AAOI versus Credo: who keeps the profit from AI connectivity?

AAOI and CRDO invite the same investment story: larger AI systems need better connections, so the companies supplying those connections should prosper. The important question is what shareholders receive from that prosperity.

My assessment favors Credo on demonstrated business economics. AAOI offers a more demanding manufacturing transformation, with potentially substantial upside if capacity becomes profitable output. Neither conclusion establishes which stock offers the better return at today's price. The distinction is between the quality already visible in the accounts and the improvement an investor must still underwrite.

This comparison uses Applied Optoelectronics' calendar Q2 2026, ended June 30, and Credo's fiscal Q1 2027, ended August 1. The quarters differ in timing; they are the latest reported periods reviewed here, not matched calendar quarters. Dollar amounts are U.S. dollars. Market references are the displayed October 1 close, retrieved October 2. ^{[3] [7] [12] [13]}

The products overlap more than the labels suggest

Applied Optoelectronics designs and manufactures optical networking products, including its own lasers and light engines. Its FY2025 filing describes proprietary MBE and MOCVD laser processes and manufacturing in the United States, Taiwan and China. Management presents this integration as a source of quality, cost and product-development advantages. The filing also describes intense competition and names Coherent, InnoLight, Eoptolink and Lumentum among its rivals. ^[1]

Credo starts with its own SerDes and DSP technology: circuitry that sends, recovers and processes high-speed signals. Its products include active electrical cables, optical DSPs and retimers. The FY2026 filing describes AECs for copper connections up to seven meters and a fabless model using manufacturing partners. Credo also identifies Broadcom, Marvell and Astera Labs among its competitors. ^[2]

These positions create different routes to value. A manufacturer's advantage must survive the cost of producing a qualified module at scale. A silicon designer's advantage must survive the next design cycle and remain valuable enough for customers to keep buying its technology. In either case, a good specification is the beginning of the commercial test.

The simple trade of buying AAOI for fiber and Credo for copper is already incomplete. Credo completed its acquisition of silicon-photonics developer DustPhotonics on May 28. ^[10] On September 15, it announced a 1.6T optical transceiver family combining its DSP, Kfir200 photonic integrated circuit and PILOT diagnostics. The claimed link-health benefits are vendor claims, rather than an independent performance comparison. ^[11]

My inference is that the two companies can become competitors in more optical sockets while continuing to serve different parts of a network. An investor should ask which component, module or system function each company sells in a particular deployment. The transmission medium alone no longer answers that question.

The latest quarter reveals who retains the gross profit

AAOI's Q2 revenue was $191.9 million, up 86.4% year over year. GAAP gross margin was 27.7%; non-GAAP gross margin was 29.8%. It recorded a $22.8 million GAAP net loss alongside $5.5 million of non-GAAP net income. Datacenter revenue was $107.7 million and CATV revenue $80.6 million. ^[3]

Credo reported $479.0 million of revenue, up 114.7% year over year, with GAAP gross margin of 64.5% and non-GAAP gross margin of 68.0%. GAAP net income was $129.4 million; non-GAAP net income was $236.3 million. ^[7]

Calculated from the underlying GAAP tables, Credo generated approximately 2.5 times AAOI's revenue but 5.8 times its gross profit. The GAAP gross-margin gap was about 36.8 percentage points. These are whole-company comparisons; they do not establish margins for a particular cable or transceiver. ^{[3] [7]}

That is the most useful starting point for the investment comparison. More sales can support more research, distribution and shareholder earnings when a business retains more of each sales dollar. AAOI's opportunity is partly to change its economics as production scales. Credo's opportunity is partly to preserve attractive economics as its product range expands.

The mix also matters. AAOI's reported datacenter and CATV revenue imply approximately 56% and 42% of Q2 sales, respectively. Those percentages are calculations, and datacenter revenue itself does not identify how much was specifically AI-related. ^[3]

I would therefore assess AAOI through two demand cycles rather than attaching an AI label to the whole income statement. The broadband business can help fund expansion or complicate the margin mix. Its presence gives AAOI another commercial engine, but also makes a direct comparison with an AI connectivity narrative less precise.

