## Executive takeaway: the “beat-and-raise” is being reinforced by *hard* execution signals in the latest disclosures What’s driving the next leg isn’t just a one-quarter print—it’s that Applied Optoelectronics’ most recent 10‑K and late‑February 8‑Ks quietly line up around the same message: **the demand inflection is data-center/AI-led, the product cycle is moving to 800G+, and management is actively locking in U.S. capacity and financing to meet it**. ["fastest-growing business"](0001437749-26-005875|0) ["speeds reach 800 Gbps and above"](0001437749-26-005875|0) ["elevated data center demand will likely continue into 2026"](0001437749-26-005875|0) --- ## 1) The “real” read-through: AOI is framing 2026 as a *technology transition year* (not a one-off spike) Between the lines of the 10‑K, management is effectively telling you the upside skew comes from a mix shift that can surprise to the high side: **they expect 800G product sales to become the dominant driver later in 2026**, and they explicitly tie 800G to AI compute interconnect demand. ["800 Gbps products will likely exceed"](0001437749-26-005875|0) ["800G products in particular"](0001437749-26-005875|0) The underappreciated “alpha” angle: **transitions create volatility, but they also create share-take windows**—and AOI highlights advantages that matter specifically when hyperscalers ramp fast (in-house “light engines,” automation, and ability to scale production rapidly). ["highly-automated production process"](0001437749-26-005875|0) ["ability to scale production rapidly"](0001437749-26-005875|0) --- ## 2) Demand durability: they’re leaning on *customer forecasts + backlog* to justify momentum into 2026 AOI doesn’t just attribute FY2025 growth to a temporary environment; it points to **“customer forecasts and order backlog”** as the basis for expecting elevated data center demand to continue into 2026. That’s the kind of language you use when visibility is improving (even if not contractually guaranteed). ["Based on customer forecasts and order backlog"](0001437749-26-005875|0) And the causality is spelled out clearly: demand is coming from **new data center construction plus upgrades to enable AI**—a stronger narrative than generic “cloud growth,” because it implies capex is being directed at bandwidth-heavy architectures. ["data center construction"](0001437749-26-005875|0) ["enable new technologies like AI"](0001437749-26-005875|0) --- ## 3) Capacity is being pulled forward: Texas footprint expansion is showing up as signed commitments (not aspirations) If the market is rewarding “bullish guidance,” the filings show why management can say it with a straight face: **they are actively expanding Texas manufacturing/assembly capacity via long-duration leases with meaningful tenant improvement allowances**. ["lease approximately 153,928 rentable square feet"](0001683168-26-001360|0) ["tenant improvement allowance"](0001683168-26-001360|0) Two details that read like conviction (and are easy to miss on a headline skim): - The Houston lease includes **a purchase option with a stated price and near-term deadlines**, which is a stronger signal than a generic expansion plan—it’s a path to convert operating footprint into owned strategic capacity if the ramp sustains. ["one-time purchase option"](0001683168-26-001360|0) ["purchase price of $30,257,846.86"](0001683168-26-001360|0) - The 10‑K also discloses a separate Sugar Land, TX manufacturing lease (executed Sept. 2025) that **commences in 2026** and implies sizeable future fixed commitments—i.e., capacity is being staged ahead of demand. ["lease will commence in 2026"](0001437749-26-005875|0) ["209,665 square feet"](0001437749-26-005875|0) --- ## 4) Financing optionality: the ATM shelf is a “quiet accelerant” (and also a tell) The Feb. 26, 2026 8‑K discloses an **“at-the-market” equity program up to $250 million** with Raymond James and Needham as sales agents. That matters because it gives AOI the ability to fund working capital + capex needed for a steep ramp without waiting for a single, binary financing window. ["aggregate offering price of up to $250 million"](0001104659-26-020600|0) ["“at the market” offerings"](0001104659-26-020600|0) Hidden between the lines: this also reinforces that management views the opportunity set as large enough to justify **keeping the capital spigot available**—especially in a model where vertical integration and scaling can consume cash before it throws off peak margins. ["high fixed cost base"](0001437749-26-005875|0) ["substantial capital will be required"](0001437749-26-005875|0) --- ## 5) The “torque” (and the risk): vertical integration cuts both ways—so utilization is the swing factor AOI is explicit that its vertically integrated model creates a **high fixed-cost base**, meaning upside can be explosive when volumes ramp (operating leverage), but downside also bites if demand timing slips. The market tends to price the upside first; the filings remind you what the key swing variable really is: **utilization + yield as the mix shifts into new products.