Credo Technology is emerging as a high-growth AI infrastructure semiconductor story, with revenue surging to $1.335B in fiscal 2026 and profitability expanding sharply. The stock remains a Buy, but premium valuation and customer concentration keep the risk profile elevated.
Credo Technology Group Holding Ltd (CRDO) is one of the cleaner AI infrastructure growth stories in semiconductors: revenue scaled from $436.8M in fiscal 2025 to $1.335B in fiscal 2026, Q4 fiscal 2026 revenue reached $437.0M, and Q1 fiscal 2027 guidance of $465.0M to $475.0M points to another sequential step higher. That kind of growth would be impressive on its own. What makes it more investable is that Credo is pairing that top-line expansion with unusually strong profitability for a company still early in a product-cycle land grab, including 68.2% GAAP gross margin in Q4 fiscal 2026, $472.3M of FY2026 net income, and $1.4B in cash and short-term investments at year end.
The core bull case rests on three linked facts. First, Credo is already winning in active electrical cables and high-speed connectivity for hyperscale AI clusters. Second, management is extending that position into optical DSPs, ZeroFlap optical transceivers, PCIe retimers, and OmniConnect memory connectivity. Third, the company’s design philosophy is not just about raw speed. It is built around reliability, telemetry, and power efficiency, which matter when AI clusters are expensive enough that downtime becomes a budget item, not a technical footnote.
The main risk is valuation. CRDO trades at 128.0x trailing earnings, 48.1x forward earnings, and 39.6x EV/revenue. Those are premium multiples even in an AI market that has been generous with optimism. Customer concentration is another real issue. In fiscal 2025, one customer represented 67% of revenue, and in Q3 fiscal 2026 the top three end customers each contributed more than 10% of revenue. This is not a wide, slow-moving industrial franchise. It is a fast, concentrated, execution-sensitive semiconductor name.
For a balanced, moderate-risk investor with a medium-term horizon, the stock still earns a Buy rather than a Hold because the business momentum is strong enough to justify a premium. But that premium already prices in a lot of success. The right posture is constructive, not careless.
Get AI-powered research reports, daily market intelligence, and a personal analyst in your pocket.
Get Full Access →Credo Technology Group Holding Ltd (CRDO) is a fabless semiconductor company focused on high-speed connectivity solutions for optical and electrical Ethernet, PCIe, and AI data center applications. The company was founded in 2008, went public on January 27, 2022, and is based in Grand Cayman, with operations and customers spanning the U.S., Taiwan, Mainland China, Hong Kong, and other international markets. It had 622 employees, including 507 engineers as of May 3, 2025.
Its product set includes HiWire active electrical cables, optical PAM4 DSPs, line card PHYs, serializer/deserializer chiplets, SerDes IP licensing, PCIe retimers, and telemetry software. In plain English, Credo sits in the plumbing layer of modern AI and cloud infrastructure. It does not sell the GPU that gets the headlines. It sells the high-speed links and signal-conditioning pieces that help those systems talk to each other without wasting power or falling apart under load.
That niche is getting more valuable. AI clusters are scaling in bandwidth, density, and complexity. Credo’s 10-K says its products address 100G, 200G, 400G, 800G, and emerging 1.6T Ethernet markets, plus PCIe5 and upcoming PCIe6. Management’s recent messaging pushes the same theme harder: Credo wants to span the full connectivity fabric of AI infrastructure, from die-to-die and chip-to-chip links to rack-scale copper and facility-wide optical solutions.
Financially, the company has moved from development-stage losses to meaningful profitability at unusual speed. Annual revenue rose from $58.7M in fiscal 2021 to $106.5M in fiscal 2022, $184.2M in fiscal 2023, $193.0M in fiscal 2024, $436.8M in fiscal 2025, and then $1.335B in fiscal 2026. Net income swung from a $28.4M loss in fiscal 2024 to a $52.2M profit in fiscal 2025 and then to $472.3M in fiscal 2026.
