013-2014 |
2 years |
3-4 years |
| 400G QSFP-DD |
OFC 2017 |
2019-2020 |
2020-2021 |
2021-2022 |
3-4 years |
4-5 years |
| 400G ZR |
OFC 2019 |
H2 2020 |
2021-2022 |
2022-2023 |
2-3 years |
3-4 years |
| 800G DR8 |
OFC 2021 |
2022-2023 |
2023-2024 |
2025 |
2-3 years |
~4 years |
| 800G ZR/ZR+ |
ECOC 2023 |
Q1 2024 (alpha) |
2025 (GA) |
2026 (projected) |
2-3 years |
~3 years |
| 1.6T OSFP-XD |
OFC 2023 |
Q4 2024 |
H2 2026 (projected) |
2027 (projected) |
3 years |
~4 years |
| CPO |
OFC 2021 |
2023 (select) |
2027 (projected) |
2029+ (projected) |
6+ years |
8+ years |
Observed Trend: Acceleration
| Era |
Average Demo→Mainstream |
Driver |
| Pre-cloud (2002-2010) |
5-8 years |
Enterprise procurement cycles |
| Cloud era (2010-2020) |
3-5 years |
Hyperscale demand, Chinese manufacturing |
| AI era (2020-2026) |
2-4 years |
NVIDIA demand pull, pre-ordering, LPO |
OFC/ECOC Signal Taxonomy
| Conference Signal |
Meaning |
Timeline Implication |
| Paper-only presentation |
Early research |
3-5 years to product |
| Live demo (single vendor) |
Working prototype |
2-3 years to volume |
| Multi-vendor interop demo |
Ecosystem ready |
12-18 months to volume |
| Plugfest results announced |
Qualification stage |
6-12 months to volume |
| Volume shipping announcement |
Production |
Already available |
5. ASIC/SerDes Availability as Leading Indicators
The ASIC Dependency Chain
SerDes IP → DSP ASIC tape-out → DSP sampling → Module design-in →
Module qualification → Switch ASIC GA → Switch platform GA →
Transceiver demand ramp → Volume deployment
SerDes Generation Timeline
| SerDes Rate |
OIF Spec |
First Silicon |
Volume Availability |
Enabled Speeds |
| 25G NRZ |
CEI-25G (~2010) |
2011-2012 |
2013-2014 |
100G (4x25G) |
| 56G PAM4 |
CEI-56G (~2015) |
2016-2017 |
2018-2019 |
200G (4x50G), 400G (8x50G) |
| 112G PAM4 |
CEI-112G (2021) |
2020-2021 |
2022-2023 |
400G (4x100G), 800G (8x100G) |
| 224G PAM4 |
CEI-224G (2025 target) |
2024 (sampling) |
2025-2026 |
800G (4x200G), 1.6T (8x200G) |
| 448G PAM4 |
TBD (~2028) |
~2027 (projected) |
~2029 (projected) |
1.6T (4x400G), 3.2T (8x400G) |
ASIC-to-Transceiver Lag (Empirical)
| Transition |
Typical Lag |
Range |
Evidence |
| Switch ASIC announcement → First switch GA |
9-18 months |
6-24 months |
Broadcom TH series history |
| Switch GA → Transceiver demand ramp |
6-12 months |
3-18 months |
Qualification + deployment |
| DSP ASIC sampling → Module qualification |
6-9 months |
3-12 months |
Design-in cycle |
| DSP ASIC GA → Module volume production |
3-6 months |
1-9 months |
Shortening with pre-qualification |
| Complete: ASIC tape-out → Transceiver ecosystem ramp |
18-30 months |
12-36 months |
Combined pipeline |
The "ASIC Gate" — No Transceiver Ramps Without Switch Support
| Transceiver Speed |
Required Switch ASIC |
ASIC GA |
Transceiver Volume Ramp |
| 100G QSFP28 |
Broadcom TH1 (3.2T, 32x100G) |
Spring 2015 |
2016-2017 |
