AI Optics After Earnings: $POET, $AAOI, $LITE and $COHR — What Counts as Commercial Proof?

00 Executive snapshot

The four companies are not direct peers. They differ in scale, product breadth, customer mix, manufacturing intensity and accounting maturity. That is exactly why the comparison is useful. They sit at different points on a commercialization curve, allowing investors to separate six concepts that are often collapsed into one: technical interest, qualification, purchase order, shipment, recognized revenue and profitable scale.

For POET, that distinction has become more important after the latest earnings cycle. The company reported Q2 revenue of approximately $569,925, up 112% from $268,469 a year earlier and 13% from $503,389 in Q1. It also reported roughly $796.3 million in cash and short-term investments. Those figures show a company with a large liquidity cushion and a still very small revenue base. The combination is unusual: financing pressure is materially lower, while the commercial conversion test is becoming much cleaner.

01 The question changed: market risk versus execution risk

For most of the AI-infrastructure cycle, optical investors had to argue from first principles. AI clusters were scaling, bandwidth per accelerator was rising, copper faced distance and power constraints, and faster networking architectures would need more optical connectivity. The investment debate therefore revolved around forecasts: how quickly 800G would migrate to 1.6T, when co-packaged or near-packaged optics would matter, and how much hyperscaler capital expenditure would reach optical suppliers.

That debate is no longer purely theoretical. In its latest fiscal fourth quarter, Lumentum reported approximately $1.01 billion of revenue, up 109% year over year, adjusted EPS of $3.23 and an adjusted gross margin of 50.4%. It guided the September quarter to $1.225-$1.275 billion of revenue. Coherent reported roughly $2.05 billion of fiscal Q4 revenue, up 34% year over year, adjusted EPS of $1.74 and next-quarter revenue around $2.3 billion at the midpoint. Applied Optoelectronics reported $191.9 million of Q2 revenue and said 800G volumes more than doubled sequentially.

Those results do not guarantee a straight-line optical cycle. Supply constraints, customer concentration, pricing, architecture changes and capex timing still matter. But they materially weaken one bearish explanation for an early-stage optical company: perhaps the market simply is not developing.

Strong results from $LITE, $COHR and $AAOI reduce uncertainty around the first question. They do not automatically reduce the second question for $POET. In fact, they can raise the standard: if peers are already converting AI demand into hundreds of millions or billions of quarterly revenue, then POET’s future rerating should increasingly depend on company-specific conversion evidence rather than sector enthusiasm alone.

02 Four companies, four standards of commercial proof

The table is not a league table. A company at an early commercialization stage should not be judged against a multi-billion-dollar supplier on absolute revenue. The relevant comparison is the type of evidence the market can demand at each stage.

A development agreement proves that two organizations see technical merit. A qualification program proves that a product is being tested. A purchase order establishes a stronger commercial commitment. But a purchase order is still not identical to revenue, and revenue is not identical to a profitable recurring business. This sounds obvious; in high-momentum sectors, it is one of the distinctions investors most often lose.

03 $POET: commercial proof starts with conversion, not TAM

The Q2 number is growing, but the denominator is still small

POET’s latest Q2 release reported approximately $569,925 of NRE and product revenue, up 112% year over year and 13% sequentially. That was the company’s sixth consecutive quarter of sequential revenue growth. The direction is constructive, but the absolute base remains small enough that percentage growth alone cannot carry the thesis.

The right question is not whether $569,925 is “good” or “bad” in isolation. It is whether the commercial commitments already announced can begin to appear in the income statement at a rate that changes the scale of the company.

The Lumilens distinction: $50M purchase order, >$500M conditional framework

The May agreement with Lumilens is the clearest example. POET disclosed an initial $50 million purchase order for EOI-based optical engines. The companies also outlined a broader framework that POET says could target more than $500 million of cumulative revenue over five years. POET’s own July FAQ explicitly states that the broader revenue opportunity is contingent on successful development, qualification and manufacturing scale-up milestones.

At Q2’s reported revenue level, the $50 million order is roughly 88 times one quarter of recognized revenue. That huge ratio is not automatically bullish or bearish. It simply shows how much information future conversion rates can carry. If material portions of the order begin shipping and being recognized, the income statement can change quickly. If conversion takes substantially longer than expected, the gap between announced commercial value and reported revenue will remain the central debate.

