what are your thoughts on AAOI as a stock?

The orders have begun turning into revenue

The demand case deserves a fair hearing. AAOI’s data-center revenue reached $107.7 million in Q2 2026, versus $44.8 million a year earlier. Chief executive Thompson Lin said 800G volume “more than doubled sequentially.” Data-center products accounted for 56% of quarterly sales, with cable-network equipment contributing another $80.6 million. The company is benefiting from both faster data-center connections and cable infrastructure spending. [1][2]

The transition is visible in individual customer announcements. In March, AAOI disclosed an initial order exceeding $200 million for 1.6T transceivers from a longstanding hyperscale customer, with deliveries scheduled for the third and fourth quarters after qualification. The initial $53 million 800G order announced that month came from the same customer. By April, announced 800G orders from that customer totaled approximately $124 million, including a new $71 million order. [4][5][6]

There is evidence of a second relationship progressing beyond qualification. In the April announcement, AAOI said it had shipped the first 10,000 units of an 800G single-mode transceiver order to another hyperscaler. The large orders establish customer intent; shipped products and the subsequent revenue growth show execution beginning to catch up. [6][2]

The market is expanding fast enough to accommodate a meaningful new supplier. Cignal AI estimated that more than 42 million 400G-and-faster datacom modules shipped worldwide in 2025, up more than 80%. LightCounting’s April outlook forecast 65% Ethernet transceiver market revenue growth for 2026. These are different measures—module shipments and market revenue—but both support the demand backdrop. [7][8]

Our skepticism therefore starts further down the income statement. Q2 sales were real. The investment question is how far sales and margins must advance to earn a return from this price.

The price requires several times the guided sales pace

The identified price and latest available basic-share snapshot imply approximately $9.7 billion of equity value. Subtracting June net cash produces indicative enterprise value of roughly $9.39–$9.42 billion, depending on whether bank-acceptance payables are treated as debt-like. That is a substantial valuation for a company reporting $191.9 million of quarterly revenue. [3][1][2]

Management guided Q3 revenue to $255 million–$290 million. Annualizing the $272.5 million midpoint gives a $1.09 billion sales pace. At the lower enterprise-value estimate, investors are paying approximately 8.6 times that annualized guided quarter. The multiple already anticipates considerable expansion. [2][1][3]

To judge the prospective return, we use a two-year horizon and a 15% annual appreciation hurdle. That would take the stock to approximately $152 in October 2028. At 25× earnings and 95 million economic shares, the required $577 million of net income translates into approximately $3.65 billion of revenue at a 20% operating margin and 21% tax rate. At a 20× multiple, required revenue rises to $4.56 billion. The operating margin is an ambitious assumption: Q2’s GAAP operating margin was negative 12.9%. [3][1][2]

This is the gap we are underwriting. Revenue must rise severalfold while the business moves from operating losses to substantial profitability. Either change is plausible on its own. The valuation requires them together.

Our base case already gives AAOI considerable credit

Our base case assumes $2.5 billion of 2028 revenue and a 35% gross margin. Sales would exceed five times the company’s $455.7 million of 2025 revenue, while gross margin would improve by 7.3 percentage points from Q2 2026. This is a successful expansion in operating terms. [9][2]

We allow operating expenses to rise to $450 million as the business scales. That leaves $425 million of operating profit and approximately $336 million of net income after normalized tax. Divided among 95 million economic shares, earnings reach $3.53 per share. A 25× multiple produces an $88 stock, approximately 23% below the starting price two years later. The assumptions and alternatives are laid out below. [1][2][3]

The multiple deserves scrutiny too. We regard 25× as generous for a business still proving that expansion spending can produce durable earnings. But it is not a ceiling. Coherent’s premium valuation shows that optical businesses can retain considerable market enthusiasm, while Fabrinet’s $473 million of FY2026 GAAP net income on $4.64 billion of sales demonstrates the scale of realized profitability available elsewhere in the industry. Their different products and economics make them context rather than interchangeable valuation comparables. [3][10][11]

Even assigning our base earnings a 35× multiple yields approximately $124 per share, only 3.8% annual appreciation from $114.72. To reach our 15% return hurdle on those same earnings, the exit multiple must approach 43×. A buyer needs materially higher earnings or a persistently exceptional valuation. [3][1][2]

The factory ramp must improve margins as well as volumes

The strongest argument against our caution is the scale of the manufacturing step ahead. In August, chief financial officer Stefan Murry described capacity “approaching 200,000 units per month” and targeted approximately 650,000 monthly units of 800G and 1.6T products by year-end. That would increase nameplate capacity 3.25 times, to an annualized 7.8 million units. Lin forecast that demand would exceed production capacity through mid-2027. [2]

That is the mechanism behind the bull case. Larger volumes can spread factory and development costs across more products. Higher-speed product mix and improved yields can lift the gross profit earned on each shipment. But installed capacity has to pass through qualification, production, customer acceptance and payment before shareholders receive the benefit.