AAOI's upside requires capacity to become cash

In August, AAOI said 800G volume had more than doubled sequentially, manufacturing capacity was approaching 200,000 units per month, and it targeted roughly 650,000 monthly units of 800G and 1.6T products by year-end. These are reported ramp commentary and capacity expectations, not proof that the target has been reached or that every unit will sell. Its July Pearland announcement described nearly 400,000 square feet of additional manufacturing capacity. ^{[4] [5]}

The bullish interpretation is straightforward: a qualified supplier with available capacity could turn strong demand into a much larger business. The harder questions concern yield, utilization and the cash needed between purchasing equipment and collecting from customers. Installed equipment, technically available capacity and revenue-producing shipments are different milestones.

AAOI's six-month cash-flow statement makes the funding requirement visible. Through June 30 it used $73.8 million in operating cash, spent $335.1 million on property and equipment, and recorded $289.7 million of equipment and other prepayments. Net common-stock offering proceeds were approximately $1.03 billion. ^[6]

The equity funding supports the expansion; it also means shareholders must evaluate progress per share. I would want growth in sustainable earnings and cash generation to outrun dilution, rather than treating a larger factory or a higher cash balance as sufficient proof of investment success.

The accounting improvement deserves similar scrutiny. AAOI's Q2 bridge to non-GAAP profit included a $14.3 million tax adjustment and $4.9 million of share compensation. Adjusted EBITDA remained slightly negative at approximately $0.5 million. Management guided Q3 revenue to $255–290 million and non-GAAP gross margin to 29–30.5%. Guidance remains an expectation. ^[4]

The evidence I would seek next is simultaneous progress in shipments, margins and cash conversion. A revenue beat with little margin improvement would leave much of the transformation ahead. Better margins accompanied by rising receivables would require another check before crediting the improvement as durable cash generation.

Credo's quality has a substantial adjustment bill

Credo's latest quarter produced $90.2 million of operating cash flow and $7.3 million of property-and-equipment purchases. Subtracting those two reported figures gives approximately $82.9 million, a simple cash-flow calculation before acquisitions and other investing activity. The same quarter included $735.6 million of acquisition cash spending, net of cash acquired. ^[8]

This is a useful distinction between running the existing business and buying a larger opportunity. Positive operating cash generation gives Credo resources to invest. It does not make the acquisition free, and a cash-flow measure excluding acquisitions cannot answer whether the full capital-allocation program creates value.

Share-based compensation was $88.0 million in the quarter, approximately 18.4% of revenue by calculation. ^[8] Credo's release reconciles $129.4 million of GAAP net income to $236.3 million of non-GAAP income through stock compensation, acquisition costs, amortization and tax adjustments. Q2 revenue guidance was $525–535 million, with non-GAAP gross margin of 67–69%. ^[7]

For valuation, I would retain both earnings views. Adjusted results help examine operations, while GAAP results keep compensation and acquired-asset costs visible. Counting every adjustment as irrelevant would overstate what shareholders can repeatedly receive. Counting every acquisition charge as recurring would obscure another part of the picture.

The optical expansion raises the ambition further. The April acquisition announcement specified $750 million of upfront cash, approximately 0.92 million shares, and possible additional share consideration. Management then expected combined optical products to generate more than $500 million of fiscal 2027 revenue. That was a management forecast, not booked revenue. ^[9]

My test would be whether the new optical business earns an attractive return on the cash and shares committed to it. Revenue growth alone cannot settle that question. Credo's existing profitability gives the strategy credibility; it does not guarantee the next product family will reproduce the same margins.

Customer concentration is a risk for both

AAOI's FY2025 filing identifies Digicomm and Microsoft as 53.1% and 28.8% of annual revenue, respectively. It warns that purchase orders can change and that manufacturing fixed costs are difficult to reduce quickly. Those are historical annual exposures, not a statement of the current quarter's mix. ^[1] Its June 30 quarterly filing reports Digicomm at 67.2% of receivables and describes extended payment terms. ^[6]

Credo's latest quarterly filing separates contractual buyers from end customers. On the end-customer basis, four anonymized customers represented 33%, 28%, 13% and 10% of revenue, or 84% in total by calculation. The filing does not identify those letters as named hyperscalers. ^[8]

I would not confuse many product families with many independent customers. A large buyer can change deployment timing, qualify alternatives or negotiate on price across several products. That can affect both a factory ramp and a profitable semiconductor platform.