The >$5M Infinity order gives a second conversion clock

POET’s July company FAQ also reiterates a production order first announced on October 22, 2025 and valued at more than $5 million for POET Infinity 800G optical engines, targeted for shipment in the second half of 2026. That provides another useful clock because it predates the Lumilens agreement and is already described as a production order rather than a broad framework.

The key evidence to watch is sequential: qualification and production milestones, shipment, revenue recognition, then follow-on demand. A repeat order after successful shipment can be more informative than a larger first announcement because it begins to answer the question every component supplier eventually faces: did the customer come back?

The balance sheet has changed the nature of the risk

POET’s financing position is now very different from the typical early-stage hardware company. In the June 30 CEO update, management said the company had raised approximately $830 million of equity capital over the preceding 12 months and could receive up to an additional $661 million if outstanding warrants were exercised. The same update described an initial deployment of approximately $50 million for capital equipment in the second half of 2026.

The May registered direct offering alone raised $400,000,020 in gross proceeds through 19,047,620 shares and an equal number of warrants. The final closing documents set the combined unit price at $21 and the warrant exercise price at $26.25. That final strike matters because an earlier preliminary communication carried a different figure; the closing release and final filing are the numbers that should be used.

Q2’s reported cash and short-term investments of roughly $796.3 million reinforce the same point: immediate access to capital is no longer the obvious first-order constraint. Capital now needs to become equipment, capacity, qualification output and commercial throughput.

The industrial objective is measurable

POET’s June update reiterated that the production ramp was expected to begin in the second half of 2026 with volume shipments and that capacity could expand to as much as one million units per month by the end of 2027, more than ten times the then-current production capacity. That is a forward-looking target, not a current capability.

As the ramp develops, investors should expect physical evidence to emerge in multiple statements at once: capital equipment, inventory or work in process, receivables, product revenue and eventually gross margin. When a commercialization story is real, the balance sheet, cash-flow statement and income statement gradually start telling the same story.

04 $AAOI: what volume proof looks like

Applied Optoelectronics is useful because it sits between POET’s early conversion stage and the billion-dollar scale of Lumentum and Coherent. It is already shipping meaningful datacenter volume, yet manufacturing expansion remains central to the thesis.

AAOI reported $191.9 million of Q2 2026 revenue, up from $151.1 million in Q1 and $103.0 million one year earlier. Datacenter revenue reached approximately $107.7 million, or 56.1% of the quarter. The company said 800G volume more than doubled sequentially.

Management also said overall production capacity was approaching 200,000 units per month and continued to target approximately 650,000 units per month of combined 800G and 1.6T capacity by year-end. It expects high-speed optical demand to outpace production capacity through roughly mid-2027.

That is a completely different type of problem from POET’s current one. AAOI no longer has to show that customers will pay for high-speed optics at material scale. It has to prove it can install capacity, manage yields, procure components, build inventory, ship on time and keep unit economics acceptable while volumes rise quickly.

Guidance creates another layer of proof

AAOI guided Q3 revenue to $255-$290 million. At the midpoint, that would be about $272.5 million, materially above Q2. Non-GAAP gross-margin guidance was 29%-30.5%. The ability to guide a quarter in this way tells investors that the company has substantially more visibility into customer demand and production schedules than an early commercial-stage supplier.

AAOI’s balance sheet also shows the physical cost of scaling. Q2 inventory was approximately $278.8 million, up from $183.1 million at year-end 2025, while net property, plant and equipment rose to approximately $697.1 million from $376.1 million. Those increases are not automatically good or bad. They are evidence that the revenue ramp requires real capital and working capital.

05 $LITE: when optical demand becomes margin proof

Lumentum’s fiscal fourth quarter shows the next stage of maturity. The company reported approximately $1.01 billion of revenue, up 109% year over year. Adjusted EPS reached $3.23 and adjusted gross margin reached 50.4%. For the first quarter of fiscal 2027, management guided revenue to $1.225-$1.275 billion and adjusted EPS to $4.05-$4.35.

The important point is not simply that sales grew. Revenue growth and gross-margin expansion are occurring together. In fiscal Q4 2025, Lumentum had reported $480.7 million of revenue and a 37.8% non-GAAP gross margin. One year later, revenue has more than doubled while adjusted gross margin has moved above 50%.

