$POET, $COHR and $AAOI: Three Different Ways to Play the AI Photonics Race

Executive Summary

The artificial intelligence infrastructure boom is not only a story about GPUs, accelerators, memory, power, cooling, and real estate. It is also, increasingly, a story about light. As AI clusters become larger, denser, and more power hungry, the bottleneck shifts from raw compute alone to the movement of data between processors, racks, data halls, and campuses. Electrical interconnects still matter, but the further the AI buildout moves into high-bandwidth, low-latency, energy-constrained architectures, the more optical networking becomes one of the critical hidden layers of the trade.

That is where POET Technologies, Coherent Corp., and Applied Optoelectronics enter the discussion. They are not the same type of company. They do not carry the same balance-sheet quality, customer profile, manufacturing scale, product maturity, investor base, or risk level. But they all sit inside the same widening question: who supplies the optical rails of the AI data-center era?

POET Technologies is the most speculative of the three, but it is also the most asymmetric and arguably the most interesting small-cap platform story in this comparison. It is built around the POET Optical Interposer, a packaging and integration platform designed to reduce cost, complexity, size, and power consumption in optical engines. Its bull case is not based on current revenue scale. It is based on whether the company can convert years of technology development, partner announcements, product demonstrations, and customer engagements into meaningful commercial shipments. POET is closer to an option on photonic integration than a mature operating company. That makes it risky, but also genuinely compelling for readers who understand early-stage technology risk.

The company recently reported Q4 2025 revenue of only $341,202, while also highlighting large financing inflows and a move from development toward execution. The market’s view of POET remains unstable because the opportunity is large, but the evidence of sustained revenue conversion is still early. The friendlier, slightly bullish reading is that POET does not need to look like Coherent today to matter; it needs to prove that its platform can become useful inside a market that is clearly moving toward denser, faster, lower-power optical connectivity.

Coherent is the industrial benchmark in this trio. It is not a clean small-cap moonshot. It is a large photonics and materials company with scale, an existing revenue base, a broad technology stack, and a real position in data-center and communications optics. Coherent reported fiscal Q3 2026 revenue of $1.81 billion, with non-GAAP gross margin of 39.6% and non-GAAP diluted EPS of $1.41. Its data-center and communications exposure makes it one of the more credible institutional ways to express the AI photonics theme. Coherent is not likely to move like POET on a single speculative customer headline, but it also does not need to prove that it can manufacture at commercial scale.

Applied Optoelectronics, or AOI, sits between those two profiles. It is more commercial than POET and far more volatile than Coherent. It has real revenue, real customers, and a direct role in optical transceivers for data-center and broadband markets. In Q1 2026, AOI reported record revenue of $151.1 million, up 51% year over year, with datacenter revenue of $81.4 million compared with $32.0 million in the prior-year quarter. The company also announced in March 2026 that it received its first volume order for 1.6T data-center transceivers from a major hyperscale customer. That matters because 1.6T is exactly where the next stage of AI networking demand is moving. But AOI is not a low-risk story either. Margins, execution, customer concentration, production ramp timing, and quarterly guidance sensitivity can move the stock violently.

The cleanest way to frame this article is simple: POET is the technology-option story, Coherent is the scaled photonics-infrastructure story, and AOI is the revenue-ramp transceiver story. Each name offers exposure to the AI optical networking boom, but each does so through a very different risk/reward structure. The tone toward POET should be constructive rather than dismissive: the company is still early, yes, but it is early in a market where the technical problem it is targeting is becoming more important, not less.

For traders and investors, the key is not to pretend that these three are interchangeable. They are not. POET is about validation and commercialization. Coherent is about scale, leadership, and margin leverage. AOI is about hyperscaler demand, 800G/1.6T transceiver ramp, and whether record revenue can become durable operating profitability.

The opportunity is real. The risks are real too. AI data centers need more optical bandwidth, but that does not automatically mean every optical stock wins equally. The winners will likely be the companies that can deliver the right speed, power profile, cost structure, manufacturability, reliability, and customer integration at the right time. In photonics, timing is everything. A brilliant platform that arrives too early can starve. A scaled supplier that executes late can lose share. A fast-ramping transceiver company can look explosive until one customer pauses, one qualification slips, or one margin assumption breaks.

That is why $POET, $COHR, and $AAOI make a compelling trio. Together, they give readers three different windows into the same AI infrastructure race: innovation, scale, and ramp. Among the three, POET remains the least proven — but also the one where even modest commercial validation could matter most to the equity story.

Why Photonics Has Become a Real AI Infrastructure Story

The AI trade began publicly with chips. Nvidia became the icon. Memory suppliers followed. Server vendors, liquid cooling names, power equipment companies, and data-center landlords later entered the story. But underneath the obvious winners, there is a quieter infrastructure layer that has become harder to ignore: optical connectivity.

Large AI clusters do not behave like traditional enterprise servers. Training and inference workloads push enormous volumes of data across GPUs, CPUs, memory pools, switches, storage systems, and network fabrics. The more compute a cluster contains, the more critical the communication layer becomes. A GPU that waits for data is not an efficient GPU. A cluster that cannot move data at sufficient speed wastes capital. A data center that adds compute without solving interconnect power and bandwidth eventually runs into physical and economic limits.

This is where optical networking moves from background plumbing to strategic infrastructure. Copper has advantages at short distances, but its limitations become more painful as speed, distance, density, and power requirements increase. Optical links offer higher bandwidth over distance, lower signal loss, and, in many architectures, better power efficiency as data rates scale. That does not mean optical replaces everything everywhere. It means optical becomes more important as AI networks move into 800G, 1.6T, 3.2T, co-packaged optics, optical circuit switching, and eventually more radical photonic interconnect models.

The industry is moving through several overlapping transitions at once. First, 400G and 800G optical modules have become increasingly central to cloud and AI data-center upgrades. Second, 1.6T transceivers are moving from roadmap language into actual customer orders and ramp planning. Third, the market is already discussing 3.2T and beyond because AI cluster growth is forcing suppliers to think several generations ahead. Fourth, the industry is experimenting with co-packaged optics, near-packaged optics, external light sources, and optical engines that bring optics closer to compute and switching silicon.

This is not merely a technical upgrade. It is a capital-allocation shift. Hyperscalers are spending aggressively to build AI capacity. Those budgets flow through the entire supply chain: semiconductors, switches, optical components, modules, lasers, materials, connectors, testing equipment, and manufacturing infrastructure. When Nvidia agreed to invest in photonics firms such as Coherent and Lumentum, the message was not subtle. The AI leader wants influence and supply assurance across the optical stack because future AI systems depend on more than accelerators alone.