Credo reports revenue across Product, License, and Product Engineering Services. In fiscal 2025, Product revenue was $412.2M, or 94.4% of total revenue, while License contributed $12.5M, or 2.9%, and Product Engineering Services added $12.1M, or 2.8%. That mix matters because it shows the business is now overwhelmingly driven by shipping physical products into production environments rather than relying on licensing revenue to smooth results.
The Product segment is the engine. It includes AECs, optical DSPs, retimers, line card PHYs, chiplets, and related connectivity silicon. In fiscal 2024, Product revenue was $145.0M, or 75.2% of total revenue. By fiscal 2025, it had nearly tripled in dollar terms and expanded sharply as a percentage of sales. That is the signature of a company moving from design wins into scaled deployment.
License revenue has become a smaller piece of the story. It fell from $28.0M in fiscal 2024 to $12.5M in fiscal 2025. Product Engineering Services also declined from $19.9M to $12.1M over the same period. Normally, shrinking service and license lines might look like a warning. Here, they look more like a sign that the center of gravity has shifted to product shipments, where the larger and more durable opportunity sits.
Inside the Product segment, management’s commentary points to three current growth pillars. The first is AECs, where Credo says demand is accelerating across hyperscalers and emerging neocloud providers. The second is ICs, including retimers and optical DSPs. The third is the early optical expansion through ZeroFlap optics, which management said is ahead of schedule and expected to begin a significant production ramp in the first quarter of fiscal 2027.
There is also a staged roadmap beyond the current revenue base. Management expects ALC products to sample and qualify in fiscal 2027 with production ramp in fiscal 2028, while the first OmniConnect gearbox is also expected to ramp in fiscal 2028. That gives Credo a layered growth profile: current AEC and IC scale, near-term optical ramp, and medium-term expansion deeper into memory and near-package connectivity.
Get AI research on any stock
Instant reports, daily intelligence, and an AI analyst in your pocket.
Credo’s flagship franchise today is its HiWire and ZeroFlap active electrical cable platform. The company’s 10-K describes HiWire AECs as plug-and-play copper interconnects for 100G, 200G, 400G, 800G, and emerging 1.6T data speeds. The practical value proposition is simple: lower power than optical alternatives, better density than DACs, and enough reach for the most important short-distance links inside AI racks and between racks.
Management’s strongest product claim is around reliability. Brennan said ZeroFlap AECs deliver up to 1,000x better reliability than commodity laser-based optics while consuming roughly half the power. That is a bold statement, but it is also central to Credo’s positioning. In AI clusters, reliability is not a marketing flourish. If a link flap stalls training or inference workloads, the cost is measured in wasted compute time and lower utilization of expensive accelerators.
The second flagship family is optical DSPs and transceivers. Credo’s recent product launches include Robin 800G optical DSPs, Cardinal 1.6T optical DSPs, and 800G ZeroFlap optical transceivers. These launches matter because they move Credo beyond the copper side of the rack and into the optical layers where AI network spending is also rising fast. The company is not abandoning copper. It is building a broader toolkit for whichever link type makes the most sense at a given distance and power budget.
The third flagship product line is PCIe retimers, especially the Toucan family and the Blue Heron 224G multiprotocol AI scale-up retimer. Toucan achieved PCI-SIG compliance at 32.0 GT/s in February 2026, and management said PCIe Gen6 retimers remain on track with fiscal 2026 design wins expected to convert to production revenue in fiscal 2027. Blue Heron supports UALink, ESUN, and Ethernet, which broadens Credo’s relevance as AI architectures diversify.
Credo’s competitive advantage starts with SerDes and DSP design. The 10-K says its proprietary mixed-signal and DSP architectures let it deliver leading-edge performance and power efficiency in mature process nodes, giving it an n-1 manufacturing advantage versus competitors that rely more heavily on bleeding-edge nodes. In semiconductors, that is the difference between building a race car and building one that can also be manufactured at scale without setting money on fire.
