KEY

Lasers. Pew pew! 1. Introduction 2. The Company 3. Giga-Chads and Giga-Watts 4. Co-Packaged Optics 5. Optical Circuit Switching 6. Taking Share from Competitor 7. Industry Consolidation 8. Tariffs 9. Short Interest 10. Catalysts 11. Positions or Ban **1. Introduction : The Exploding Demand for Optical Bandwidth** As AI clusters and cloud infrastructure demand more and more networking bandwidth, electronic links in the datacenter are being confined to shorter and shorter distances. Yet as these datacenters promise to consume exponentially more power to satisfy this massive bandwidth, operators are disaggregating them into disparate physical sites up to hundreds of miles apart in order to reliably deliver them the giga-Watts of power they require and adequately cool them. Networking infrastructure both within and between datacenters is being increasingly built with photonic links. For the same reasons -- power, latency and reliability being paramount -- operators are giving renewed attention to optics for datacenter switching in a slow but growing move away from traditional electrical packet switching. I believe Lumentum ($LITE) is the optical product maker best-positioned to capitalize on these near- and far-term transformations. **2. The Company : Cloud Revenue Inflecting Upward, Massive Purchase from NVIDIA** Lumentum earns roughly 90% of its \~$500MM quarterly revenue from its optical transceivers and lasers chips for cloud and networking applications. It earns the remaining 10% from industrial and consumer segments, neither of which are covered here. Its primary moneymakers are transceivers capable of up to 800G transmission and lasers capable of up to 200G modulation. Its EM Lasers are used as the light source for optical datacenter interconnect (DCI) and its narrow linewidth lasers light up 400ZR and 800ZR coherent optical links capable of 100km transmission. Operators like Meta, Microsoft, and AWS are deploying these 400ZR, 800ZR and eventually 1600ZR coherent links in massive quantities to connect up their datacenters and AI factories. These fiber links are lit by narrow-linewidth lasers, of which Lumentum's are considered best-in-class due to the higher yields on them they realize compared to competitors' lasers, per management commentary. This capability gives the company favorable "pricing latitude" and "price levers" per the CEO that, per the CFO, would provide further upside to the company's already-improving margins. Lumentum's experience in developing high-power lasers for subsea fiber links gave them the foundation for development of the ultra-high-power lasers that NVIDIA has turned to for its 200G photonic-enabled switches with co-packaged optics (CPO). This commitment from NVIDIA, announced last quarter, is Lumentum's largest-ever purchase order. This arrangement is just in its early innings. Lumentum's cloud-facing revenue is inflecting upward and drove Q42025 revenue above even the company's revised guidance. Cloud revenue grew 16% QoQ and 67% YoY. Management cites "exceptional" demand from hyperscalers for both their transceiver modules and laser components. Their modules revenue grew 50% QoQ, and management expects their components business to have a breakout year in 2026. The company has expanded its US manufacturing capacity in order to produce more of its components State-side and meet the heavy demand for its products. **3. Giga-Chads and Giga-Watts : Distributed AI Factories and "Scale-Across" Infrastructure** That giga-chad Mark Zuckerberg is building giga-Watt datacenters. NVIDIA is building networking switches with integrated photonics that connect "AI Factories" to millions of GPUs across physical sites. Microsoft is spending $80bn on its regional Azure datacenters. The massive energy consumption of these networked elements necessitates that they be disaggregated into multiple physical sites, all of which are linked by high-speed optics. NVIDIA terms this the "scale-across" network, expanding on their "scale-up" and scale-out" capabilities. These \~100km links are built today with 400ZR/ZR+ coherent optical technology, which is ramping 30% YoY, while the industry is rapidly transitioning to 800ZR/ZR+ with 1600ZR/ZR+ on the horizon. Coherent optics require the high-power narrow-linewidth lasers that Lumentum specializes in: coherent module and DSP makers like Cisco, Ciena and Marvel are shipping as many of these as they can build. Lumentum signaled strong demand for these lasers in its 3Q24 report. Lumentum's 100G and 200G EM Lasers are also in high demand. Providers are turning to 100G and increasingly 200G optics to satisfy the demands of their AI workloads in scale-out networks. Lumentum's 200G EM Lasers are on allocation and allow the company to take price as mentioned above. **4. Co-Packaged Optics : Key Ecosystem Partner in NVIDIA's "Scale-Out" Networks** Electrical interconnects within the datacenter are also being migrated to optical technologies. Co-packaged optics (CPO) are the long-term answer to meeting AI networks' requirements for low latency, low power, high reliability and high faceplate density. Major switch makers like Broadcom, NVIDIA and Cisco are integrating optics right next to the switch silicon to simplify the connection to the optical fiber and eliminate the cost and power of the DSP that the switch otherwise has to connect to in a traditional external pluggable optical module. NVIDIA demonstrated that their switches with CPO photonics engines consume 67% less power than their pluggable switches. CPO will eventually cannibalize the intra-DC pluggable module business. CPO requires high-power lasers to light up the link's optical fiber. Last quarter, Lumentum reported their largest-ever PO, for their high-power lasers for a 200G CPO application. This is almost certainly for NVIDIA's CPO photonic switches. Lumentum appears for now to be the only vendor approved by them for these lasers: "We believe we have a good competitive moat that will keep us in that position (of being sole-sourced)". I expect to hear of follow-on orders for CPO lasers in Lumentum's upcoming quarterly call: The company expects a "significant revenue ramp in CPO by the second half of calendar 2026." Lumentum manufactures these lasers in a US fab, exempting them from any tariffs on this revenue, while also expanding capacity. So, they are the vendor with the