For investors, this creates a difficult but attractive setup. Photonics is not a simple category. It includes lasers, modulators, photodiodes, transceivers, optical engines, silicon photonics, indium phosphide devices, packaging, test systems, materials, switching architectures, and integration platforms. Some companies sell finished modules. Some sell components. Some sell platforms. Some are vertically integrated. Some outsource heavily. Some are already producing at scale. Others are still waiting for commercial validation.

The market often groups them together whenever the AI optics trade heats up. That can create opportunities, but also confusion. A company with $1.8 billion in quarterly revenue is not the same risk as a company with a few hundred thousand dollars in quarterly revenue. A company with a 1.6T volume order is not the same as a company demonstrating a future optical engine. A company with a broad materials and photonics stack is not the same as a platform company still proving that customers will adopt its architecture at scale.

That is the central tension in comparing POET, Coherent, and AOI. They all participate in the AI photonics race, but they are not running in the same lane.

The Three Lanes: Platform, Scale, and Ramp

The most useful way to understand this trio is not by ranking them immediately, but by assigning each company to its natural lane.

POET is the platform lane. The company’s core thesis is that its Optical Interposer can change how optical engines are designed and manufactured. The promise is integration: bring optical components, electronics, waveguides, coupling structures, and packaging features together in a way that reduces assembly complexity and supports scalable production. If POET’s platform becomes broadly adopted, the upside could be disproportionate to its current revenue base. But if adoption remains slow, fragmented, delayed, or dependent on partner roadmaps, the stock can remain highly volatile.

Coherent is the scale lane. It already has the footprint, product breadth, manufacturing capability, and customer access to be treated as a major AI optical infrastructure supplier. It is not merely trying to prove that photonics matters. It is already selling into markets where photonics matters. The investment debate is therefore less about survival and more about growth quality: how much of the AI optics cycle flows through Coherent, at what margin, with what level of durability, and how the company balances growth with capital allocation and operational complexity.

AOI is the ramp lane. Applied Optoelectronics has a direct commercial hook into the datacenter transceiver cycle. It has reported strong year-over-year revenue growth, a record Q1 2026, and a first 1.6T volume order from a major hyperscaler. That makes the story easier for traders to understand: demand is rising, the company is ramping, and the question is whether production, yields, margins, and customer volumes can support a much larger revenue base. AOI is more tangible than POET because it has meaningful revenue now, but less diversified and less institutionally stable than Coherent.

This platform-scale-ramp framework is important because it prevents a common mistake: comparing these companies only by ticker performance. A stock can rise for very different reasons. POET can rise on a partnership or product-validation headline. Coherent can rise on margin expansion, guidance, Nvidia-related photonics demand, or broader AI infrastructure optimism. AOI can rise on 800G shipments, 1.6T orders, hyperscaler demand, or raised revenue guidance. The market may call all three “AI optical stocks,” but the operating evidence behind each one is different.

For a reader trying to understand the sector, the better question is not “which one is best?” The better question is: what kind of AI photonics exposure does each ticker actually offer?

POET Technologies: The High-Beta Optical Interposer Bet

POET Technologies is the most asymmetric and most debated name in this group. Its appeal is obvious: the company is positioned around one of the most important problems in high-speed optical networking, and its market capitalization has historically allowed traders to imagine very large upside if commercial adoption accelerates. Its risk is also obvious: revenue is still tiny relative to the scale of the story, the company has relied on financing to fund execution, and recent customer-order controversy has increased governance and disclosure scrutiny. But a fair article should not reduce POET to those risks alone. The company is attempting to solve a real industrial problem at the exact moment AI infrastructure is putting more pressure on optical bandwidth, power efficiency, packaging density, and manufacturing scalability.

The core of POET’s business is the POET Optical Interposer, a platform intended to simplify the integration and packaging of photonic and electronic components. In optical modules and engines, assembly complexity can be a major cost and manufacturing bottleneck. Traditional approaches often require precise alignment between lasers, waveguides, fibers, modulators, photodiodes, and electronic drivers. Active alignment steps can be expensive, labor intensive, and difficult to scale. POET’s pitch is that its interposer platform uses design and packaging techniques that can reduce those pain points and enable compact, cost-effective, thermally optimized optical engines.

In simple terms, POET is not selling the same kind of story as AOI. AOI is largely about transceiver manufacturing and customer demand. POET is about a deeper packaging and integration layer that could sit inside next-generation optical modules, external light sources, and co-packaged or near-packaged architectures. If POET succeeds, it may become an enabling platform rather than a commodity module vendor. That is the bull case.

But this is also why POET is difficult to value. Platform stories are powerful when customers adopt them. They are dangerous when adoption is delayed. Investors can see technical demonstrations, partnerships, prototypes, and press releases, but the market ultimately wants purchase orders, revenue, gross margin, customer concentration clarity, and production evidence. Without that conversion, the stock remains vulnerable to speculation cycles.

POET’s Recent Financial Picture

POET’s Q4 2025 results show the current gap between narrative and revenue scale. The company reported Q4 2025 revenue of $341,202, compared with $29,032 in Q4 2024. That is a large percentage increase, but the absolute number remains very small. The company also reported continued operating expenses, including research and development, depreciation and amortization, professional fees, and wages and benefits.

The more important headline from the company’s own messaging was financing. Management said it secured more than $225 million in financing during Q4 2025 and an additional $150 million in January 2026. That capital gives POET more room to scale manufacturing, support product development, and pursue customer programs. It also changes the near-term survival conversation. A small technology company with insufficient cash can be forced into unfavorable financing or slow execution. A better-funded POET can move more aggressively.

However, the presence of capital does not eliminate execution risk. It shifts the debate from “can POET fund itself?” to “can POET convert this funding into commercial traction?” That is a better debate, but still a serious one.

The company’s Q4 commentary framed 2025 as a transition from development to execution. That language is important because POET’s investor base has waited a long time for commercialization. The market is no longer satisfied by technology potential alone. It wants evidence that customers are not only interested, but ordering, qualifying, integrating, and scaling.

Product and Technology Positioning

POET’s recent product narrative centers on high-speed optical engines and light-source solutions for AI applications. The company has discussed products and demonstrations tied to 800G, 1.6T, external light sources, hybrid lasers, and co-packaged optics. It has also announced joint development work with LITEON for optical modules for AI applications. That collaboration is relevant because LITEON has manufacturing and module experience, while POET contributes its Optical Interposer platform.