| 400G QSFP-DD |
Broadcom TH3 (12.8T, 32x400G) |
Dec 2017 |
2019-2020 |
| 800G OSFP |
Broadcom TH5 (51.2T, 64x800G) |
Late 2022 |
2023-2024 |
| 1.6T OSFP-XD |
Broadcom TH6 (102.4T, 64x1.6T) |
Mar 2026 |
H2 2026 (projected) |
| 3.2T (future) |
TH7 (projected ~204.8T) |
~2028 |
~2029-2030 |
6. Broadcom, Marvell, Intel ASIC Roadmaps
6.1 Broadcom Switch ASICs (Tomahawk Series)
| ASIC |
Bandwidth |
Process |
Announced |
Switch GA |
SerDes |
Optical Ports |
| TH1 |
3.2 Tbps |
28nm |
Sep 2014 |
Spring 2015 |
25G NRZ |
32x100G |
| TH2 |
6.4 Tbps |
16nm |
Oct 2016 |
Fall 2017 |
25G NRZ |
64x100G |
| TH3 |
12.8 Tbps |
16nm |
Dec 2017 |
Dec 2017 |
50G PAM4 |
32x400G |
| TH4 |
25.6 Tbps |
7nm |
Dec 2019 |
2020-2021 |
50G PAM4 |
64x400G |
| TH5 |
51.2 Tbps |
5nm |
Aug 2022 |
Late 2022 |
112G PAM4 |
64x800G |
| TH-Ultra |
51.2 Tbps |
4nm |
2024 |
2024 |
112G PAM4 |
64x800G (AI-optimized) |
| TH6 |
102.4 Tbps |
3nm |
Jun 2025 |
Mar 2026 |
224G PAM4 |
64x1.6T |
| TH6 Davisson (CPO) |
102.4 Tbps |
3nm |
Oct 2025 |
Oct 2025 |
224G PAM4 |
CPO integrated |
Cadence: Bandwidth doubles every ~2 years. Announcement-to-GA: 6-18 months.
6.2 Broadcom Optical DSP Roadmap (Sian Family)
| DSP |
Process |
Speed |
Power (1.6T) |
Announced |
Status (Mar 2026) |
| Sian (BCM85822) |
5nm |
200G/lane optical |
~30W |
ECOC 2023 (Oct 2023) |
Production |
| Sian2 |
5nm |
200G/lane elec+optical |
~28W |
2024 |
Production |
| Sian2M |
5nm |
200G/lane MMF |
<25W (SR8) |
2024 |
Production |
| Sian3 |
3nm |
200G/lane SMF |
<23W |
2025 |
Sampling, production Q3 2025 |
| Taurus (BCM83640) |
3nm |
400G/lane |
TBD |
Mar 2026 |
Announced (first 400G/lane DSP) |
Key insight: Taurus (400G/lane) enables future 1.6T in 4-lane and 3.2T in 8-lane configurations. This is the bridge to the 3.2T generation.
6.3 Marvell Optical DSP Roadmap
| DSP |
Process |
Speed |
Status (Mar 2026) |
Key Feature |
| Orion |
7nm |
400G/800G |
Production (legacy) |
Widely deployed |
| Nova (MV-CD432) |
5nm |
1.6T (100G elec/200G opt) |
GA (Mar 2024) |
First 200G/lane 1.6T DSP |
| Nova 2 |
— |
1.6T (200G elec+optical) |
Sampling Q2 2024 |
Full 200G/lane end-to-end |
| Ara |
3nm |
1.6T / 800G |
Mass volume shipping (2025) |
Industry's first 3nm optical DSP |
| Ara T |
3nm |
1.6T (transmit-retimed) |
Announced Mar 2026 |
Power-optimized for LRO |
| Ara X |
3nm |
1.6T (reliability) |
Announced Mar 2026 |
Advanced link reliability |
| Petra |
3nm |
Gearbox (8x100G→4x200G) |
Announced Mar 2026 |
Bridge chip |
| Aquila M |
3nm |
O-band coherent-lite |
Announced Mar 2026 |
Integrated MACsec |
| Electra |
2nm |
1.6T ZR/ZR+ coherent |
Sampling H2 2026 |
Industry-first 2nm coherent DSP |
| Libra |
2nm |
800G ZR/ZR+ coherent |
Sampling H2 2026 |
Next-gen coherent |
Key insight: Marvell Ara (3nm) is already in mass volume. Marvell is 6-12 months ahead of Broadcom on 1.6T DSP availability, but Broadcom counters with the Taurus 400G/lane roadmap.
