top of page
OPTICS · SEMICONDUCTORS · AI INFRASTRUCTURE | stt@simpletechtrend.com
Articles: Blog2
ARTICLES
Articles
Explainers, conference notes, earnings-call reads and market insights on optical communications, silicon photonics, CPO and the AI infrastructure supply chain.
Search


The Fastest-Growing Cloud Fell the Most: Hyperscalers' Bottleneck Shifts From Demand to Delivery and Cash — Q3 2026 Cloud Earnings Review
All three clouds accelerated, yet Alphabet, with the fastest cloud growth, fell nearly 10% after hours. Capex came in light for supply-side reasons, and the market is now asking who can deliver and who can pay.
Nov 92 min read


OCP Global Summit 2026 | Meta | Granthana Rangaswamy, Siamak Amiralizadeh, Xu Wang | Integrated Optics at Scale: From 51.2T Validation to 102.4T
Meta's 51.2T CPO test is closed: zero optical-engine failures and 6x fewer link interruptions than pluggables, but 80% of what remains traces to the external laser module, root-caused to one laser-driver SMT part. At 102.4T, CPO vs NPO becomes a supply-chain question.
Oct 237 min read


OCP Global Summit 2026 | NVIDIA, Meta | Harish Krishnamoorthy, Parvez Shaikh | NVIDIA SDK + FBOSS: Production-Ready AI Network Fabrics
Meta's first NVIDIA switch program runs Meta's own open-source FBOSS. The hardest number is the schedule: with OCP SAI and the SimX simulator, tests hit 90% a week after hardware arrived and 100% two quarters later.
Oct 237 min read


OCP Global Summit 2026 | Meta, AMD | Arvind Srinivasan, Sandeep Nagaraja, Balakrishnan Raman | MetaRoCE: Rethinking RDMA for AI
Meta moves RDMA intelligence from the switch into the NIC: switches need only ECN and ECMP, while multipath, retransmission and reordering live in the NIC. Small-message All-Reduce runs about 25% faster than RoCEv2 at 64 GPUs, but All-to-All has regressions.
Oct 238 min read


OCP Global Summit 2026 | Meta | Santosh Janardhan | Concrete to Code: AI Infrastructure at Gigawatt Scale
Meta reframes gigawatt AI around two clocks: how fast a site comes online and how often it breaks. Tents bring data centers up in 3-5 months, but even at target a 100k-GPU cluster still breaks about once an hour, so fault-tolerant software does the rest.
Oct 227 min read


Meta Platforms (META) Q3 2026 Earnings: Record $34.9B Quarterly Capex, Negative Free Cash Flow, and Muse Pushes Compute from Training to Inference
Meta's revenue grew 26% and the ad business is firing on volume and price; the GAAP profit drop is a settlement charge. The real signal is a record $34.9B quarter of capex and negative free cash flow, with Q4 implying $49–59B more. Muse is shifting Meta's compute demand from training to inference.
Oct 229 min read


OCP Global Summit 2026 | Meta | Nan Ge | Beyond Pluggables: Building the NPO Ecosystem
Meta's technical sourcing manager skipped the technology and explained the buying logic: power is a SerDes count, NPO pilots in scale-out where pluggables are the backup, and scale-up stays on copper. The hardest part is that NPO still has no standard.
Oct 227 min read


OCP Global Summit 2026 | Meta | Amin Yekani, Qing Wang | Operating LPO at Hyperscale: Fleet Reliability Data
Meta's first hyperscale LPO fleet data: removing the DSP cut interruptions 2x versus retimed optics, and 8.4x in the last 60 days. The bigger shift is the yardstick: interruption rate replaces failure rate, raising the bar for NPO and CPO.
Oct 227 min read


OCP Global Summit 2026 | Meta | Rongchun Zhou, Reza Niazmand | Inside Meta's 102.4T IceCube Switch
The hardest part of Meta's 102.4T IceCube switch is not the ASIC but 340 m of copper inside the box and 8,192 fibers in the rack. One shared main board lets Meta swap both the ASIC and the ODM.
Oct 217 min read