The LITEON announcement described a jointly developed optical engine that would use the POET Optical Interposer to integrate optical components, drive electronics, and coupling structures into a compact and thermally optimized module. The goal is scalable and cost-efficient production for advanced optical modules used in next-generation co-packaged optics, AI systems, and high-bandwidth data-center applications.

That is exactly the zone investors want POET to occupy. The AI data-center market needs more bandwidth and better power efficiency. Optical engines that reduce complexity and improve manufacturability could become valuable if adopted by module makers, system vendors, and hyperscaler supply chains. The problem is not whether the use case exists. The problem is whether POET’s solution becomes a commercial standard, a niche component, or a promising technology that larger competitors route around.

POET has also highlighted products such as Blazar and Starlight external light-source concepts for AI applications. External light sources matter because co-packaged optics and related architectures may separate the light source from the switching or compute package. Instead of embedding lasers directly in every module or package, systems can use external laser sources to feed optical engines. This can improve thermal management, serviceability, and architecture flexibility, but the market is still evolving.

The fact that POET is discussing these topics puts it in the right conversation. But being in the right conversation is not the same as winning commercial share.

The Marvell/Celestial AI Order Cancellation Issue

Any serious POET article in May 2026 must address the Marvell/Celestial AI issue directly. Ignoring it would make the piece look promotional rather than credible.

In April 2026, POET disclosed that Marvell Semiconductor, which had acquired Celestial AI, provided written notice canceling all purchase orders related to Celestial AI. According to POET’s own update, Marvell cited disclosures of information related to the purchase order and shipping information as the basis for the cancellation, alleging that those disclosures violated confidentiality obligations.

This matters for several reasons.

First, the cancellation touched a customer relationship that had become symbolically important to the market. When a small company is still proving commercialization, each named or implied customer relationship carries outsized weight. Losing or jeopardizing one relationship can affect investor confidence far beyond the immediate dollar amount.

Second, the stated reason for cancellation raises disclosure and governance questions. In photonics supply chains, confidentiality is not a side issue. Customers, especially large semiconductor and AI infrastructure companies, are highly sensitive to product roadmaps, shipment timing, partner identities, and technical details. If a supplier is perceived as too promotional or loose with sensitive information, it can damage trust.

Third, the episode reinforces the difference between POET and larger peers. Coherent can absorb customer friction more easily because it has diversified revenue. AOI can also face customer concentration risk, but it has a larger existing revenue base. POET, by contrast, depends heavily on the market believing that partner and customer relationships will turn into revenue. A broken relationship therefore hits the narrative core.

To be fair, POET has stated that it continues to pursue other customer orders and opportunities. The cancellation does not automatically invalidate the Optical Interposer platform. Technology can remain valuable even after one commercial dispute. But it does raise the burden of proof. The company now needs cleaner execution, clearer communication, and more evidence from other customers to rebuild confidence.

POET Bull Case

The bullish case for POET is that AI data centers need new optical architectures, and POET owns a differentiated integration platform at the right time. If the Optical Interposer can reduce cost, simplify manufacturing, improve thermal behavior, and support high-speed optical engines, it could become attractive to module manufacturers and system-level customers that need scalable solutions.

The financing position strengthens this case because POET is no longer operating from the same level of cash scarcity. With more capital, the company can invest in manufacturing scale-up, product readiness, customer support, and supply-chain execution. In platform companies, underfunding can kill even good technology. Better funding improves the odds of reaching commercial milestones.

The LITEON relationship is another important bull-case element. A small platform company often needs manufacturing and ecosystem partners to reach the market. POET does not need to become Coherent overnight if it can embed its technology into partners’ module programs. If those programs move from prototypes to high-volume production, POET’s revenue model could change quickly.

Finally, the speculative appeal is straightforward. POET’s current revenue base is tiny relative to the scale of AI optical infrastructure spending. If revenue begins to inflect, the percentage growth could be dramatic. That is why the stock attracts retail attention. A company moving from hundreds of thousands of dollars in quarterly revenue to millions, then tens of millions, would look very different.

POET Bear Case

The bear case is equally direct. POET has been a technology story for years, but the revenue base remains small. The company has repeatedly needed to convince the market that commercialization is approaching. Until sustained revenue appears, skeptics can argue that the story is still mostly promise.

The Marvell/Celestial AI cancellation adds a second layer of concern, and it should be treated seriously. Even if the technology works, customers must trust the company. In AI infrastructure, relationships with large semiconductor and cloud ecosystem players are delicate. A confidentiality-related dispute is not just another missed order; it can create a reputational overhang. Still, it should not be treated as a final verdict on the Optical Interposer platform. Early-stage technology companies can absorb setbacks if they respond with cleaner execution, stronger communication discipline, and broader customer validation.

There is also competitive risk. POET is not operating in an empty field. Large suppliers, vertically integrated photonics companies, silicon photonics specialists, optical module makers, semiconductor vendors, and hyperscaler-backed initiatives are all working on similar problems. A small company can have strong technology and still be squeezed if the ecosystem standardizes around another architecture.

Financing can also be a double-edged sword. Capital helps execution, but dilution history matters. Investors in small-cap technology stories often face repeated capital raises before commercial scale arrives. If POET’s growth takes longer than expected, even a stronger balance sheet may not prevent future dilution.

The final risk is timing. AI optics demand is real, but product qualification cycles are not instant. Hyperscalers and system vendors do not adopt critical optical infrastructure casually. Reliability, manufacturability, yield, cost, thermal performance, and supply assurance all matter. POET must move at the speed of the market while passing the slow, rigorous tests that infrastructure customers require.

POET Bottom Line

POET is the highest-risk, highest-beta member of the trio, but also the name with the most direct small-cap leverage to a potentially differentiated photonic integration platform. It has the least current revenue support, so it cannot be treated as a proven operating story. But the company’s story can work if commercial validation broadens beyond press releases and demonstrations into repeatable revenue, production scale, and trusted customer relationships. In a friendly but realistic reading, POET is not a stock to judge by maturity metrics alone; it is a platform story that must now prove it can cross the bridge from promising technology to commercial relevance.

For a trader, POET is a volatility instrument tied to AI photonics headlines, customer updates, financing perception, and sentiment. For a fundamental investor, it is a commercialization watchlist name that must be judged by orders, revenue conversion, margin path, and governance improvement. It is not a mini-Coherent. It is a different beast entirely.