** ["high fixed costs"](0001437749-26-005875|0) ["lower yields... on new products"](0001437749-26-005875|0) That’s why the most “alpha” read of the current surge is: investors aren’t only buying last quarter’s result—they’re buying the possibility that **the 800G/AI cycle pulls utilization up fast enough that the fixed-cost structure works *for* them, not against them**. ["sales of 800 Gbps products"](0001437749-26-005875|0) ["AI... fueling a new wave of investment"](0001437749-26-005875|0)

## Executive takeaway: the “beat-and-raise” is being reinforced by *hard* execution signals in the latest disclosures What’s driving the next leg isn’t just a one-quarter print—it’s that Applied Optoelectronics’ most recent 10‑K and late‑February 8‑Ks quietly line up around the same message: **the demand inflection is data-center/AI-led, the product cycle is moving to 800G+, and management is actively locking in U.S. capacity and financing to meet it**. ["fastest-growing business"](0001437749-26-005875|0) ["speeds reach 800 Gbps and above"](0001437749-26-005875|0) ["elevated data center demand will likely continue into 2026"](0001437749-26-005875|0) --- ## 1) The “real” read-through: AOI is framing 2026 as a *technology transition year* (not a one-off spike) Between the lines of the 10‑K, management is effectively telling you the upside skew comes from a mix shift that can surprise to the high side: **they expect 800G product sales to become the dominant driver later in 2026**, and they explicitly tie 800G to AI compute interconnect demand. ["800 Gbps products will likely exceed"](0001437749-26-005875|0) ["800G products in particular"](0001437749-26-005875|0) The underappreciated “alpha” angle: **transitions create volatility, but they also create share-take windows**—and AOI highlights advantages that matter specifically when hyperscalers ramp fast (in-house “light engines,” automation, and ability to scale production rapidly). ["highly-automated production process"](0001437749-26-005875|0) ["ability to scale production rapidly"](0001437749-26-005875|0) --- ## 2) Demand durability: they’re leaning on *customer forecasts + backlog* to justify momentum into 2026 AOI doesn’t just attribute FY2025 growth to a temporary environment; it points to **“customer forecasts and order backlog”** as the basis for expecting elevated data center demand to continue into 2026. That’s the kind of language you use when visibility is improving (even if not contractually guaranteed). ["Based on customer forecasts and order backlog"](0001437749-26-005875|0) And the causality is spelled out clearly: demand is coming from **new data center construction plus upgrades to enable AI**—a stronger narrative than generic “cloud growth,” because it implies capex is being directed at bandwidth-heavy architectures. ["data center construction"](0001437749-26-005875|0) ["enable new technologies like AI"](0001437749-26-005875|0) --- ## 3) Capacity is being pulled forward: Texas footprint expansion is showing up as signed commitments (not aspirations) If the market is rewarding “bullish guidance,” the filings show why management can say it with a straight face: **they are actively expanding Texas manufacturing/assembly capacity via long-duration leases with meaningful tenant improvement allowances**. ["lease approximately 153,928 rentable square feet"](0001683168-26-001360|0) ["tenant improvement allowance"](0001683168-26-001360|0) Two details that read like conviction (and are easy to miss on a headline skim): - The Houston lease includes **a purchase option with a stated price and near-term deadlines**, which is a stronger signal than a generic expansion plan—it’s a path to convert operating footprint into owned strategic capacity if the ramp sustains. ["one-time purchase option"](0001683168-26-001360|0) ["purchase price of $30,257,846.86"](0001683168-26-001360|0) - The 10‑K also discloses a separate Sugar Land, TX manufacturing lease (executed Sept. 2025) that **commences in 2026** and implies sizeable future fixed commitments—i.e., capacity is being staged ahead of demand. ["lease will commence in 2026"](0001437749-26-005875|0) ["209,665 square feet"](0001437749-26-005875|0) --- ## 4) Financing optionality: the ATM shelf is a “quiet accelerant” (and also a tell) The Feb. 26, 2026 8‑K discloses an **“at-the-market” equity program up to $250 million** with Raymond James and Needham as sales agents. That matters because it gives AOI the ability to fund working capital + capex needed for a steep ramp without waiting for a single, binary financing window. ["aggregate offering price of up to $250 million"](0001104659-26-020600|0) ["“at the market” offerings"](0001104659-26-020600|0) Hidden between the lines: this also reinforces that management views the opportunity set as large enough to justify **keeping the capital spigot available**—especially in a model where vertical integration and scaling can consume cash before it throws off peak margins. ["high fixed cost base"](0001437749-26-005875|0) ["substantial capital will be required"](0001437749-26-005875|0) --- ## 5) The “torque” (and the risk): vertical integration cuts both ways—so utilization is the swing factor AOI is explicit that its vertically integrated model creates a **high fixed-cost base**, meaning upside can be explosive when volumes ramp (operating leverage), but downside also bites if demand timing slips. The market tends to price the upside first; the filings remind you what the key swing variable really is: **utilization + yield as the mix shifts into new products.** ["high fixed costs"](0001437749-26-005875|0) ["lower yields... on new products"](0001437749-26-005875|0) That’s why the most “alpha” read of the current surge is: investors aren’t only buying last quarter’s result—they’re buying the possibility that **the 800G/AI cycle pulls utilization up fast enough that the fixed-cost structure works *for* them, not against them**. ["sales of 800 Gbps products"](0001437749-26-005875|0) ["AI... fueling a new wave of investment"](0001437749-26-005875|0)