OCP Global Summit 2026 | Meta | Omar Baldonado | Betting on the Network: Meta's Case for Full-Stack, Open AI Networking
Meta is betting AI on the network: Ethernet scale-up via ESUN, 102.4T liquid-cooled switches with a two-tier fabric, and SRv6 on FBOSS across buildings. Open specs make it easier for white-box ODMs to get in, and harder to stay differentiated.
Oct 217 min read


OCP Global Summit 2026 | Cisco | Rakesh Chopra, Tom Wiliams | Breaking the Walls: Open Scale Across for the AI Era
Cisco and Meta promote cross-data-center networking to a third tier of the AI fabric. Meta's slides show 10,000s of 800G ZR per side, coherent plus DWDM beyond 10 km, while deep-buffer routing and multi-rail OLS return to the table.
Oct 217 min read


OCP Global Summit 1/6 | NPO Wins on Buying Power, Not Volume
At OCP Global Summit the buyers took the stage: Meta's sourcing, Microsoft's and Oracle's acceptance bars, Dell's OEM view. STT argues NPO wins on purchasing, not physics, and breaks down who buys and when.
Oct 202 min read


OCP Global Summit 2026 | Meta | Vimal Vasudevan, Ashutosh Grewal | Tahan: An Open, Ethernet-Based Scale-Up Switch for MTIA
Meta's Tahan is an Ethernet scale-up switch built by subtraction: same FBOSS, routing and multipath off. The switch holds about 98.5% of the latency budget, and next gen grows links 13.5x while per-accelerator bandwidth grows only 1.5x.
Oct 207 min read


OFC 2026 - AI Interconnect Scale-Up and CPO Reliability Deep Dive - Meta Platforms
When does copper run out? At OFC 2026, Meta answered with 90 million device-hours of field data: CPO cuts link power by 65% and, excluding one fixable component issue, is more than 10x more reliable than pluggable modules.
Oct 53 min read


OFC 2026 - DCI and Scale Across in the AI Era: A Full Breakdown - Meta, Google, Zayo, Lumen, Nokia
As AI training outgrows a single site, Scale Across is the new normal. How Meta defines the next generation of interconnect, why Google says "fiber is the new wavelength," and the strategy behind Lumen's push to add 41 million fiber miles.
Oct 54 min read


OFC 2026 - Ethernet's Accelerating Evolution for AI: 400G per Lane and the Road to 6.4T - Meta, Cisco, Synopsys, Dell
AI compute demand is forcing Ethernet to evolve faster than ever, and OFC 2026 marked the official start of 400G-per-lane work. Experts from Meta, Cisco, Synopsys and Dell'Oro explain how 1.6T and 3.2T networks will scale under hard power and space limits.
Oct 53 min read


Earnings Call Highlights: Corning (GLW) | FY26 Q1
Corning is no longer just a glass leader. Riding AI-driven optical demand and a surging solar business, it raised its Springboard growth plan through 2030 for the second time, as more hyperscalers sign on beyond Meta.
Oct 54 min read


Paper Analysis | Meta, Broadcom and AMD Define the OCI 200G Line Spec: A Scale-Up Bet on Bidirectional Single Fiber, Micro-Ring DWDM and External Lasers
Meta, Broadcom and AMD jointly define the OCI 200G line interface spec, betting scale-up optical interconnect on micro-ring DWDM, bidirectional single fiber, external lasers and NRZ. It is a buyer-supplier alliance taking on NVIDIA's vertical integration, and its external laser clause draws a clear spec for CW DFB and ELSFP suppliers.
Oct 57 min read


OFC 2026 - New AI/ML Interconnect Architectures: The CPO, NPO and OCI-MSA Power Play - Oracle, Meta, Microsoft, AMD, Arista
As AI models head toward Yottascale, optical interconnect architecture faces an unprecedented shift. From soldered to socketed, from external lasers to optical backplanes, how are Meta, AMD, Microsoft and other giants defining the next AI interconnect standard?
Oct 54 min read
SUBSCRIBE
Want the full supply-chain breakdown?
Subscribe to STT for premium articles: in-depth earnings-call reads, supply-chain maps and weekly industry insights.
bottom of page