The company also has breadth within a narrow but valuable niche. It offers AECs, optical DSPs, retimers, line card PHYs, chiplets, SerDes IP, and telemetry software. Credo’s 10-K explicitly says it believes it is the only company in its industry offering a complete suite of high-performance connectivity solutions. Whether one agrees with that wording or not, the portfolio breadth is real and gives the company multiple ways to deepen customer relationships.
Telemetry and reliability are another edge. Credo’s PILOT software platform is designed to detect, diagnose, and resolve signal degradation before it affects workloads. Management says ZeroFlap optics combine hardware, optics, firmware, and PILOT software with switch-level SDK integration to detect and mitigate potential link flap events before they impact the cluster. That is a useful form of differentiation because it shifts the conversation from component specs to system outcomes.
The patent base adds some support. As of May 3, 2025, Credo owned 78 issued U.S. patents and 48 issued patents in mainland China, with additional pending applications in both markets. Patents are not a moat by themselves in semiconductors, but they are a sign that the company has built real technical depth rather than simply packaging commodity parts with a fresh label.
The planned DustPhotonics acquisition adds another layer. Credo said the deal would bring silicon photonics in-house and create a more vertically integrated stack spanning SerDes, DSP, silicon photonics, and system integration. The company also said the combined optical portfolio could generate more than $500M of optical revenue in fiscal 2027. If executed well, that would strengthen both product breadth and supply-chain control.
Credo runs a fabless model. In fiscal 2025, it used Taiwan Semiconductor Manufacturing Company for wafer fabrication, while Amkor and ASE handled packaging, KYEC and Sigurd handled testing, and BizLink manufactured AEC products. This is standard for a company of Credo’s size, but it creates a familiar tradeoff: lower fixed cost and capital intensity in exchange for dependence on a small set of external manufacturing partners.
The recent operating data shows the model is scaling well. In Q3 fiscal 2026, cash flow from operations reached a record $166.2M, CapEx was $26.5M, and free cash flow was $139.7M. Management said the CapEx was driven largely by purchases of production mask sets, which is exactly the sort of spending a fast-scaling semiconductor company should be making when demand is real and design cycles are moving.
Inventory has risen with demand. Q3 ending inventory was $208M, up $57.8M sequentially. That can be a warning sign in semiconductors when end demand is rolling over. Here, it sits alongside 52% sequential Q3 revenue growth, Q4 guidance of $425M to $435M that was later delivered at $437.0M, and Q1 fiscal 2027 guidance of $465.0M to $475.0M. In that context, higher inventory looks more like preparation for scale than a warehouse full of regret.
Supply concentration remains a risk. The 10-K notes exclusive use of TSMC for wafer production in fiscal 2025, and the company’s broader manufacturing footprint leans heavily on Asia. That is efficient, but it leaves Credo exposed to foundry bottlenecks, packaging constraints, and geopolitical disruptions around Taiwan and cross-border trade. The company’s growing cash position helps absorb shocks, but it does not remove them.
Credo operates in the AI data-center interconnect market, a subset of semiconductors that is growing faster than the industry average because AI clusters need more bandwidth, more signal conditioning, and tighter power budgets. Gartner said worldwide semiconductor revenue grew 21% in 2025 to $793B, with AI semiconductors accounting for nearly one-third of total sales. Deloitte also highlighted AI data-center networks scaling to 51.2 Tbps and above, forcing a rethink of copper, optics, and packaging choices.
For Credo specifically, the most relevant market is not the whole semiconductor sector. It is the connectivity layer inside hyperscale and AI infrastructure. The company’s own materials describe a multibillion-dollar data infrastructure market. More targeted external anchors are also useful. Credo cited an analyst projection that the scale-up networking market will exceed $40B by 2030, and said the silicon photonics PIC market is expected to grow to $6B by 2030.