That is evidence of operating leverage and product-mix power. It tells investors that the optical cycle is capable of producing not just volume, but high-value revenue with improving economics for at least one major supplier.

It also changes what should be demanded from a smaller company. An early-stage supplier does not need to match Lumentum’s margin immediately. But as product revenue becomes material, the business must eventually show that the engineering proposition creates economics: lower assembly cost, good yield, defensible pricing, or another measurable advantage. Technology can be differentiated without creating attractive shareholder economics; gross margin is one of the places where the distinction eventually becomes visible.

06 $COHR: scale itself becomes a form of proof

Coherent reported approximately $2.05 billion of fiscal Q4 revenue, up 34% year over year, with adjusted EPS of $1.74. Management’s next-quarter outlook implies roughly $2.3 billion of revenue at the midpoint and adjusted EPS of $1.85-$2.05.

Coherent’s relevance to this comparison is not that POET should become Coherent. The companies are built differently. Coherent spans a broad photonics and materials portfolio, large-scale manufacturing and multiple end markets. The value of the comparison is that Coherent shows the far end of the commercial chain: AI-driven optical demand can be absorbed by a diversified industrial platform at multi-billion-dollar quarterly scale.

At this stage, the question is capacity efficiency. Coherent has been investing in datacenter and communications capabilities and, earlier in 2026, received a $2 billion strategic investment from NVIDIA to support R&D, capacity expansion and operational capabilities. The company therefore faces a mature version of the same basic problem: how to turn capital expenditure into enough reliable optical output to satisfy demand without eroding margins or returns.

That matters for POET because it demonstrates that the optical opportunity is not a single-product speculative narrative. There is a very large industrial market forming around AI connectivity. Capturing a piece of it, however, requires execution across manufacturing, customer qualification, supply chain and economics.

07 The commercial-proof ladder

$POET is moving through the middle of this ladder. $AAOI has moved firmly into volume and forward visibility. $LITE is showing margin conversion. $COHR is operating at the scaled industrial end. Investors should therefore resist using the same metric to judge all four companies.

For POET, another partnership can be useful, but it is no longer the only evidence the market needs. The existing portfolio of orders and partners is already large enough to create measurable conversion milestones. The next incremental dollar of recognized product revenue can now carry more information than another large TAM estimate.

Revenue quality matters as much as revenue growth

There is one more reason the order-to-revenue framework is useful: not all revenue has the same informational value. An early photonics company can report non-recurring engineering revenue, prototype revenue, qualification-unit revenue and recurring production revenue inside the same broad top-line discussion. Those dollars can have very different implications for the durability of the business.

NRE revenue can be strategically valuable because it demonstrates that customers are paying POET to solve real engineering problems. It can also fund part of development. But NRE is usually project-specific and may not repeat at the same level. Production revenue, by contrast, begins to answer whether a qualified architecture is moving into customer deployments. Once production revenue is followed by repeat orders from the same customer, the evidence becomes stronger again.

As POET scales, four disclosures will therefore matter beyond the headline revenue number:

• Product versus NRE mix. A rising proportion of recurring product revenue would make year-over-year comparisons more economically meaningful.
• Accounts receivable. Revenue growth accompanied by receivables can be normal during a ramp, but collection quality and customer concentration will eventually matter.
• Inventory and work in process. Rising inventory ahead of orders can support growth; inventory rising much faster than shipments can also signal timing or demand risk.
• Deferred or contract liabilities. Depending on contract structure, these can help investors understand whether cash or customer commitments are arriving before accounting recognition.

This is where the AAOI comparison becomes especially useful. AAOI already has enough production scale for investors to see the physical footprint of the ramp in inventory and PP&E. Lumentum and Coherent have enough scale for investors to judge margins and forward guidance. POET’s disclosures should gradually migrate in the same direction—not because the business models are identical, but because a real hardware commercialization cycle leaves increasingly visible traces across all three financial statements.

Why valuation should not be inferred from peer revenue multiples

Another common mistake is to take the peer read-through one step too far. If Lumentum or Coherent reports exceptional growth, it can be tempting to apply a peer revenue multiple directly to POET’s announced orders or to the >$500 million Lumilens framework. That would mix different stages of evidence.