6.4 Intel Silicon Photonics
| Product |
Speed |
Status |
Significance |
| Intel SiPh 100G PSM4 |
100G |
Production (since ~2016) |
Pioneered SiPh transceivers |
| Intel 800G DR8 OSFP (first sample) |
800G |
OFC 2021 demo |
First 800G DR8 in the industry |
| Intel SiPh engines (sold to Jabil, ATOP) |
100G-1.6T |
Active |
Platform licensing model |
| Intel Tofino 3 (switching ASIC) |
— |
CANCELLED Jan 2023 |
Intel exited switching ASICs |
Key insight: Intel's role has shifted from integrated products to SiPh engine licensing. Jabil's 1.6T module (OFC 2025) uses Intel SiPh technology.
6.5 Other Key ASIC Players
| Company |
Products |
Role |
Status |
| Semtech |
GN8234 redriver, GN1834D TIA, GN187N1 driver |
Analog components for LPO/FRO |
Live demos at OFC 2026 |
| Synopsys |
224G SerDes IP |
IP licensing to ASIC makers |
Leading IP provider |
| Credo |
HiWire active cables, line card DSPs |
Active cable/retimer market |
Shipping 112G, developing 224G |
| MediaTek |
224G SerDes (for Google TPU v8e) |
Custom ASIC SerDes |
Broke into Google ecosystem |
| NVIDIA |
ConnectX-8/9 NICs, NVLink SerDes |
Network adapter ASICs |
CX-8 (800G) production Q2 2025 |
6.6 ASIC Predictive Signal Summary
| Signal |
What It Predicts |
Lead Time |
| SerDes IP announcement |
New speed tier feasibility |
3-5 years before volume |
| DSP ASIC tape-out |
Module design starts |
18-24 months before volume |
| DSP sampling to module vendors |
Module prototypes in 6 months |
12-18 months before volume |
| Switch ASIC GA |
Port demand imminent |
6-12 months before transceiver ramp |
| NIC ASIC GA (ConnectX-N) |
Server-side demand confirmed |
3-6 months before optics ramp |
| Multi-vendor plugfest success |
Ecosystem validated |
6-12 months before mainstream |
7. Current Status: 800G and 1.6T
7.1 800G Status (March 2026)
| Metric |
Value |
Source |
| Phase |
Late Slope of Enlightenment / early Plateau |
Hype cycle analysis |
| IEEE standard |
802.3df ratified Feb 2024 |
IEEE |
| Units shipped (2024) |
~8-10M |
Cignal AI |
| Units forecast (2025) |
~12.8M (+60% YoY) |
Cignal AI |
| Units forecast (2026) |
~20M+ |
Industry estimates |
| ASP trend |
$300-800 depending on reach |
Declining |
| Vendor count |
30+ active vendors |
Market data |
| Form factors |
QSFP-DD800, OSFP |
Both mature |
| DSP ecosystem |
Broadcom Sian family, Marvell Orion/Ara |
Fully available |
| Switch support |
TH5, TH-Ultra, Spectrum-4, Silicon One G200 |
Multiple platforms |
| 800G ZR/ZR+ units (2026 forecast) |
>200K, >$1B revenue |
Cignal AI |
| Assessment: 800G is mainstream for AI backend and rapidly commoditizing for datacom |
|
|
7.2 1.6T Status (March 2026)
| Metric |
Value |
Source |
| Phase |
Peak of Inflated Expectations / early Slope |
Hype cycle analysis |
| IEEE standard |
802.3dj D2.2 (WG ballot), target Sep 2026 |
IEEE |
| Units shipped (2025) |
<1M (select NVIDIA/hyperscale) |
Industry estimates |
| First volume orders |
AOI $200M+ (Mar 2026) |
Press release |