Coherent Corp.: The Scaled Photonics Infrastructure Benchmark

Coherent is the adult in the room. That does not automatically make it the best stock, but it makes it the most established operating company in this comparison. Unlike POET, Coherent does not need to prove that it can generate meaningful revenue. Unlike AOI, it is not narrowly defined by one transceiver ramp or one hyperscaler cycle. Coherent is a broad photonics, materials, and optical systems supplier with exposure to several end markets, including data-center and communications infrastructure.

The company’s relevance to the AI photonics race has increased sharply because AI clusters require more optical connectivity, and Coherent has the technology stack to serve multiple layers of that demand. Its portfolio spans materials, lasers, components, modules, transceivers, and optical networking technologies. That vertical breadth is a strategic advantage when customers want performance, reliability, supply depth, and roadmap continuity.

At OFC 2026, Coherent highlighted technologies for next-generation pluggable optical transceivers spanning 1.6T, 3.2T, and emerging architectures for 12.8T and beyond. It also discussed AI-scale optical innovations including 400G-per-lane technologies, 3.2T transceivers, co-packaged optics, optical circuit switching, multi-rail transport, and open optical networking platforms. This matters because Coherent is not only selling into today’s 800G and 1.6T cycle; it is trying to position itself for the next several network generations.

Latest Financial Performance

Coherent’s fiscal Q3 2026 results provide the strongest financial foundation among the three companies. The company reported revenue of $1.81 billion for the quarter ended March 31, 2026. GAAP gross margin was 37.7%, GAAP diluted EPS was $0.97, non-GAAP gross margin was 39.6%, and non-GAAP diluted EPS was $1.41.

Those numbers are not speculative. They are operating scale. They show a company already converting demand into revenue and earnings. The data-center and communications segment has been a major driver, supported by AI infrastructure demand and the broader shift toward higher-speed optical networking.

The company also guided for continued strength. Management projected fiscal Q4 2026 revenue in the range of $1.91 billion to $2.05 billion, with non-GAAP gross margin expected to continue improving. The market may still debate valuation, cycle risk, and the sustainability of growth, but Coherent is already playing at a level POET hopes to reach one day.

Why Coherent Matters in AI Photonics

The reason Coherent matters is not merely that it sells optical products. It matters because it controls parts of the technology stack that are hard to replicate. Photonics is not software. Manufacturing quality, materials expertise, laser performance, thermal behavior, optical coupling, packaging, testing, and reliability all determine whether products can ship at scale.

AI data centers are pushing optical suppliers toward higher density and higher speed. A 1.6T transceiver is not just a faster version of an older module. As bandwidth rises, design challenges multiply. Signal integrity, power consumption, heat, component yield, and module reliability become harder. When the roadmap moves toward 3.2T, 6.4T, and 12.8T-related architectures, only suppliers with deep technical and manufacturing capability can realistically compete.

Coherent’s advantage is that it can participate across multiple product types. If the market remains dominated by pluggable optical modules, Coherent can supply into that. If co-packaged optics gains share, Coherent has relevant technology. If external light sources become more important, Coherent’s laser and photonics capabilities matter. If optical circuit switching becomes more central to AI cluster architecture, the company has exposure there too.

That breadth is important because the AI optical market is still evolving. No one knows exactly which architecture will dominate every layer of the network. Hyperscalers may use different approaches depending on distance, workload, cost, vendor relationships, and system design. A broad supplier can adapt more easily than a narrow one.

Nvidia and the Strategic Value of Photonics

One of the biggest signals for the sector has been Nvidia’s willingness to commit capital and strategic attention to photonics. Nvidia’s investment plans involving Coherent and Lumentum reinforced the idea that optical technology is no longer peripheral to AI infrastructure. When the dominant AI accelerator company moves to secure future access to advanced optics, lasers, and photonic manufacturing capacity, investors should pay attention.

For Coherent, this type of strategic validation matters. It suggests that the company’s photonics capabilities are not only relevant to legacy telecom or industrial markets, but also to the highest-growth area in technology infrastructure. It also reinforces Coherent’s role as a trusted supplier to major players rather than a speculative newcomer trying to get noticed.

However, investors should not treat strategic investment headlines as automatic profit guarantees. Large customers and ecosystem partners can create opportunity, but they can also exert pricing pressure, demand capital investment, and shift product roadmaps. The question for Coherent is whether strategic AI demand translates into durable high-margin revenue, not just headline relevance.

Coherent Bull Case

The bull case for Coherent rests on several pillars.

First, AI data-center demand is real and growing. Higher-speed optical connectivity is becoming more important as clusters scale. Coherent is positioned directly inside that trend.

Second, the company already has revenue scale. Unlike POET, Coherent does not need to prove that customers will buy its products in principle. It needs to prove that growth can accelerate and margins can improve.

Third, Coherent has a broad and vertically integrated technology stack. This gives it multiple shots on goal as optical architectures evolve. It can serve pluggables, components, lasers, optical circuit switching, and future co-packaged or near-packaged systems.

Fourth, the company’s recent financial performance shows both revenue growth and profitability. In a market where many AI-adjacent stories are still burning cash, Coherent offers a more mature profile.

Fifth, institutional validation from AI ecosystem leaders supports the idea that photonics is becoming strategic. Coherent is one of the names investors naturally look at when they want scaled exposure to that theme.

Coherent Bear Case

The bear case is not that Coherent lacks relevance. It clearly has relevance. The bear case is that expectations may become too high, margins may face pressure, and the business is more complex than a pure AI optics story.

Coherent is diversified. That is a strength, but also a complication. Not every segment has the same growth profile. Some parts of the business may be cyclical, exposed to industrial or telecom spending, or affected by customer inventory cycles. Investors buying Coherent only as an AI optics stock may underestimate the broader business mix.

The company also operates in a competitive market. Lumentum, Broadcom, Marvell, Innolight, Eoptolink, Fabrinet-linked ecosystems, and many other players compete across the optical supply chain. Coherent has scale, but scale does not guarantee pricing power forever.

Capital intensity is another issue. Photonics manufacturing requires investment. Scaling to meet AI demand can require capacity expansion, equipment, engineering, and working capital. If customers demand rapid ramp but pricing is tight, the company may have to manage the tradeoff between growth and margin.

Valuation can also become a risk. When a mature company becomes an AI infrastructure favorite, the market can quickly re-rate it. If earnings merely meet expectations rather than beat them dramatically, the stock can pull back even on solid results. That is what often happens when a good business becomes a crowded trade.

Coherent Bottom Line

Coherent is the scaled benchmark for this comparison. It offers the most mature exposure to AI photonics among the three names, with real revenue, profitability, technology breadth, and strategic relevance. It is less explosive than POET and probably less retail-volatile than AOI, but it is also the company with the clearest current operating proof.

