The Optical Tell — Huang's Copper-vs-Optics Call Repriced the Photonics Layer the Digest Named on 31 May, While Credo Grew 157% and Still Fell

The Lead

Issue 09 ran on 31 May. Its photonics thread argued that the AI thesis was migrating off the chip and into the connectivity layer — that when “the GPU monopolist and the leading ASIC challenger both move on photonics in the same five days, the connectivity thesis stops being a contrarian’s hedge and becomes consensus infrastructure.” Marvell had just closed its Celestial AI photonic-interconnect acquisition and booked the charge; the press was carrying “Nvidia Bets Big on Photonics.” We named the layer. Two trading days later, the market repriced it in a single session.

On Tuesday, Jensen Huang used a Computex appearance in Taipei to make the optical case explicit: copper, he said, “has its limits” — “you use optics wherever you must, you use copper wherever you can.” Coherent rose roughly 16–17% to an all-time closing high near US$426; Lumentum gained 13%; Corning, which makes the fibre, climbed 13%; Applied Optoelectronics added 8%; and Marvell caught a fresh bid on the same tape, with Huang reportedly floating it as a potential “next trillion-dollar company.” This was not a rumour-driven melt-up. It sat on top of hard capital: over roughly three months NVIDIA has put about US$2 billion each into Lumentum and Coherent, US$500 million into Corning, and joined a US$500 million round in Ayar Labs — on the order of US$6.5 billion committed to the optical layer. When the monopolist spends six billion dollars wiring the next generation and then tells you on stage why, the market listens.

The discipline arrives in the same breath. The exact same shift — optics displacing copper — is a question mark over the one copper-centric name in our universe. Credo Technology reported Monday night, grew quarterly revenue 157% year-on-year to $437 million, beat on every line, guided above consensus — and the stock fell about 10%. Its core franchise is the copper active-electrical-cable (AEC) business that Huang’s remark implicitly puts on a clock. Growth was never the question. Priced growth, in the wrong part of the stack, was.

So this is an interim issue, triggered by the repricing rather than the calendar. The two verdicts Issue 09 promised — Broadcom’s custom-silicon number and Palo Alto’s free-cash-flow read — both land at or just after press time: Broadcom reports tonight. The bars are set below, and the forecasts stay open, honestly, until the prints are in. What we can settle is the theme, the counter-tape, and a candid accounting of what this system is actually worth to someone trading it.

Threads

The optical melt-up — Huang’s copper-vs-optics call repriced the photonics layer

The cleanest way to state what happened: a thesis the Digest had been building for three issues went from “early and slightly contrarian” to “consensus” in one session, on a senior executive’s framing plus the largest customer in the industry voting with its balance sheet.

Name	What it is	Tuesday move	The read
COHR	Coherent — vertically integrated optical components/materials	~+16–17%, all-time closing high ~$426	The platform player; owns materials peers buy. NVIDIA ~$2bn invested.
LITE	Lumentum — optical transceivers/lasers	+13%	The agile transceiver play; NVIDIA ~$2bn invested; 200G lasers into NVIDIA’s CPO platform.
GLW	Corning — optical fibre	+13%	Makes the fibre itself; NVIDIA $500m for US capacity expansion.
AAOI	Applied Optoelectronics	+8%	The higher-beta, lower-quality optical proxy — rode the broadening.
MRVL	Marvell — custom silicon + optical interconnect (Celestial)	fresh bump	Huang’s “next trillion-dollar company” framing; our largest verified anomaly this window (z = 7.2).

Three things make this more than a one-day pop:

First, it is funded, not just talked about. The ~US$6.5bn NVIDIA has committed across Lumentum, Coherent, Corning and Ayar Labs in three months is the substantive signal; Huang’s Computex line is the narration. Capital commitments of that size are a multi-year supply-chain bet, not a trading catalyst, and they are already on the filings/press record — which is precisely the kind of thing this system is built to catch and connect.

Second, it is the same arc Issue 09 traced. Marvell bought a photonic-interconnect company (Celestial) and took the charge; NVIDIA was already funding the optical suppliers; the press flagged “Nvidia Bets Big on Photonics.” Issue 09’s call was that the constraint was moving from compute to bandwidth. Tuesday was the market agreeing, loudly. The honesty caveat is in the next thread — being directionally early on a layer is not the same as a backtested, repeatable signal, and I will not pretend it is.

Third, the structure has a built-in loser. Optics-over-copper is not a rising tide that floats every connectivity name. It is a rotation — and rotations have a short side. That short side, this week, was Credo.

Forecast [2026-06-02-001]: Through Broadcom’s print and the five trading days after, the optical pair (COHR, LITE) outperforms the copper-AEC name (CRDO). Confidence 0.55, horizon 7 days. This is the falsifiable version of the whole thesis: if “optics over copper” is real and sustained, the spread holds; if Tuesday was a one-day overshoot, it closes.