A dollar of recognized, recurring revenue from a scaled supplier with demonstrated gross margins is not economically equivalent to a dollar inside a conditional five-year framework at an early-stage supplier. The latter may eventually become more valuable if the company has superior technology and rapid growth, but it carries qualification, timing, manufacturing, concentration and margin uncertainty that the current income statement has not yet resolved.

The most defensible way to use the peers today is therefore as a market-validation benchmark, not as an automatic valuation shortcut. Their earnings help answer whether optical connectivity is attracting real spending. POET’s own filings must answer what percentage of that spending it can capture and on what economics.

08 What strong peers prove for $POET — and what they do not

What the peer results increasingly support

• AI optical demand is real. The revenue acceleration at Lumentum, Coherent and AAOI is too large to explain only through narrative.
• 800G/1.6T migration is commercially monetizable. AAOI’s 800G ramp and the higher-speed product commentary across peers show customers are spending real capital.
• Optics can carry attractive economics. Lumentum’s 50.4% adjusted gross margin demonstrates that high-value optical products can translate demand into strong margin.
• Capacity is becoming strategically valuable. Multiple suppliers are expanding output because the limiting factor is increasingly how much qualified product can be delivered.

What the peer results cannot prove for POET

• That a POET qualification will pass on a particular timetable.
• That the $50M Lumilens order will convert into recognized revenue on a particular quarterly schedule.
• That the >$500M conditional framework will be realized in full.
• That POET will achieve one million units per month by end-2027.
• That POET will earn peer-like gross margins once volume production begins.
• That customers will place recurring orders after first shipments.

09 The five-number POET dashboard from here

1. Recognized product revenue

Not the headline value of a framework. Not a total addressable market. The cleanest variable is how much product revenue actually reaches the financial statements. Q2’s approximately $0.57 million establishes the current scale.

2. Order-to-revenue conversion

The initial Lumilens order is $50 million. The older Infinity order is more than $5 million. Future quarters should allow investors to measure how quickly those commitments become shipments and recognized revenue. The conversion schedule matters almost as much as the headline value.

3. Repeat orders

Follow-on orders are a quality signal. They indicate that a customer has moved beyond initial engagement and is willing to commit again. Over time, customer recurrence will matter more than the number of names in a partnership slide.

4. Manufacturing evidence

Management plans an initial ~$50 million of H2 capital-equipment deployment and is targeting a much larger capacity footprint by end-2027. Investors should watch whether capital spending, installed equipment, production output and revenue begin to move together.

5. Gross margin and cash conversion

Once product revenue is large enough, the debate will shift from “can POET sell it?” to “can POET sell it at attractive economics?” The Optical Interposer thesis includes claims of assembly simplicity and cost advantages. Eventually those advantages need to appear in gross profit, working capital and cash conversion.

10 Three scenarios — without turning the framework into a forecast

11 Bottom line: the optical market is being proved; POET must prove participation

The newest optical earnings cycle changes the burden of proof.

Lumentum has crossed the billion-dollar quarterly-revenue threshold while expanding adjusted gross margin above 50%. Coherent is operating above $2 billion a quarter and guiding higher. Applied Optoelectronics has already reached the point where 800G volume, manufacturing capacity and working capital are central operating questions.

POET remains much earlier. Q2 revenue is still measured in hundreds of thousands of dollars, yet the company has a $50 million initial Lumilens purchase order, another production order above $5 million, substantial liquidity and a stated plan to scale capacity sharply. The disparity is not a reason to compare valuation mechanically with the large suppliers. It is the reason to focus on conversion.

The next decisive proof is therefore straightforward:

If those steps occur, POET can begin closing the gap between an engineering story and an operating business. If they do not, strong peer earnings will not rescue the company-specific thesis. That is the useful lesson from comparing $POET with $AAOI, $LITE and $COHR today.

12 Related Merlintrader research

• POET Technologies ($POET) Stock Hub 2026 — the evergreen company hub for orders, capital structure and execution milestones.
• AI Optics 2026: The Memory-to-Fiber Rotation — $CIEN, $LITE and $POET — sector rotation and the optical spending shift.
• $POET, $COHR and $AAOI: Three Different Ways to Play the AI Photonics Race — platform, scale and hyperscaler-ramp framework.
• Why AI Needs Optics — the optical-interconnect stack and where POET sits technically.