For investors who want AI optical infrastructure exposure without relying entirely on a commercialization inflection, Coherent is the cleaner large-company candidate. For traders looking for extreme upside from a small base, it may feel less exciting. That is exactly the point: Coherent is not the lottery ticket. It is the infrastructure supplier.

Applied Optoelectronics: The Hyperscaler Transceiver Ramp Story

Applied Optoelectronics is the most direct revenue-ramp story in this trio. It is not as speculative as POET because it already has meaningful revenue and customer demand. It is not as broad or scaled as Coherent because its story is more concentrated around optical transceivers, data-center demand, and production execution. That makes AOI a fascinating middle lane: more tangible than POET, more volatile than Coherent.

AOI designs and manufactures optical networking products used in data-center, CATV, telecom, and other markets. Its relevance to AI infrastructure comes mainly through high-speed data-center transceivers. As hyperscalers upgrade networks to support AI workloads, demand shifts toward faster modules such as 800G and 1.6T. AOI’s ability to capture that demand is the core of the current investment thesis.

Q1 2026: Record Revenue and Datacenter Acceleration

AOI’s Q1 2026 results were strong on top-line growth. The company reported total revenue of $151.1 million, up from $99.9 million in the first quarter of 2025. Datacenter revenue was $81.4 million, compared with $32.0 million in the prior-year period. CATV revenue was $66.8 million, up modestly from $64.5 million. Telecom and other revenue were much smaller contributors.

This mix matters. The datacenter segment is the part investors care about most because it connects AOI to AI infrastructure spending. A jump from $32.0 million to $81.4 million year over year is not cosmetic. It shows that AOI’s datacenter business is becoming the growth engine.

The company also reported total gross profit of $43.9 million compared with $30.5 million in the prior-year quarter. However, the market reaction to results can still be volatile because investors focus not only on revenue growth but also on EPS, margin, guidance, and the timing of production ramps.

1.6T Volume Order: Why It Matters

In March 2026, AOI announced that it received its first volume order for 1.6T data-center transceivers from a major hyperscale customer. That is one of the most important facts in the AOI story.

1.6T is not a marketing buzzword. It represents the next stage of high-speed optical connectivity as AI networks require more bandwidth. If 800G is the current major ramp, 1.6T is the forward-looking proof point. A volume order suggests that AOI is not merely demonstrating technology; it is being pulled into customer deployment planning.

For traders, this is exactly the type of headline that can change the stock’s perception. It connects AOI to hyperscaler capex, AI networking upgrades, and the broader theme of optical scarcity. It also raises expectations. Once a company receives a 1.6T volume order, the market will expect follow-through in revenue, margins, shipment volume, and future orders.

The risk is that production ramps are not linear. Customer orders can be lumpy. Qualification timelines can shift. Supply constraints can appear. Yields can affect margin. A company can have demand and still disappoint investors if the timing or profitability of the ramp does not match expectations.

AOI’s Manufacturing and Customer Position

AOI’s story is not only about products; it is about manufacturing. Optical transceiver demand is only valuable if a company can produce at scale, meet customer specifications, and deliver reliably. The AI optical market is unforgiving. Hyperscalers do not want theoretical capacity. They want qualified, tested, high-performance modules that can support enormous data-center deployments.

AOI has emphasized production capacity expansion and high-speed product ramps. The company has also benefited from market dynamics that may favor non-Chinese suppliers in certain customer supply chains. Geopolitical and supply-chain considerations can matter in optical networking because large cloud providers often want supplier diversity, regional resilience, and reduced dependency on restricted vendors.

That said, AOI’s customer profile can be both an opportunity and a risk. Large hyperscaler customers can drive huge volume, but they can also create concentration risk. If one customer slows orders, shifts timing, changes design preferences, or pressures pricing, the effect can be significant. Smaller suppliers may benefit from hyperscaler growth, but they do not control the customer roadmap.

AOI Bull Case

The bull case for AOI is straightforward and powerful.

First, datacenter revenue is growing fast. The Q1 2026 datacenter number shows real momentum.

Second, AOI has exposure to the 800G and 1.6T upgrade cycle. These are not distant concepts; they are active customer-driven markets.

Third, the first 1.6T volume order gives the company a credible forward-looking hook. If that order becomes the start of a larger ramp rather than a one-off event, AOI’s revenue base could expand significantly.

Fourth, AOI is more levered to optical transceiver demand than a diversified company like Coherent. That means if the transceiver cycle is strong, AOI can move dramatically.

Fifth, the company’s profile fits the current market appetite. Investors want AI infrastructure names beyond Nvidia. AOI offers a direct, understandable angle: AI data centers need faster optical links, and AOI supplies them.

AOI Bear Case

The bear case begins with execution. Revenue growth is not enough if margins and profitability do not follow. Optical modules can be competitive, technically demanding, and price-sensitive. A company can grow revenue but still disappoint if gross margin fails to expand or operating expenses consume the benefit.

Customer concentration is another major concern. Hyperscaler orders can be transformative, but they can also create dependency. If a major customer delays a ramp, changes suppliers, or renegotiates terms, AOI’s quarterly numbers can swing.

Guidance sensitivity also matters. AOI’s stock can react sharply to small differences between management guidance and investor expectations. When a stock becomes tied to a ramp narrative, the market often prices in acceleration before it fully appears. Any hint of delay can cause a pullback.

Competition is intense as well. AOI is not the only company chasing 800G and 1.6T opportunities. Larger, better-capitalized, and more vertically integrated players are also targeting the same demand. AOI must prove that it can win share, maintain quality, and defend margins.

Finally, AOI has historically been a cyclical and volatile name. Investors should not assume that one strong AI-driven cycle eliminates business cyclicality. Optical demand can be powerful, but inventory digestion, product transitions, and customer timing have repeatedly created volatility across the sector.

AOI Bottom Line

AOI is the clearest revenue-ramp story in this trio. It has record revenue, a rapidly growing datacenter segment, and a meaningful 1.6T hyperscaler order. That gives it far more near-term commercial evidence than POET. But AOI also carries significant execution and customer concentration risk, and it does not have Coherent’s scale or breadth.

For traders, AOI may be the most attractive middle ground: enough real business to avoid being purely speculative, but enough volatility and ramp potential to offer major upside if execution is strong. For long-term investors, the key question is whether AOI can turn AI optical demand into durable profitability rather than just impressive revenue growth.

Direct Comparison: What Each Company Actually Offers

A useful comparison must separate technology, revenue, customers, scale, risk, and timing.