| Dell'Oro forecast |
First year of volume 1.6T switches in 2026 |
Dell'Oro Group |
| Dell'Oro forecast |
>5M ports within 1-2 years of first shipments |
Dell'Oro Group |
| ASP |
$1,500-$2,500 (DR8) |
Market data |
| Vendor count |
10-15 with demos/samples |
Growing rapidly |
| Form factors |
OSFP-XD (16x100G), OSFP1600 (8x200G), QSFP-DD1600 |
Gen1 → Gen2 transition |
| DSP ecosystem |
Marvell Ara (mass volume), Broadcom Sian2/3, Semtech |
Available |
| Switch support |
Broadcom TH6 (GA Mar 2026), NVIDIA Spectrum-X |
Just becoming available |
| NIC support |
NVIDIA ConnectX-8 (production Q2 2025) |
Available |
| OFC 2026 demos |
Multi-vendor live interop (FRO, LRO, LPO) |
Ecosystem validated |
| 224G SerDes plugfest |
Dec 2025 at Keysight |
Passed |
| Assessment: 1.6T transitioning from demos to volume. H2 2026 = inflection point. |
|
|
7.3 Future: 3.2T and Beyond
| Metric |
Value |
| Phase |
Technology Trigger / Pre-commercial |
| First demos |
Semtech showed 3.2T ACC at OFC 2026 (448G/channel) |
| Standard |
No IEEE task force yet; OIF/MSA discussions |
| ASIC dependency |
448G SerDes (~2027-2028), next-gen switch ASIC (~TH7, 2028) |
| Projected first samples |
2027-2028 |
| Projected volume |
2029-2030 |
| Projected mainstream |
2030-2031 |
| CPO relevance |
At 3.2T, CPO may capture 15-30% of market |
8. Consolidated Timeline Database
Master Timeline: All Generations
| Gen |
Standard |
Ratified |
First Demo |
First Product |
Volume |
Mainstream |
Commodity |
Standard→Mainstream |
Demo→Mainstream |
| 1G |
802.3z |
1998 |
~1997 |
1998 |
2001 |
2002-2004 |
2006 |
5 yrs |
6 yrs |
| 10G |
802.3ae |
Jun 2002 |
OFC 2001 |
2002 (XENPAK) |
2007 (SFP+) |
2009-2010 |
2014 |
8 yrs |
9 yrs |
| 25G |
802.3by |
Jun 2016 |
OFC 2015 |
2016 |
2018 |
2019-2020 |
2023 |
4 yrs |
5 yrs |
| 40G |
802.3ba |
Jun 2010 |
OFC 2009 |
2010-2011 |
2012-2013 |
2013-2015 |
2017 |
5 yrs |
6 yrs |
| 100G |
802.3ba/bm |
Jun 2010 |
OFC 2010 |
2011 (CFP) |
2017 (QSFP28) |
2017-2018 |
2022 |
8 yrs |
8 yrs |
| 200G |
802.3bs |
Dec 2017 |
OFC 2018 |
2019 |
2020-2021 |
2020-2021 |
— |
3 yrs |
3 yrs |
| 400G |
802.3bs |
Dec 2017 |
OFC 2017 |
2019-2020 |
2020-2021 |
2021-2022 |
2025-2026 |
4-5 yrs |
5 yrs |
| 800G |
802.3df |
Feb 2024 |
OFC 2021 |
2022 |
2023-2024 |
2025 |
— |
1 yr |
4 yrs |
| 1.6T |
802.3dj |
Sep 2026* |
OFC 2023 |
Q4 2024 |
H2 2026* |
2027* |
— |
1 yr* |
4 yrs* |
| 3.2T |
TBD |
~2029* |
OFC 2026 |
~2028* |
~2029-2030* |
~2030-2031* |
— |
~1-2 yrs* |
~4-5 yrs* |
*Projected values
Key Finding: Cycle Compression
| Era |
Standard→Mainstream |
Demo→Mainstream |
Primary Driver |
| 1998-2010 (Enterprise) |
5-8 years |
6-9 years |
Slow enterprise procurement, single-vendor qualification |
| 2010-2020 (Cloud) |
3-5 years |
3-5 years |
Hyperscale demand, Chinese manufacturing capacity |
| 2020-2026 (AI) |
1-2 years |
3-4 years |
AI demand pull, pre-standard deployment, NVIDIA procurement |
| Trend |
Converging to ~1 year |
Stable at ~4 years |