The shared exposure also limits diversification. Holding both companies may spread product and execution risk, but it cannot be assumed to remove dependence on the same broad infrastructure spending cycle. Their different businesses make them imperfect substitutes; they do not automatically make them independent investments.

The stock price asks for more than a good business

The October 1 quote snapshots displayed AAOI at $107.32 with approximately $9.11 billion of equity value, and CRDO at $210.17 with approximately $39.50 billion. These are dated market references, not current execution quotes. ^{[12] [13]}

Using each company's latest reported quarterly revenue multiplied by four produces equity-value-to-sales ratios of approximately 11.9 for AAOI and 20.6 for Credo. This deliberately simple calculation uses different reporting periods, ignores future growth, and is not enterprise value to sales. Its purpose is to expose how much expansion is already being purchased. ^{[3] [7] [12] [13]}

A sales multiple is especially incomplete here because retained profit differs so much. Credo deserves consideration for the stronger conversion already demonstrated; AAOI deserves consideration for the improvement still possible. Neither observation determines a fair premium without assumptions about growth, margins and future shares.

For Credo, annualizing the quarter's GAAP diluted EPS of $0.67 gives $2.68 and roughly 78 times the quoted price. Annualizing non-GAAP EPS of $1.20 gives $4.80 and roughly 44 times. Repeating a quarter four times is a calculation, not an earnings forecast. The large difference shows why the chosen earnings definition matters. ^{[7] [13]}

What future profitability would have to support

The following scenarios are my illustrative assumptions, not management guidance, analyst estimates or price targets. They hold today's rounded equity values constant and ignore later dilution, buybacks and changes in net cash. Net margin means profit after all expenses, not gross margin.

• AAOI: $2 billion of annual revenue at a 10% net margin would produce $200 million of earnings. A $9.11 billion equity value would then be about 46 times earnings. At $3 billion of revenue and a 15% net margin, earnings would be $450 million and the ratio about 20 times. ^[12]
• Credo: $2.5 billion of annual revenue at a 30% net margin would produce $750 million of earnings. A $39.50 billion equity value would then be about 53 times earnings. At $3 billion and a 35% net margin, earnings would be $1.05 billion and the ratio about 38 times. ^[13]

These are hurdles to interrogate, not outcomes to adopt. An AAOI investor needs conviction about a much more profitable manufacturing business. A Credo investor needs conviction that strong profitability can survive growth and support an already substantial valuation. Later share issuance would make preserving the same per-share value more demanding than these fixed-equity-value illustrations imply.

What would change my assessment

For AAOI, I would upgrade the business assessment when profitable shipments replace capacity targets as the central evidence: improving GAAP margins, reliable customer acceptance, cash collections and a reduced need for external capital. I would become more cautious if expansion keeps requiring fresh equity while the earnings available per share fail to improve.

For Credo, I would look for optical adoption that preserves attractive returns, operating cash generation that keeps pace with earnings, and compensation dilution that remains manageable. I would become more cautious if new optical revenue comes at materially weaker economics or if customer concentration turns a deployment delay into a persistent growth problem.

Architecture changes deserve attention in both cases. DustPhotonics' stated roadmap includes near-packaged and co-packaged optics. ^[9] I would therefore avoid assuming that every future increase in network bandwidth translates into the same number or economics of today's pluggable modules. Credo's broader portfolio may offer more ways to participate; AAOI's manufacturing expertise may still travel to new formats. Both propositions require evidence from actual qualified products and orders.

Credo currently offers the more convincing business economics. AAOI offers greater room for a transformation in those economics, accompanied by heavier funding and execution demands. The investment decision depends on the price paid for that distinction. I would underwrite AAOI through cash generation per share and Credo through the durability of profit after compensation and acquisition costs. That is a more useful comparison than choosing a winner between copper and fiber.