Technology Exposure

POET offers exposure to photonic integration and optical engines. Its key asset is the Optical Interposer platform. The story is about packaging innovation, integration, and the possibility of becoming an enabling technology inside next-generation optical modules and AI interconnects.

Coherent offers exposure to a broad photonics stack. Its technology base includes materials, lasers, optical components, transceivers, and advanced optical systems. It is less of a single-platform bet and more of a diversified photonics infrastructure supplier.

AOI offers exposure to high-speed optical transceivers and optical networking products. Its near-term story is less abstract: can it ship 800G and 1.6T products into hyperscaler demand at scale and acceptable margins?

Revenue Proof

Coherent has the strongest revenue proof by far, with quarterly revenue above $1.8 billion in fiscal Q3 2026.

AOI has meaningful and growing revenue, with Q1 2026 revenue of $151.1 million and datacenter revenue of $81.4 million.

POET has the weakest revenue proof, with Q4 2025 revenue of $341,202. Its story depends on future conversion rather than present scale.

Customer and Partner Validation

Coherent has broad institutional and strategic validation. Its role in AI photonics is supported by major market demand and ecosystem attention.

AOI has direct hyperscaler validation through its 1.6T volume order and datacenter revenue growth.

POET has partner and customer engagement validation, including LITEON and other announced relationships, but it also faces a recent negative validation event through the Marvell/Celestial AI cancellation.

Balance Sheet and Financing

POET’s financing inflows improve its runway and execution capacity, but the company still needs to show that capital turns into revenue. Financing is helpful, but not the same as customer-funded growth.

Coherent has a mature operating structure and scale, but also the capital intensity of a large manufacturing and photonics company.

AOI is in a ramp phase where working capital, production investment, and customer timing matter. Its financial profile will be judged by whether revenue growth converts into margin and operating income.

Risk Profile

POET is the highest risk. Its upside depends on technology adoption and commercialization.

AOI is medium-to-high risk. It has revenue and orders, but also customer concentration, margin, and execution sensitivity.

Coherent is lower risk relative to the other two, but not risk-free. It has valuation, margin, competitive, and cyclical risks.

Timing

POET’s timing is still uncertain. The market needs to see customer programs move into commercial shipments.

AOI’s timing is immediate. The 800G and 1.6T ramp is already being watched quarter by quarter.

Coherent’s timing is ongoing and multi-year. It is already benefiting from AI optics demand while preparing for future network generations.

The AI Data-Center Optical Stack: Where the Three Companies Fit

To understand the investment debate, it helps to visualize the AI data-center optical stack as several layers.

At the top are system-level AI clusters: GPUs, accelerators, CPUs, memory, switches, servers, and storage. These are the visible pieces of the AI infrastructure boom. Below that sits the networking layer: Ethernet, InfiniBand-like fabrics, switching architectures, routing, optical circuit switching, and high-speed interconnects. Below that are modules and engines: 800G, 1.6T, and future optical transceivers, co-packaged optics, near-packaged optics, and external light sources. Beneath those are components: lasers, modulators, photodiodes, drivers, TIAs, waveguides, substrates, interposers, connectors, and packaging. Below all of it is manufacturing: epitaxy, wafer processing, assembly, testing, alignment, thermal management, reliability qualification, and supply-chain control.

Coherent touches several layers. It can supply components, modules, lasers, materials, and advanced optical technologies. That gives it broad optionality.

AOI sits strongly in the module and transceiver layer, with manufacturing and customer supply-chain exposure. Its value is tied to customer deployments of high-speed optical links.

POET sits in the integration and optical-engine layer. Its value depends on whether its platform becomes a preferred way to assemble and scale optical engines or external light-source solutions.

This distinction is important because the most valuable layer can shift over time. During one phase, module suppliers may capture the best economics because demand exceeds supply. During another phase, component suppliers may benefit because lasers or photodiodes become bottlenecks. During a later phase, packaging and integration platforms may become more valuable because power and density constraints make conventional assembly too expensive or inefficient.

A diversified supplier like Coherent can participate in multiple shifts. A focused company like AOI can outperform if its specific layer is in shortage and it executes well. A platform company like POET can outperform dramatically if the industry adopts its architecture, but may lag if the market standardizes elsewhere.

Customer Concentration: The Hidden Risk in Optical Stocks

Optical networking companies often look most exciting when they announce hyperscaler wins. A major cloud customer can transform revenue, validate technology, and attract investor attention. But customer concentration is also one of the biggest risks in this sector.

Hyperscalers are powerful buyers. They control volume, timing, qualification, pricing pressure, and roadmap decisions. They often dual-source or multi-source where possible. They may push suppliers to invest in capacity before full demand visibility is guaranteed. They can also change architecture if internal needs shift.

For POET, customer concentration risk appears in a different form. Because the company is still early in commercialization, each customer or partner matters more. A single order, collaboration, or cancellation can move the market’s perception dramatically. The Marvell/Celestial AI cancellation shows how fragile sentiment can be when a company’s revenue base is still small.

For AOI, customer concentration risk is tied to actual revenue. The company can benefit enormously from major hyperscaler demand, but the same dependence can create volatility if one customer changes timing or mix.

For Coherent, the risk is diluted by scale and diversification, but not eliminated. Large AI customers and strategic partners can still influence capacity planning, pricing, and product priorities.

This is why investors should treat customer announcements carefully. A named or implied hyperscaler relationship is valuable, but it must be translated into revenue, margin, and repeat demand. A purchase order is better than a memorandum. A production shipment is better than a prototype. Repeat orders are better than a first order. Diversified customers are better than one giant buyer.

Manufacturing Scale: The Difference Between a Good Idea and a Business

Photonics is a manufacturing discipline. That may sound obvious, but in stock-market narratives it is often forgotten. The market loves the idea of faster optical links, lower power, and AI-ready interconnects. But the real winners must manufacture at scale.

Manufacturing optical products is difficult because tolerances are tight and performance requirements are high. Lasers must behave reliably. Coupling losses must be controlled. Modules must pass thermal and mechanical stress. Yields must be good enough for margin. Testing must be efficient. Supply chains must be reliable. Customers must trust that the supplier can deliver not only a sample, but thousands or millions of units depending on the product category.

This is where Coherent’s advantage is most visible. The company already operates at industrial scale. It has experience across photonics manufacturing and can support large customers. That does not make execution easy, but it reduces the existential question.

AOI has a real manufacturing base and is ramping high-speed products. Its challenge is to scale specific product generations while maintaining margin and meeting customer timing. A successful ramp can transform the company. A messy ramp can punish the stock.