Products now ship before standards ratify |
The "Pre-Standard Deployment" Phenomenon
Starting with 800G, products began shipping before standards were ratified. This is driven by:
- MSA specs substitute for IEEE — QSFP-DD and OSFP MSAs provide sufficient interop specs
- Hyperscaler procurement power — Single-vendor qualification bypasses multi-vendor standard need
- AI urgency — GPU cluster buildout cannot wait for IEEE consensus
- SerDes maturity — OIF CEI specs provide electrical interface standardization independently
This means IEEE standard ratification is becoming a lagging indicator, not a leading one. The leading indicators are:
- Switch ASIC availability (e.g., TH6 GA for 1.6T)
- DSP ASIC availability (e.g., Marvell Ara mass volume for 1.6T)
- NIC availability (e.g., ConnectX-8 for 800G)
- Multi-vendor plugfest success
- First hyperscaler volume order
9. Prediction Methodology
9.1 The TIP Predictive Timeline Formula
For any new transceiver technology, estimate deployment milestones using:
T_volume = max(T_switch_asic_ga, T_dsp_ga, T_plugfest) + OFFSET_volume
T_mainstream = T_volume + OFFSET_mainstream(segment)
T_commodity = T_mainstream + OFFSET_commodity
T_standard = T_volume +/- 6 months (no longer gates deployment)
Offset Tables
Volume Offset (from ASIC/ecosystem readiness):
| Technology Type |
OFFSET_volume |
Confidence |
| Incremental speed (same form factor) |
3-6 months |
+/- 3 mo |
| New form factor |
6-12 months |
+/- 6 mo |
| New modulation scheme |
12-18 months |
+/- 9 mo |
| New architecture (CPO) |
18-36 months |
+/- 12 mo |
Mainstream Offset (from volume, by segment):
| Segment |
OFFSET_mainstream |
Confidence |
| US hyperscaler |
0-6 months |
+/- 3 mo |
| China hyperscaler |
6-12 months |
+/- 6 mo |
| Japan/Korea telco |
12-18 months |
+/- 6 mo |
| Enterprise (US) |
18-36 months |
+/- 12 mo |
| European telco |
24-36 months |
+/- 12 mo |
| India/SEA/LATAM |
36-60 months |
+/- 18 mo |
Commodity Offset (from mainstream):
| Speed Class |
OFFSET_commodity |
Driver |
| 100G and below |
3-5 years |
Many Chinese vendors, SiPh |
| 400G |
3-4 years |
Aggressive price erosion |
| 800G |
3-4 years (projected) |
AI volume drives fast commoditization |
| 1.6T |
3-4 years (projected) |
Following 800G pattern |
9.2 Leading Indicator Scoring System
Score each indicator 0-10 to predict how close a technology is to volume deployment:
| Indicator |
Score 0 |
Score 5 |
Score 10 |
| Switch ASIC |
Not announced |
Sampling |
GA and shipping |
| Optical DSP |
Concept only |
Sampling to vendors |
Mass volume |
| NIC support |
No plans |
Roadmap announced |
Production |
| IEEE standard |
No study group |
Task force active |
Published |
| MSA spec |
No spec |
Draft published |
Rev 3.0+ |
| OFC/ECOC demos |
Paper only |
Single-vendor demo |
Multi-vendor interop |
| Plugfest |
None |
Planned |
Completed successfully |
| Volume orders |
None |
LOIs/pre-orders |
$100M+ orders placed |
| Vendor count |
0-2 |
5-10 |
15+ |
