At 1:05pm today, Credo announced the acquisition of DustPhotonics, a fabless silicon photonics company developing Photonic Integrated Circuits for high-speed optical transceivers. Per Credo’s press release, the deal “will bring industry-leading Silicon Photonics PIC technology in-house, expanding Credo’s addressable market and deepening its optical interconnect portfolio across 800G, 1.6T, and 3.2T NPO and CPO,” positioning Credo with a vertically integrated connectivity stack spanning SerDes, DSP, silicon photonics, and system integration. The deal terms: $750 million cash plus approximately 0.92 million Credo shares upfront, with up to 3.21 million additional contingent shares based on financial milestones. Total upfront consideration is approximately $870 million at current CRDO levels, with a maximum deal value approaching $1.3 billion if the contingent shares vest in full. Credo expects the transaction to be accretive to non-GAAP EPS in fiscal 2027, with closing in Q2 calendar 2026 subject to customary regulatory approvals.
Seventeen hours earlier, at 8:07pm Pacific last night, I published Credo Was the Wrong Question. The piece argued that Credo had already transitioned from a copper-and-AEC incumbent into a system-level optical product company, attacking the optical transceiver TAM historically owned by Coherent, Lumentum, AAOI, and Marvell. The follow-up framework was built on the Gilad Shainer GTC footage, the reader-corrected NVIDIA roadmap, and the ZeroFlap 800G general availability from March 17. The piece called the direction. Credo just confirmed it with a balance-sheet decision less than a day later, using language that matches the framing almost word-for-word.
The Framework Called It
From last night’s piece, in the ZeroFlap section: “Credo is no longer a copper-and-AEC company defending its turf against optics. Credo is now selling optical transceivers directly. The company has moved up the stack from SerDes, DSP, and AEC supplier to a system-level optical product company.”
From Credo’s press release at 1:05pm today: “vertically integrated connectivity stack spanning SerDes, Digital Signal Processing (DSP), Silicon Photonics and system integration.”
The two sentences are saying the same thing, one as analysis and one as corporate strategy. The only difference is that the follow-up piece listed the layers Credo was moving up (SerDes, DSP, AEC, system integration) and Credo’s announcement adds the missing layer between DSP and system integration: silicon photonics. That is the layer DustPhotonics fills. That is the capability gap that would have kept Credo as a reliability-differentiated reseller of commodity PICs rather than a margin-owning vertical integrator. Credo just closed the gap by acquisition.
From three layers to four. DustPhotonics fills the capability gap between DSP and system integration. Credo now controls every layer of the optical transceiver stack end-to-end, the same vertical integration Lumentum, Coherent, and Marvell have at their respective layers.
What DustPhotonics Actually Brings
DustPhotonics is an Israeli silicon photonics company based in Modi’in, founded specifically to build merchant PICs for hyperscale datacenter optics. Per Cignal AI’s optical component startup tracker, the company’s differentiator is a proprietary Low Loss Laser Coupling (L3C) technology that allows a single laser source to drive four optical channels, dramatically improving the laser-to-PIC coupling efficiency that has historically been the limiting factor in silicon photonics power budgets. Their flagship products include the Tamar200 (the industry’s first merchant 1.6T 2xFR4 PIC, announced at OFC 2025) and the Carmel chip, both of which use proprietary low-loss Mach-Zehnder Modulator architectures that eliminate the need for thermoelectric coolers, which is the kind of system-level efficiency improvement that compounds at hyperscaler cluster sizes.
The product portfolio matters because it maps directly onto Credo’s existing roadmap. DustPhotonics builds PICs for 800G, 1.6T, and the next-generation 3.2T optical transceivers, which are the exact speeds Credo’s ZeroFlap line supports today plus the speed transition that defines the 2027-2028 deployment window. Even more pointed: Credo’s Cardinal 1.6T optical DSP family, announced in March 2026, was already engineered with what the press release described as “high-swing integrated EML and silicon photonics drivers.” Credo’s newest flagship DSP was already designed to drive silicon photonics PICs. The DustPhotonics acquisition closes a roadmap loop that was already partially built at the chip design level a month ago.