That market is being reshaped by two engineering realities. First, AI workloads are driving transitions from 100G to 200G per lane and eventually 400G per lane. Second, power and reliability constraints are becoming as important as raw bandwidth. This is where Credo’s positioning makes sense. The company is not trying to be all things to all customers. It is targeting the parts of the interconnect stack where low power, signal integrity, and telemetry command a premium.
Copper and optical are both part of the opportunity. Management has been explicit that the market is evolving toward a heterogeneous mix of short-reach copper, pluggable optics, near-package optics, and eventually co-packaged optics. That is an important point because it keeps Credo from being trapped in a false either-or debate. The company can win where copper is best, and it is building products for the optical layers where copper reach becomes limiting.
Like what you're reading?
Get full access to AI-powered research reports, market analysis, and portfolio tools.
Credo sells to hyperscalers, OEMs, ODMs, optical module manufacturers, enterprise customers, and HPC buyers. The 10-K says the company has engagements with all major hyperscalers and more than 20 blue-chip clients. That is a strong customer list for a company of this size, but it comes with concentration risk that investors should not wave away.
In fiscal 2025, the top 10 customers accounted for about 90% of total revenue, and one customer accounted for 67% of revenue. In Q3 fiscal 2026, management said the top three end customers were each greater than 10% of revenue, and the largest customer represented 39% of revenue, the second largest 32%, and the third 17%. That concentration is extreme by most standards. It also helps explain the speed of the revenue ramp. When hyperscaler programs hit, they hit hard.
The customer mix is broadening, at least gradually. Management said AEC growth was driven by existing customers and new wins, including a fifth hyperscaler, and said ZeroFlap optics had begun production shipments with Tensor Wave while qualifying with three additional customers including hyperscalers and neocloud operators. That does not eliminate concentration, but it does show the company is not standing on one leg and hoping no one notices.
Ownership data also points to a market that takes the story seriously. Institutional ownership stands at 76.462%, insider ownership at 11.137%, and short interest is low at 0.0683% of float with a 1.73 short ratio. Large holders include BlackRock, Vanguard, Point72, State Street, and Wellington. That is not proof of quality, but it does indicate the name is well followed and widely owned by sophisticated capital.
Credo’s 10-K names Broadcom (AVGO), Marvell (MRVL), and Astera Labs (ALAB) as principal competitors, along with various cable suppliers. That list makes sense. Broadcom brings scale, broad platform reach, and deep customer relationships. Marvell competes across AEC DSPs, optical interconnect, Ethernet PHYs, and PCIe retimers. Astera is especially relevant in AI infrastructure connectivity and PCIe/CXL signal conditioning.
Credo’s advantage against those larger rivals is specialization. It is focused on the high-speed connectivity layer where power efficiency, reliability, and telemetry matter enough to influence architecture decisions. The company’s recent product launches, including 224G multiprotocol retimers, 800G DSPs, 1.6T DSPs, and ZeroFlap optics, show it is aiming at the fastest-growing parts of the market rather than trying to out-bloat the giants.
Its disadvantage is scale. Broadcom and Marvell have broader portfolios, larger R&D budgets, and more room to price aggressively or bundle solutions. In semiconductors, a focused specialist can win technical battles and still lose commercial leverage if a larger rival decides the category matters enough. That is why Credo’s execution speed and product differentiation have to stay sharp. A niche leader cannot afford to become merely good.
The company’s own framing of the market is useful here. Brennan pushed back on the idea that AECs and optical are direct substitutes in all cases, arguing instead for the right technology at the right reach and power budget. That is strategically smart. It positions Credo as an architect of mixed interconnect systems rather than a single-format evangelist. In a market full of absolutist narratives, that is refreshingly practical.