157% growth and the stock fell — Credo is the counter-tape, and the copper-AEC casualty

Credo Technology reported fiscal Q4 2026 after Monday’s close. The numbers, on their own, are extraordinary:

Metric	Q4 FY26	Detail
Revenue	$437M	+157% YoY (from $170M); beat $433M consensus
Adjusted EPS	$1.16	beat $1.03 consensus
Q1 FY27 guide	$465–475M	above ~$465M consensus
FY26 revenue	~$1.3B	roughly tripled year-on-year
FY26 non-GAAP net income	$662M	more than five-fold increase

And the stock fell ~4% in the session, then a further ~10–12% after hours. Two forces, and both matter for how you read the watchlist:

The priced-for-perfection force. Credo trades around 38× forward earnings after a parabolic run. The market is no longer paying for growth; it is paying for acceleration of beats. Q3 was +201%; Q4 was +157%; the magnitude of the beat is compressing even as the absolute growth stays spectacular. GAAP operating margin slipped from 36.8% to 35.7% quarter-on-quarter. When expectations are stretched that far, an in-line-to-slightly-above print is a sell. This is the same profitability/expectations filter that took Zscaler down 19% (Issue 09) — the AI complex is increasingly graded on clean beats and margins, not headline growth.

The thematic force — and this is the one that ties the issue together. Credo’s core is copper AECs. The bear case, in the market’s words, is that “the industry is moving to optics… and Credo’s copper-based AEC business is cooked.” Huang’s Tuesday remark is a direct articulation of that bear case from the most authoritative possible source. Credo is not standing still — management guided the combined optical business above $500m next fiscal year and total growth above 75% for FY27, explicitly pivoting from a single-product AEC company to an end-to-end connectivity platform across copper, mid-reach optical and co-packaged optics. But the timing was cruel: it reported a copper-heavy quarter into the week the market fell in love with optics.

The watchlist read: CRDO stays bullish on fundamentals but moves to “transition-watch.” The franchise is pivoting in the right direction; the question is whether the optical ramp outruns the AEC de-rating. That is exactly what Forecast [2026-06-02-001] tests.

The verdicts pending — AVGO tonight, PANW’s FCF filter — bars set

Issue 09 promised this issue would carry the Broadcom custom-silicon verdict and the Palo Alto FCF read. Honest status: both print at or just after press time, so both forecasts stay open. What I can do is set the bars precisely, so the resolution in the next update is unambiguous.

Broadcom (AVGO) — reports tonight, 3 June, after the close. Forecast [2026-05-31-001] said AI/custom-silicon revenue would grow more than 50% YoY and be the fastest of the fortnight’s ASIC reporters. The setup:

• Q2 revenue guided to ~$22.0bn (+47% YoY) — already a beat of the ~$20.6bn consensus that stood before the guide.
• AI chip revenue guided to $10.7bn for the quarter — a 27% sequential jump from Q1’s $8.4bn, and Q1 itself was +106% YoY. On any reasonable prior-year base, the AI line clears the 50% bar comfortably if the guide is met.
• CEO Hock Tan has claimed “line of sight” to >$100bn of AI chip revenue in 2027; the AI-switch backlog alone is >$10bn.
• The market is pricing an ~8% one-day move. Consensus EPS ~$2.40 (+52% YoY).

What “busts” it: AI revenue printing below the guide, or management walking back the 2027 trajectory. What confirms it: the $10.7bn AI number lands or beats, and the custom-silicon (Google/Meta ASIC) commentary stays on the >$100bn-2027 glide path. The verdict, and the grade, come in the next update.

Palo Alto (PANW) — reported ~2 June, after the close. The Issue 09 question was whether the free-cash-flow profitability filter that hit Zscaler (–19%) also bites the cybersecurity leader. The frame to grade against: FY26 guidance of 37% adjusted FCF margin, non-GAAP operating margin 28.5–29.0%, and the CyberArk/Chronosphere integration that management has already flagged will compress second-half margins. If PANW holds the 37% FCF line and RPO growth (was $16.0bn, +23%) stays intact, the ZS de-rating was idiosyncratic. If the FCF margin guide slips, the “FCF is the new growth filter” thesis generalises across the whole AI-software complex. Result pending at press time.

The positioning hasn’t moved — and an honest read on what this is worth to a trader

The convergence layer (the 11-fund 13F universe) is essentially unchanged from Issue 09, which is itself the signal: the smart money is still concentrated at the top of the stack.

Ticker	Convergence	Note
NVDA	62 (highest)	Still the most crowded long in the universe.
GOOG	51	Six-fund consensus; unchanged.
COHR	39	Now high in convergence — see the caveat below.
LITE	29	Same.
META	25	Crowded.
DLR / EQIX	20 / 6	The physical-layer bet; still early, still one-sided (Forecast [2026-05-31-002]).
CRWV	9	Contested; riskiest expression.

One honest caveat on COHR/LITE’s convergence scores: they are measured after Tuesday’s move, and the press feed feeds the salience layer. A high score on an optical name this week partly reflects the rally it is supposed to anticipate. Treat the positioning read on the optical pair as confirmation, not as an independent forward signal — the forward signal was the 31 May thesis and the NVIDIA capital, both of which predate the pop.