POET’s manufacturing challenge is more foundational. The company must prove that its platform can move from development and partner demonstrations to reliable commercial production. If the Optical Interposer meaningfully simplifies assembly, then manufacturing could become a strength. But the market needs proof.

In optical stocks, investors should always ask: who can build it, who can build it cheaply, who can build it repeatedly, and who can build it when customers need it?

Margin Quality: Revenue Is Not Enough

AI optics demand can drive revenue, but revenue alone does not make a good business. Margin quality matters.

Coherent’s Q3 fiscal 2026 non-GAAP gross margin of 39.6% gives it a stronger profitability profile than many smaller optical names. The company’s margin trajectory will be closely watched because AI demand should, in theory, support better mix and scale. But if competition, pricing pressure, or capacity costs offset growth, investors may become more cautious.

AOI’s margin story is more sensitive. As high-speed datacenter products ramp, the company needs to show that revenue growth brings operating leverage. If gross margin expands and operating expenses remain controlled, the stock can earn a higher-quality multiple. If revenue grows but margins disappoint, investors may question whether AOI is capturing enough value from the AI cycle.

POET’s margin story is still mostly theoretical because revenue scale is not yet meaningful. The company’s platform could potentially support attractive economics if adopted broadly, but there is not yet enough commercial history to judge sustainable gross margin. For POET, the first step is revenue conversion. Margin quality comes after that.

This is a critical distinction. Coherent is judged on margin improvement. AOI is judged on margin conversion. POET is judged on whether it can reach a revenue base where margin analysis becomes meaningful.

Valuation Logic: How the Market May Price Each Name

The market is likely to value these companies using different mental models.

POET may trade like a technology option. Investors may assign value based on total addressable market, partner announcements, product demonstrations, cash runway, and perceived probability of commercialization. Traditional valuation metrics are difficult because current revenue is minimal. This can produce dramatic stock moves in both directions. If confidence rises, the stock can disconnect from current financials. If confidence breaks, it can fall sharply because there is little revenue support.

Coherent may trade on earnings power, revenue growth, margin trajectory, AI optics exposure, and strategic positioning. It can receive an AI infrastructure premium, but investors will still look at actual financial performance. Because it is larger and more mature, the valuation debate is more conventional.

AOI may trade on revenue ramp, customer wins, gross margin, guidance, and the perceived size of its 800G/1.6T opportunity. It can receive a high-growth multiple if investors believe revenue will scale rapidly and profitability will follow. But if guidance disappoints or margins lag, the multiple can compress quickly.

In other words, POET trades on probability, Coherent trades on execution quality, and AOI trades on ramp credibility.

Retail Sentiment: Why These Names Attract Different Crowds

Retail sentiment is especially important for POET and AOI. Both names can move sharply on news, social discussion, and theme-driven momentum. Coherent also benefits from AI optics sentiment, but its larger size and institutional shareholder base make it less purely retail-driven.

POET attracts retail investors because it looks like an undiscovered platform story in a hot market. The stock has the ingredients that online communities like: AI relevance, photonics buzz, a small-cap structure, partnership headlines, large addressable markets, and the possibility of a major re-rating if commercialization succeeds. The danger is that this kind of sentiment can become emotional. When retail investors treat every partnership as proof of inevitable success, risk gets underpriced. When a negative event occurs, sentiment can reverse brutally.

AOI attracts a different kind of retail and trader crowd. Its story is more tied to actual revenue and hyperscaler demand. Traders can point to record revenue, 800G and 1.6T products, and customer orders. That makes the thesis easier to explain. But because expectations are already linked to near-term results, earnings reactions can be violent.

Coherent attracts more institutional and thematic investors. It can be owned as part of the AI infrastructure supply chain, but it is less likely to be treated as a meme-like binary story. That can make it less explosive, but also more stable.

Retail sentiment can be useful as a signal of attention, but it should not be treated as fact confirmation. Social media can identify what traders care about. It cannot prove that a product will scale, a customer will repeat orders, or a margin target will be achieved.

Competitive Landscape: The Race Is Bigger Than Three Companies

POET, Coherent, and AOI do not exist in a closed three-way race. The optical networking and photonics landscape is broad and highly competitive.

Large semiconductor and networking companies such as Broadcom, Marvell, and Nvidia influence system architecture and supply-chain priorities. Optical specialists such as Lumentum, Coherent, and others supply key components and technologies. Module makers and contract manufacturers play major roles in transceiver production. Silicon photonics companies, indium phosphide specialists, laser suppliers, and packaging innovators all compete for position.

This matters because AI photonics is not a winner-take-all market in the simple consumer-internet sense. Different architectures can win in different zones. One company may dominate high-performance lasers. Another may win transceiver share. Another may supply components. Another may control switching silicon. Another may own a packaging method that becomes attractive for co-packaged optics.

For POET, competition comes from alternative integration approaches and larger companies with internal photonics resources. If a hyperscaler, switch vendor, or module maker decides to use a different architecture, POET may not participate.

For Coherent, competition is broad but the company’s scale gives it resilience. Its challenge is to maintain leadership while customers demand higher performance and lower cost.

For AOI, competition is direct in optical transceivers. Winning 800G and 1.6T demand requires cost, quality, speed, and customer trust. The market can be large, but it will not be uncontested.

The important point is that AI optics growth does not automatically make every supplier a winner. The market may grow rapidly while pricing remains competitive. Some companies may win volume but not margin. Others may have excellent technology but fail to gain customer adoption. Some may be acquired. Some may be designed out.

Catalysts to Watch

For POET, the most important catalysts are customer order updates, production shipments, partner program progress, 800G and 1.6T optical-engine milestones, external light-source validation, and any clarification around governance or customer trust after the Marvell/Celestial AI cancellation. Investors should also watch financing use, cash burn, and whether reported revenue begins to scale beyond early commercial levels.

For Coherent, catalysts include quarterly results, data-center and communications growth, margin expansion, AI-related customer demand, updates on 1.6T and 3.2T product adoption, optical circuit switching progress, co-packaged optics roadmap milestones, and any additional strategic agreements with AI infrastructure leaders.

For AOI, catalysts are more near-term and measurable: 800G shipment ramp, 1.6T order conversion, Q2 and Q3 2026 revenue progression, gross margin trends, customer concentration disclosures, capacity updates, and whether full-year guidance remains intact or improves.

Across all three, broader sector catalysts include hyperscaler capex commentary, Nvidia and cloud infrastructure roadmaps, Ethernet AI networking adoption, supply-chain constraints, optical component shortages, and any evidence that co-packaged optics or external light-source architectures are moving closer to deployment.