The macro backdrop is supportive. AI infrastructure spending remains the dominant growth engine in semiconductors. Gartner projected semiconductor revenue to exceed $1.3T in 2026, driven by AI processing, networking, and power. For Credo, that matters because its products sit directly in the network and interconnect layers that become more valuable as clusters get larger, denser, and more power constrained.
There are also clear geopolitical and trade risks. Credo’s manufacturing chain is concentrated in Asia, with TSMC as the wafer foundry in fiscal 2025 and packaging and testing partners in Taiwan and the broader region. The 10-K also notes that the business is subject to U.S. export controls, sanctions laws, customs regulations, and local trade rules. Management’s Q4 fiscal 2026 guidance explicitly said expectations were based on the current tariff regime, which remained fluid. That is a reminder that policy can move margins just as quickly as engineering can.
For a company with exposure to Taiwan, Mainland China, and global hyperscaler demand, geopolitical risk is not theoretical. It is embedded in the operating model. The good news is that Credo’s balance sheet gives it room to absorb disruptions better than a cash-poor peer. The less good news is that no amount of cash can make a foundry line in the wrong place suddenly local.
Credo ended fiscal 2026 with $1.4B in cash and short-term investments, giving it a strong liquidity cushion despite the rapid growth phase.
Unlock the full analysis
Subscribers get the complete breakdown — pick rationale, financial metrics, and recent earnings detail.
Get Full Access →Revenue climbed from $436.8M in fiscal 2025 to $1.335B in fiscal 2026, while GAAP gross margin held at 68.2% in Q4 fiscal 2026.
Unlock the full analysis
Subscribers get the complete breakdown — pick rationale, financial metrics, and recent earnings detail.
Get Full Access →Management guided Q1 fiscal 2027 revenue to $465.0M-$475.0M, signaling another sequential step up after a record fiscal 2026.
Unlock the full analysis
Subscribers get the complete breakdown — pick rationale, financial metrics, and recent earnings detail.
Get Full Access →CRDO trades at 128.0x trailing earnings, 48.1x forward earnings, and 39.6x EV/revenue, leaving little room for execution missteps.
Unlock the full analysis
Subscribers get the complete breakdown — pick rationale, financial metrics, and recent earnings detail.
Get Full Access →The report keeps a Buy stance because growth and profitability are strong enough to support a $195 fair value, even with premium multiples.
Unlock the full analysis
Subscribers get the complete breakdown — pick rationale, financial metrics, and recent earnings detail.
Get Full Access →Credo has moved out of the promising-upstart phase and into the stage where the numbers demand respect. Fiscal 2026 revenue of $1.335B, Q4 revenue of $437.0M, $472.3M of annual net income, and $1.4B in cash and short-term investments put real weight behind the narrative. This is not a concept stock floating on AI fumes. It is a profitable, scaling semiconductor company with a clear role in the buildout of next-generation data-center connectivity.
The investment case comes down to a simple judgment. If Credo continues to convert hyperscaler demand into broader platform adoption across AECs, optical DSPs, ZeroFlap optics, and PCIe retimers, the company can keep growing into its premium. If growth stumbles or customer concentration bites, the stock has enough valuation risk to punish complacency. That is why the right stance is positive but selective.
For medium-term investors, CRDO remains a Buy. The business is strong, the roadmap is credible, and my fair value estimate of $195 leaves room for upside from levels that do not already assume a flawless future. In this market, that counts as a rare form of discipline.
Credo Technology Group Holding Ltd (CRDO) falls sharply after earnings despite posting explosive revenue growth and strong margins. The selloff appears driven by high expectations and valuation, as investors wanted even stronger guidance from the AI networking leader.
These seven utilities stand out for regulated exposure, infrastructure investment potential, and the financial quality needed to benefit from the sector’s capex-driven growth cycle.
New home sales and mortgage applications are weakening under elevated borrowing costs, but construction remains resilient and price growth is cooling. The latest data point to a housing market that is slowing in an orderly way, with softer demand and less inflationary pressure rather than a sharp downturn.