One more detail worth surfacing because it ties directly back to the Photonic Divergence framework. DustPhotonics’ own product positioning at OFC 2025 explicitly described their PIC products as designed to “play a significant role in relieving some of the ongoing EML supply chain constraints in the industry.” The EML supply constraint is the structural argument I laid out in The Great Photonic Divergence: “At 200G per lane, Electro-absorption Modulated Lasers (EMLs) become the required technology.” EML supply has been the binding constraint that gave Lumentum its laser moat, and DustPhotonics’ silicon photonics with integrated laser approach is one of the credible architectural alternatives at the rack level. Credo just bought it. This is the layer below the Lumentum laser supply, not in tension with it.
One detail that deserves naming because semi readers will recognize it: DustPhotonics’ chairman is Avigdor Willenz. Willenz founded Galileo Technology, sold it to Marvell for $2.7 billion in 2000, then founded Annapurna Labs and sold it to Amazon for $370 million in 2015. Annapurna is the foundation of the entire AWS Graviton and Nitro silicon stack today. Willenz has built and exited two of the most consequential semiconductor companies of the last twenty-five years, and he chose to put his name on DustPhotonics. That is not a vanity board seat. It is one of the highest-credibility signals in semiconductor M&A, and it is part of why a 70-person fabless company with seven years of operating history just commanded an $870 million upfront price.
What Credo Just Told the Market About Optical Revenue
The single number from this morning’s release that institutional readers should be modeling against everything else: Credo expects its combined optical portfolio of ZeroFlap Optical Transceivers, Optical DSPs, and Silicon Photonics products to generate greater than $500 million in optical revenue in fiscal 2027. That is not a Jefferies estimate. That is not a sell-side model. That is the company’s own forward guidance, in a press release, on a deal day, with a follow-up conference call tomorrow morning. For context, Jefferies’ BUY initiation on Sunday afternoon framed ZF Optics as a $300 million-plus revenue opportunity by C27. Credo just told the market the combined optical business will be more than $500 million in their fiscal 2027 — a higher number than the sell-side initiation that triggered the rally Sunday night, and one that includes the contribution Jefferies could not have modeled because the DustPhotonics deal had not been announced yet.
Credo also disclosed that, per LightCounting and Credo’s own estimates, the silicon photonics PIC market is expected to grow to $6 billion by 2030. That is the TAM Credo is now buying its way into at the manufacturing level rather than just participating in at the integration level. Even single-digit share of a $6 billion category is meaningfully accretive to a company doing roughly $1.6 billion in annual revenue today.
Brennan and Gilad Are Saying the Same Thing
From Bill Brennan, Credo’s Chairman and CEO, in this morning’s press release: “We are building a vertically integrated connectivity platform that spans from copper to optical and from chip to cluster — allowing us to solve for the two constraints that matter most at scale: reliability and power efficiency, while deepening our role as a strategic partner to our customers.”
Now read what Gilad Shainer told me on camera at GTC last month, which I published in last night’s follow-up: “Power is the key element that determines my compute capacity in a given data center. I want to minimize power forever. I can.”
Brennan and Gilad are framing the same constraint from two different sides of the supply chain. NVIDIA’s networking SVP says power per compute is the binding constraint of the gigawatt buildout. Credo’s CEO says reliability and power efficiency are the two constraints he is solving for at scale, and naming them in that order in a press release is not accidental. ZeroFlap solves for reliability. The DustPhotonics PIC integration unlocks the power efficiency leg, because silicon photonics with integrated laser coupling consumes meaningfully less power per bit at the rack-level distances Credo addresses than EML-based pluggable modules do. The supplier and the customer are now publicly aligned on the same two-variable optimization problem.
Camp One Just Got Tighter
Last night’s piece repositioned Credo inside the Optical Cold War framework from Friday, arguing that the corrected view “does not put Credo in tension with the Lumentum thesis. It puts both companies inside Camp One, playing at different layers of the same architecture, benefiting from the same NVIDIA-anchored stack.” The DustPhotonics acquisition tightens that Camp One placement at the supply-chain layer. As I wrote in The Foundry Behind the Photonic Boom in March, “Lumentum makes the light. Tower makes the chip that uses it.” Credo acquiring a SiPho PIC design company means Credo now plugs into the same foundry backbone that Lumentum, Coherent, and NVIDIA’s optical engine partners already use. Camp One is no longer a framework claim from last night. It is a supply-chain fact as of this afternoon.