Bull Scenario for the Trio

In the bullish sector scenario, AI data-center demand continues to expand faster than expected, pushing hyperscalers to accelerate optical network upgrades. 800G demand remains strong, 1.6T ramps faster, and customers begin preparing seriously for 3.2T and beyond. Optical suppliers gain pricing power because capacity and technical capability become constraints.

In this scenario, Coherent benefits from broad demand across its photonics portfolio. Revenue grows, margins improve, and the market rewards the company as a strategic AI infrastructure supplier. AOI benefits from customer ramps and transceiver demand, with datacenter revenue rising sharply and operating leverage improving. POET benefits from increased urgency around optical integration, winning additional customer programs and converting partner activity into revenue.

This is the scenario in which all three stocks can work, though in different ways. Coherent may compound through earnings and multiple support. AOI may rerate through revenue acceleration. POET may move explosively if commercialization proof finally arrives.

Bear Scenario for the Trio

In the bearish scenario, AI optical demand remains real but expectations prove too aggressive. Hyperscaler deployments shift timing, inventories build, pricing pressure increases, and product transitions create uneven quarterly results.

Coherent may still grow, but if margins disappoint or guidance falls short of elevated expectations, the stock can pull back. AOI may suffer more sharply if ramp timing, customer orders, or gross margin do not match investor expectations. POET may be hit hardest if commercialization remains slow or if customer trust issues persist after the Marvell/Celestial AI cancellation.

In this scenario, the market separates proven scale from speculative promise. Coherent likely holds up best fundamentally, AOI becomes a show-me story, and POET faces the greatest sentiment risk.

Base Scenario

The base scenario is probably more nuanced than either extreme. AI optical demand continues to grow, but not every quarter is smooth. Coherent remains a major beneficiary, though the stock may react to margin details and valuation. AOI continues to ramp datacenter revenue but faces volatility around guidance and profitability. POET continues to make technical and partner progress, but investors demand more commercial evidence before granting a sustained re-rating.

This base case supports the idea of watching all three names, but not treating them equally. Coherent is the quality benchmark. AOI is the ramp candidate. POET is the asymmetric technology-option candidate.

Which Name Is Most Directly Comparable to POET?

If the question is “which of these is most directly comparable to POET,” the answer depends on what part of POET one wants to compare.

Technologically, Coherent is the more serious benchmark because it operates across photonics, lasers, components, and advanced optical architectures. It shows what scale looks like in the sector. If POET wants to become strategically relevant, Coherent is one of the giants standing in the broader ecosystem.

Commercially, AOI may be the better trading comparison because it is more exposed to optical transceiver demand and has the kind of datacenter revenue ramp that POET investors want to see eventually. AOI is not doing exactly what POET does, but it gives readers a concrete example of what happens when AI optical demand starts showing up in financial results.

For a Merlintrader-style article, that is why the trio works. POET is not being compared to two identical companies. It is being compared to two reference points: one scaled photonics leader and one high-growth transceiver ramp story. Together, they help readers understand what POET still needs to prove.

What POET Must Prove Next

POET’s next phase is about proof. The company does not need more abstract excitement around AI. It needs measurable commercial progress.

The first proof point is revenue growth. Investors need to see quarterly revenue move from early-stage levels into a meaningful trajectory. A few hundred thousand dollars in quarterly revenue does not support the scale of the AI photonics narrative by itself.

The second proof point is customer diversification. After the Marvell/Celestial AI issue, the market needs evidence that other customers and partners remain engaged and that POET can maintain trusted commercial relationships.

The third proof point is production readiness. The Optical Interposer story is powerful only if it translates into manufacturable products. Investors should watch for shipment updates, yield commentary, capacity milestones, and partner manufacturing progress.

The fourth proof point is communication discipline. A small technology company in a sensitive supply chain must be careful with disclosures. The market may forgive one controversy if execution improves, but repeated issues would be damaging.

The fifth proof point is margin potential. This comes later, but eventually POET must show that its platform is not only technically interesting but economically attractive.

What Coherent Must Prove Next

Coherent’s burden of proof is different. It must show that AI optics demand can drive sustained growth and better margins at scale.

Investors should watch the data-center and communications segment, order trends, product mix, and gross margin. If Coherent can keep expanding in AI-related optical markets while improving profitability, the stock can continue to attract infrastructure investors.

The company must also show that its broad portfolio is an advantage rather than a drag. Diversification helps reduce risk, but investors want the AI growth engine to be large enough to matter at the corporate level.

Coherent also needs to manage expectations. When a company becomes a preferred AI infrastructure name, good results may not be enough if the stock already prices in excellence. The market will want evidence of acceleration, not just stability.

What AOI Must Prove Next

AOI must prove that its revenue ramp is durable and profitable.

The first watch item is 800G and 1.6T execution. Investors should look for shipment growth, customer expansion, and confirmation that demand is translating into revenue.

The second is gross margin. AOI needs to show that high-speed datacenter products can improve profitability rather than simply increase sales.

The third is customer concentration. More customers or broader programs would reduce risk. Heavy dependence on one or two hyperscaler relationships can create volatility.

The fourth is guidance credibility. If AOI raises expectations, it must meet or exceed them. The market has little patience for ramp stories that stumble.

Final Bottom Line

$POET, $COHR, and $AAOI represent three different ways to approach the AI photonics race.

POET is the speculative platform bet. Its Optical Interposer and optical-engine strategy could become highly valuable if adopted at scale, but the company still needs to prove commercial conversion, customer trust, and production execution. The upside is potentially large because the current revenue base is small, but that same small base makes the risk high.

Coherent is the scaled photonics leader. It already has revenue, margins, customers, and a broad technology stack. It offers the cleanest institutional exposure to AI optical infrastructure among the three, though investors must still watch valuation, margins, cyclicality, and competition.

AOI is the hyperscaler transceiver ramp story. It has record revenue, strong datacenter growth, and a first 1.6T volume order. It is more tangible than POET and more explosive than Coherent, but it remains vulnerable to customer timing, margin execution, and guidance volatility.

The AI photonics race is real. The need for faster, denser, more power-efficient optical connectivity is not a retail fantasy. AI clusters are forcing the industry to rethink how data moves. But the existence of a real market does not remove company-specific risk. In fact, it makes company selection more important.

For readers following the AI infrastructure trade, this trio is worth watching not because the stocks are identical, but because the contrast is the lesson. In photonics, the future may be bright, but the winners will be determined by execution, manufacturing, customer trust, and timing — not just by being attached to the word AI.