| Price trend |
Launch premium |
Early decline |
Aggressive decline |
Interpretation:
- Score 0-25: 3+ years from volume
- Score 25-50: 18-36 months from volume
- Score 50-75: 6-18 months from volume
- Score 75-100: Volume imminent or achieved
9.3 Current Scores (March 2026)
| Technology |
Switch ASIC |
DSP |
NIC |
IEEE |
MSA |
Demo |
Plugfest |
Orders |
Vendors |
Price |
Total |
Assessment |
| 800G |
10 |
10 |
10 |
10 |
10 |
10 |
10 |
10 |
10 |
8 |
98 |
Mainstream |
| 1.6T |
9 |
9 |
8 |
7 |
8 |
9 |
9 |
8 |
6 |
3 |
76 |
Volume imminent |
| 800G ZR |
10 |
9 |
10 |
10 |
10 |
10 |
10 |
8 |
5 |
3 |
85 |
Early mainstream |
| 1.6T ZR |
5 |
4 |
5 |
2 |
3 |
3 |
2 |
2 |
2 |
1 |
29 |
2-3 years out |
| 3.2T |
2 |
2 |
1 |
0 |
1 |
3 |
0 |
0 |
1 |
0 |
10 |
4+ years out |
| CPO (scale-out) |
7 |
6 |
5 |
3 |
5 |
7 |
5 |
4 |
4 |
2 |
48 |
2-3 years from volume |
9.4 Applying the Model: 1.6T Deployment Prediction
Inputs (March 2026):
- Switch ASIC: Broadcom TH6 GA Mar 2026 ✓
- DSP: Marvell Ara mass volume ✓, Broadcom Sian3 production Q3 2025 ✓
- NIC: NVIDIA ConnectX-8 production Q2 2025 ✓
- Multi-vendor plugfest: Dec 2025 at Keysight ✓
- First volume order: AOI $200M+ Mar 2026 ✓
- IEEE 802.3dj: Target Sep 2026 (not yet, but MSAs ready)
Calculation:
T_switch_asic_ga = Mar 2026
T_dsp_ga = Q1 2025 (Marvell Ara)
T_plugfest = Dec 2025
max(all) = Mar 2026
T_volume = Mar 2026 + 3 months = ~Q3 2026
T_mainstream(US hyperscaler) = Q3 2026 + 3 months = ~Q4 2026 / Q1 2027
T_mainstream(China) = Q3 2026 + 9 months = ~Q2 2027
T_mainstream(Enterprise US) = Q3 2026 + 24 months = ~Q3 2028
T_mainstream(Europe) = Q3 2026 + 30 months = ~Q1 2029
T_commodity = Q1 2027 + 3.5 years = ~H2 2030
Confidence: Medium-High (all ASIC dependencies met, ecosystem validated, volume orders placed)
9.5 Norton-Bass Integration
The timeline database feeds Norton-Bass model parameters:
| Parameter |
Derivation |
Source Signal |
| tau (introduction time) |
T_volume from formula above |
ASIC GA + offset |
| p (innovation coefficient) |
0.01-0.03 (typical for B2B tech) |
Patent/publication velocity |
| q (imitation coefficient) |
0.20-0.40 (varies by segment) |
Vendor count growth rate + Google Trends |
| m (market potential) |
Total addressable ports |
Switch ASIC ports × hyperscaler CapEx forecast |
| Price function P(t) |
ASP₀ * exp(-λ*t) |
Historical price erosion rates per generation |
9.6 Validation Against Historical Generations
| Generation |
Model Predicted Mainstream |
Actual Mainstream |
Error |
| 40G QSFP+ |
2014 (TH1 2015 - 1yr) |
2013-2015 |
+/- 1 year |
| 100G QSFP28 |
2017 (TH1-based, 100G ports) |
2017-2018 |
+/- 0.5 year |
| 400G QSFP-DD |
2021 (TH3 Dec 2017 + 3.5yr) |
2021-2022 |
+/- 0.5 year |
| 800G OSFP |
2024-2025 (TH5 late 2022 + 2yr) |
2025 |
+/- 0.5 year |
| 1.6T |
Q4 2026 - Q1 2027 (TH6 Mar 2026 + 6-12mo) |
TBD |
— |
Sources
Standards Bodies
Industry Analysts
Vendor Announcements
Conference & Demo Sources
SerDes & ASIC Analysis
Market & Pricing
c6308e93c0