The latest quarter shows how much conversion remains. GAAP gross margin fell to 27.7%, from 30.3% a year earlier, despite the sales acceleration. AAOI reported a $24.7 million operating loss and a $22.8 million net loss. Its $5.5 million adjusted net profit was a useful milestone, but the reconciliation included $4.9 million of stock compensation and a $14.3 million tax adjustment. We want the factory ramp to produce sustained GAAP earnings as well as adjusted progress. [2]

Competitors are building into the same opportunity. Coherent plans successive expansions of internal indium-phosphide output. Lumentum identifies cloud modules and advancing 1.6T adoption as growth drivers. Innolight reports rapidly rising 800G and 1.6T shipments. These disclosures support the demand thesis while challenging the assumption that today’s supply constraints will preserve attractive pricing indefinitely. [12][13][14]

LightCounting states the cycle risk directly: “Once the suppliers add capacity and shortages ease, double ordering goes away and price declines accelerate.” Its uncertainty is timing. Our view is that this competitive mechanism matters more to the next two years than an assumption of imminent obsolescence. NVIDIA’s co-packaged-photonics roadmap remains a credible longer-term threat to some pluggable-transceiver sockets. [8][15]

Our bull case gives AAOI $3.5 billion of sales, 40% gross margin and approximately $691 million of net income. At 27× earnings, that produces roughly $196 per share and 31% annual appreciation. We give that possibility substantial weight. The evidence for demand, however, is stronger than the evidence for sustained 40% margins and the cash generation accompanying them. [1][2][3]

Shareholders are paying for the expansion ahead of its returns

The cash statement explains why accounting profitability is only part of the investment case. AAOI needs equipment before it can ship the additional products, and working capital while inventory and receivables grow. Those expenditures arrive before the earnings investors hope to capitalize.

In the first half of 2026, operations used $73.8 million of cash and PP&E purchases consumed $335.1 million. Conventional free cash flow—operating cash flow less PP&E purchases—was therefore negative $408.9 million. A separate $289.7 million equipment-prepayment line brings operating cash use plus those investment outlays to $698.7 million. Against that burden, AAOI raised $1.028 billion in net common-stock proceeds. [1]

Spending ahead of profitable demand can create value. Our concern is the size of the transition required. The base case generates approximately $176 million of 2028 free cash flow only after annual PP&E spending falls to $250 million—below the $335 million spent in the first six months of 2026 alone. Continuing equipment prepayments would reduce that modeled cash flow further. [1]

June cash of approximately $500 million provides liquidity, and the subsequently authorized $600 million ATM provides another avenue to raise capital. The latter is an authorization whose eventual use matters to per-share returns. Selling the entire amount at $114.72 would add approximately 5.23 million gross shares before fees. [1][16][3]

We test an economic share count of 90 million–100 million, with 95 million at the center. The latest identified basic snapshot was 84.57 million shares; the bridge includes convertible-note exposure, employee awards and customer warrants. The upper warrant exposure depends on future purchasing, and note settlement can involve cash, shares or both. Across our share-count range, base-case value remains approximately $84–$93. Dilution matters, but it is not the sole reason for our caution. [1][3]

Large customers retain much of the bargaining power

The funding burden becomes more consequential when a few buyers control the revenue. AAOI’s three largest customers accounted for approximately 92% of Q2 sales. Its filing states: “We do not have any long-term purchase commitments (in excess of one year) with any of our customers.” Most business is conducted through purchase orders. Factories have longer lives than those commitments. [1]

Digicomm illustrates the concentration outside AI optics. It represented approximately $147 million, or 42.8%, of first-half revenue, primarily in CATV, and 67.2% of June receivables. A large cable business gives AAOI another source of demand, while concentrating a substantial amount of sales and collections in one relationship. [1]

Amazon’s warrant shows how a powerful customer can participate in the upside. The agreement covers up to approximately 7.95 million shares, with most vesting tied to discretionary purchases reaching $4 billion over the warrant term. Amazon receives potential equity value as purchasing grows; AAOI supplies the products and funds the manufacturing expansion. The $4 billion is the purchase-linked vesting threshold. [1][17]

We think this combination—customer concentration, short purchasing horizons and competing capacity—warrants caution about capitalizing peak growth at a permanently high multiple. AAOI can win substantial revenue and still face pressure over how much of the resulting profit it keeps.

We would change our view on cash conversion or price

The next disclosures should make the investment case more concrete. Management’s Q3 revenue guide is $255 million–$290 million, and the announced 1.6T order carried a shipment schedule extending through Q4. Customer acceptance, recognized sales and the margins on those shipments will reveal more than another capacity announcement. [2][4]

For the current valuation to become attractive to us, we would want several quarters of qualified shipments and renewals supporting a credible path toward roughly $3.5 billion–$3.7 billion of annual revenue by 2028. We would also want GAAP gross margins moving into the mid-30s, prospective operating economics approaching 20%, and a controlled economic share count around 90 million–95 million. These are our underwriting tests, derived from the return hurdle and capital structure. [1][2][3]

Cash conversion is the final test: positive free cash flow after equipment prepayments, with a credible path toward at least $300 million annually and less dependence on fresh equity. Qualification delays, weaker yields, falling gross margins or persistent investment spending above our model would move us in the other direction.

Price can change the answer independently. Discounting our $88 base-case October 2028 value at 15%–20% annually gives approximately $61–$67 today. We would revisit AAOI in the low-to-mid $60s if the operating thesis remained intact. A price decline caused by lost orders would require a new model, rather than automatically creating a bargain. [1][2][3]

The next stage of AAOI’s story has to show that each new factory earns more for each share.