Since Issue 09 (31 May)

Thread	Issue 09 read	Now
Photonics / connectivity layer	”The constraint is moving from compute to bandwidth; the connectivity thesis becomes consensus.”	Confirmed, hard. Huang’s Computex optical remarks + ~$6.5bn NVIDIA optical capital → COHR ~+17% (record), LITE +13%, GLW +13%. Early on the layer; n=1.
CRDO (connectivity IC)	“bullish — Celestial validates the layer”	Grew 157%, stock fell ~10%. Right that the layer matters; the copper-AEC exposure is the wrong side of optics-over-copper. → transition-watch.
AVGO custom silicon	”cleanest read; reports 3–4 June; Forecast >50% YoY”	Reports tonight. Guide implies the bar is cleared if met; pending.
ZS FCF filter / PANW	”watch PANW for the peer read”	PANW reported ~2 June; FCF-margin guide is the thing to grade; pending.
DLR/EQIX 13F (Forecast 002)	“fund adds a DC-REIT in Q2 13Fs”	Resolves mid-August; open.

Watchlist Updates

Ticker	Direction	Δ vs prior	Driver
COHR	bullish	↑	~+17% to record high on Huang optical remarks; NVIDIA ~$2bn invested; the platform/materials player.
LITE	bullish	↑	+13%; NVIDIA ~$2bn; 200G lasers into NVIDIA’s CPO platform.
GLW	bullish	NEW	Corning — the fibre; +13%; NVIDIA $500m US-capacity commitment. The lowest-beta way to own the optical build.
MRVL	bullish	→	Fresh bump on Huang’s “next trillion-dollar” framing; largest verified anomaly this window (z=7.2). Celestial photonics thesis intact.
CRDO	bullish → transition-watch	↓ (tone)	157% growth, stock –10%: priced-for-perfection + copper-AEC overhang. Pivot to optical (>$500m next FY) is the swing factor.
AVGO	bullish	→	Verdict tonight. Guide implies AI line clears the >50% bar if met. The custom-silicon read of the fortnight.
PANW	watch	→	FCF-margin filter is the test; CyberArk integration compresses H2 margins by design. Pending.
NVDA	bullish	→	Still the most crowded long; now also the optical-ecosystem financier (~$6.5bn). Marginal-buyer risk unchanged.
AAOI	bullish (high-beta)	NEW	+8% — rode the optical broadening; the lower-quality proxy. Flagged, not endorsed.

The Week Ahead

Date	Event	What confirms / what breaks
Tonight, 3 Jun	AVGO Q2 FY26	Forecast [2026-05-31-001]. AI chip revenue ≥$10.7bn guide and the >$100bn-2027 glide path = confirm; a walk-back or a sub-guide AI line = bust.
Now–next session	PANW FQ3 read	37% adj FCF-margin line held = ZS was idiosyncratic; slip = the FCF filter generalises.
3–4 Jun	CIEN, MDB, HPE	CIEN is the optical-transport confirmation of the photonics theme — the read-through to COHR/LITE.
9–10 Jun	ORCL Q4 FY26	RPO backlog = the cloud-capex demand signal feeding the DLR/EQIX physical-layer thesis (Forecast 002).
+7 trading days	Forecast [2026-06-02-001]	COHR/LITE vs CRDO spread — does “optics over copper” hold or fade.

The next update resolves Broadcom (Forecast [2026-05-31-001]) and the PANW FCF read.

Methodology and Disclosures

Filings Intel Digest is a bona fide financial publication. Nothing here is personalised investment advice. The editor may hold positions in companies discussed. All claims are sourced to publicly filed documents or identified press sources. Calibration is partial — treat any forward statement as a hypothesis, not a forecast.

Frozen as of close-of-trading Monday 2 June 2026 (US), an interim issue covering 31 May – 2 June, triggered by the optical repricing rather than the weekly calendar. No new primary filings were ingested in this window beyond the MRVL 8-K already covered in Issue 09; the automated signal-extraction pipeline returned 0 structured signals. The optical melt-up figures (COHR, LITE, GLW, AAOI moves; NVIDIA’s optical investments; Huang’s remarks) are press-sourced (TradingView/Invezz, Barron’s, and corroborating coverage) and directional pending each company’s own disclosure. The Credo financials are from the company’s FQ4 2026 release (1 June, after close) as reported by Stocktwits/Seeking Alpha — verify against the 8-K/Exhibit 99.1. The Broadcom and Palo Alto figures are guidance and consensus only; both report at/just after press time and their forecasts remain open. Positioning is from the 13F convergence layer (Q1 2026, 11-fund universe), with the COHR/LITE post-rally caveat noted in-thread.

Watchlist now tracking 11 funds. The optical names (COHR, LITE, AMAT, LRCX, CRDO) were added to the tracked universe on 31 May; GLW and AAOI enter this issue.