Google reviews five TPU generations: an architecture set in 2017 survived the Transformer takeover and scaled 3600x. For optical readers, the key line: OCS, one of only two designs still unique to TPU, is what keeps a near-10K-chip machine alive, schedulable and deployable in stages. Power has now replaced cost as the new ceiling.

This week the market read SemiAnalysis's CPO-delay report as a death knell for optical components. It read it backwards. If CPO ramps a year later, pluggable optical modules — and the upstream EML lasers already sold out through 2027 — get an extra year of earnings. Demand hasn't weakened; only the oversold share prices have.

ZR is the meat, Coherent Lite is the bone. 2026 is the year ZR generates cash; 2028 is when Coherent Lite truly ramps. Three signals could reverse everything: whether Amazon commits, how fast LPO/CPO win sockets, and production-grade demos at OFC 2027. Taiwan's real opportunity lies in upstream components, not the DSP.

Scale-up optical interconnect looks like a seven-way melee, but there are really only two strategies, Fast-and-Narrow and Slow-and-Wide, each hitting a symmetric wall. Learn the yardstick first, then see how the seven approaches layer and coexist.

This week the bull case for optical communications got a new spokesperson: Broadcom's income statement instead of analyst decks. Demand is now locked in by earnings, but what's truly scarce can't be bought with earnings: the hardest bottleneck in the chain is a laser in short supply through 2027. The more real the demand, the more valuable the light source.

For a decade, photonics packaging was the last mile of assembly that nobody talked about. AI has pushed it center stage, with the market set to grow from $4.5B to $14.4B in six years — driven not by transceivers but by three new tracks opening at once: CPO, AR microLED and FMCW LiDAR. Part 1 of the series lays out the big picture.

Google's eighth-generation TPU splits into two architectures for the first time. STT breaks down how TPU 8t removes the training I/O bottleneck, and how TPU 8i uses large on-chip SRAM and OCS optical circuit switching to cut MoE inference network hops by 56%.

NVIDIA proposes a symmetrical, wavelength-agnostic bidirectional silicon photonics link that uses polarization multiplexing to break CPO's port-count and cost bottlenecks. Error-free 32 Gbps bidirectional operation makes "colorless" optics a reality and paves the way for scaling AI cluster networks.

As copper can no longer keep up with the data movement demands of AI clusters, Marvell answers with its Photonic Fabric optical platform. This is not just a speed upgrade, but an architectural shift that breaks through the memory wall and the power bottleneck.

In the era of gigawatt AI factories, the network defines the computer. NVIDIA unveiled 3D-stacked silicon photonics CPO and industry-leading MRM technology to break through the power and reliability limits of AI scaling, and laid out a decade-long roadmap fusing compute and optics.

Copper's physical limits have become the bottleneck to scaling AI. At OFC 2026, Broadcom showed its latest CPO and OCI progress - ultra-low 6 pJ/bit power and exceptional reliability - signaling that optics will dominate both scale-out and scale-up networks.

The AI compute race has entered its second half, and bandwidth and power are now make-or-break. Intel's OCI chiplet at OFC 2026 signals that the convergence of photonics and electronics at the chip level has arrived, rebuilding data center architecture rather than merely improving specs.

Three scale-up MSAs launched at OFC 2026, and the loudest, XPO, was championed by Arista co-founder Andy Bechtolsheim. But Meta's 90 million hours of data took apart its core selling point. Is XPO a bet on extending pluggables' life, or the real answer?

When the biggest buyer reaches upstream and index providers add optical names to their benchmarks, the industry's capital structure has shifted gears. NVIDIA's $2B stake in Coherent is not an investment; it is supply insurance.

AI traffic is reshaping network architecture. Nokia's Q1 highlight was not traditional telecom but 49% revenue growth from AI/cloud customers. With its new San Jose InP fab about to come online, how is Nokia using vertical integration to pull ahead in the 800G wave?

Silicon photonics gets the spotlight, but Intel's 0.9 pJ/b 108G PAM4 link shows VCSELs remain an underrated low-power path for short-reach CPO. We walk through this CICC 2026 paper section by section, figure by figure, and identify the last three miles VCSEL CPO must cover to reach volume: equalization, packaging and thermals.

CPO shifts value from module assembly to advanced packaging, and the biggest winners are not legacy module makers but foundries and OSATs with silicon photonics and advanced packaging capabilities. Taiwan's ecosystem is regrouping around them, and standardization will decide the outcome. But pluggables are still growing, and the steering wheel remains in the hands of the platform giants.

Earnings, timelines and shipment volumes all confirmed the optical bull case in the same week, yet Taiwan's silicon photonics stocks hit limit-down. This is a valuation correction, not a demand alarm, and the line between the two hinges on one sentence in next week's Broadcom earnings call.

While the market waits for CPO, Google wrote its answer into 134,000 TPUs with Virgo Network. What holds this fabric up is not exotic packaging but an OCS optical switch six years in the making, and the winners are pluggable optical modules and their upstream suppliers.