One of the named voices in the press release reinforces the same point and surfaces a forward signal that is worth flagging on its own. Gavin Baker, Managing Partner and CIO of Atreides Management, was a DustPhotonics investor before the deal and contributed a quote to the announcement. The line that matters for forward modeling: Atreides described the combination as building “on strong momentum across Credo’s DSP, ZF Optical Transceiver platform, and future Active LED Cable (ALC) product lines.” The phrase Active LED Cable has not been a public product line for Credo before. ALC is what the Hyperlume microLED acquisition was supposed to become. Last night’s piece flagged Hyperlume as a wildcard in the bear case. Atreides just publicly named it as a future Credo product line, which is the first market-facing confirmation that the Hyperlume integration is shipping into a productized roadmap rather than sitting as IP on the shelf. Credit where it is due: Atreides was early on DustPhotonics, called the architectural direction correctly, and the disclosure of the ALC product line by a named investor in a Credo press release is meaningfully more informative than a generic strategic-rationale paragraph would have been.
The Bear Case, Updated
The bear case I am still carrying has changed shape now that the deal terms are public. The price is real. Approximately $870 million in upfront consideration for a 70-person fabless company with seven years of operating history is a meaningful outlay against Credo’s roughly $1.6 billion annual revenue base, and the contingent earnout structure of up to 3.21 million additional shares means the maximum deal value approaches $1.3 billion if the milestones vest in full. Integration risk on a silicon photonics process engineering team is real, even with Avigdor Willenz on the board to anchor the institutional credibility. The fiscal 2027 accretion guidance means Credo is modeling first-year contribution rather than long-tail integration, which is bullish but creates a near-term execution clock that is shorter than most M&A integrations get. If integration runs late or the hyperscaler design wins do not translate into revenue at the pace Credo is modeling, the fiscal 2027 accretion call slips and the stock gets repriced. None of this breaks the framework. All of it is execution risk that the conference call tomorrow morning at 10am Pacific is the first opportunity for management to address directly.
So What?
Five independent events. One conclusion. The reader, the sell-side, the primary source video, the stock price, and the company itself all arrived at the same conclusion inside one news cycle.
The compression loop from Friday to this afternoon is worth laying out because it is the tightest sequence this publication has had. Friday afternoon the Hurlston piece published with the original Credo section flagged as “the position I’m reassessing.” Sunday morning a paid subscriber sent substantive technical pushback. Sunday evening the Friday piece was corrected. Sunday 4pm Pacific Jefferies initiated CRDO BUY at $175. Sunday 8:07pm Pacific the follow-up published with exclusive GTC footage and a two-TAM framework. Monday morning CRDO rallied on the initiation. Monday 1:05pm Credo announced the DustPhotonics acquisition.
Five independent events, one conclusion. From last night’s So What: “The execution is the signal. The narrative is already priced in.” The narrative got more priced-in this afternoon, and the execution signal now has a second leg: DustPhotonics integration timing and the combined entity’s hyperscaler qualification pipeline beyond Oracle. The next discrete catalyst is Credo’s conference call tomorrow morning, Tuesday April 14 at 10am Pacific, where management will walk through the deal economics, the integration plan, and the basis for the fiscal 2027 accretion guidance. That call is the first opportunity for management to address the execution clock the deal terms imply, and it is the first opportunity for the sell-side to update models on the combined entity. Holding with renewed conviction. Considering adding into any pullback. The framework holds, and the framework is now a balance-sheet fact.
Related BEP Research
Credo Was the Wrong Question — Last night’s follow-up with the Gilad Shainer GTC exclusive, the two-TAM framework, and the corrected Optical Cold War framework placement.
The Foundry Behind the Photonic Boom — The Tower Semiconductor deep dive on silicon photonics PIC manufacturing and why Tower sits at the center of the independent photonics ecosystem.
Credo’s Q3 Call Answered Every Question — February deep dive on the $407 million quarter, the AEC runway, and Bill Brennan’s “early in its AEC adoption” framing.
Credo conference call on the DustPhotonics acquisition: Tuesday April 14, 2026 at 10:00am Pacific Time. Dial-in: (800) 715-9871 toll-free, +1 (646) 307-1963 international. Conference ID: 5273210. Live webcast at investors.credosemi.com.
Disclosure: I hold positions in NVDA, LITE, CRDO, TSEM, and others disclosed in my portfolio. I am considering opening a position in MRVL. I do not hold positions in COHR, AAOI, AVGO, or GFS. This is not investment advice — do your own research.
Leave a Reply