largest capacity to produce these lasers, of which theirs are the industry's best-performing, while being the sole supplier approved for NVIDIA's CPO switches. A knock on CPO has been the perception of its inferior reliability compared to traditional pluggable modules. Meta dispelled this at the ECOC conference last month, stating, "While CPO’s value in substantially reducing power over traditional re-timed transceivers has been well established, the integrated nature of CPO and its impact to the reliability, availability, and serviceability at scale has been elusive... Within this initial time period spanning over 15 million device hours, we have observed zero unserviceable failures as well as a 5x improvement in link reliability compared to traditional pluggable optical modules... As we continue to explore innovative solutions offering reliable and power-efficient optical connectivity for the AI networks, these CPO reliability results reinforce its viability to address the future optical I/O challenges of data center network infrastructure." CPO has been a technology long on the horizon of the optical networking industry, but its widespread adoption is imminent. **5. Optical Circuit Switching (OCS) : An New and Expanding Market** In the face of ballooning power DC and AI factory power consumption, hyperscalers and DC operators are showing renewed and increasing interest in OCS as an alternative, if not replacement, to electrical packet switching. OCS is not new, having had its moment in the sun in 1990s and early 2000s. Google was perhaps the first major DC operator to deploy it at scale in recent years, finding that it improved power by 40%, cost by 30% and uptime by 50x. OCS is particularly suited to AI applications because of its low latency and high reliability. Last quarter, Lumentum reported the shipment of OCS evaluation units and more recently has announced additional OCS products, likely in response to customer interest. They have two hyperscalers signed onto their OCS platform with a third committed for 2026. Management said last quarter that OCS will bring "very, very significant revenue in Q1, Q2 and then certainly in the back half of calendar 2026." So that's a second significant 2H2026 revenue inflection. OCS is now a $100MM+ revenue source for the company with margins "significantly" above the corporate average and a time frame to significant revenue that is becoming "sharply clear that it's sooner rather than later." **6. Taking Share from Competitor** Coherent ($COHR) is Lumentum's primary US-based competitor vying to supply the networking industry with optical components (lasers, OCS, transceivers). We noted that Lumentum is sharply growing its cloud and datacom business. This appears to be coming at Coherent's expense, because even in this era of massive spending on network infrastructure and heavy demand for optical gear, they issued weaker-than-expected guidance last quarter. Their datacenter revenue growth continues to slow, to 24% from the prior quarters' 39%, 46% and 58%. Investors punished them severely for this miss, sending the stock down 20%. Lumentum appears to be taking share: their datacom growth stands at 44% while Coherent's is only 4%. Lumentum as noted is presently the only laser vendor approved by NVIDIA for its CPO switches, while Coherent made no mention of any new POs or wins in the CPO space. Lumentum has the incumbent advantage here. And in OCS, management states that they have been able "to capture volume opportunities earlier than competitors." **7. Industry Consolidation : Component Makers Get Snapped Up** The recent history of the optical networking industry shows numerous acquisitions. Barriers to entry have grown as bandwidths increase, dimensions shrink and semiconductor mask costs reach into the tens of millions. Smaller firms that develop key enabling technologies or components have been the ones to be snapped up. Lumentum CEO Michael Hurlston has been through to this. He led Finisar when it was acquired by II-VI (now Coherent) in 2018. On the flip side of the ledger, Lumentum acquired Neophotonics in 2021 to capture its expertise in building the narrow-linewidth lasers that have become a lynchpin of the coherent optics segment. For motivations similar to those expressed in the quotes below, I believe Lumentum is on the short list of photonics suppliers that a larger system integrator, module maker or certain GPU networking giant would seek to acquire. NVIDIA in particular could acquire Lumentum to secure its supply of lasers for its CPO switches, while also bringing in-house the engineering expertise needed to incorporate optical switching into its back-end AI network installations. Small OCS matrices like the 64x64 product Lumentum just announced would be well-suited to this application. This of course is more speculative than my arguments of where Lumentum's near-term business is headed. Nonetheless it's a feasible outcome for the company and one I want exposure to. * Nokia acquires Infinera for $2.3B, Feb. 2025: "This transaction will significantly improve our scale and profitability in optical networks, and allows us to speed up the pace of innovation to meet the requirements of the AI era." * II-VI acquires Coherent for $6.6B, July 2022: "Coherent is an innovator with a rich portfolio of some of the most advanced technologies in the world, which have been transformative in a broad range of markets." * Lumentum acquires Neophotonics for $900M, Nov. 2021: "With NeoPhotonics, we're making another important investment in better serving our customers and expanding our photonics capabilities at a time when photonics are at the forefront of favorable long-term market trends." * Marvel acquires Inphi for $10B, Apr. 2021: "Inphi’s technologies are at the heart of cloud data center networks and they continue to extend their leadership with innovative new products, including 400G data center interconnect optical modules, which leverage their unique silicon photonics and DSP technologies." * Cisco acquires Acacia for $4.5B, July 2019: "With the explosion of bandwidth in the multi-cloud era, optical interconnect technologies are becoming increasingly strategic. The acquisition of Acacia will allow us to build on the strength of our switching, routing and optical networking portfolio to address our customers' most demanding requirements." * Cisco acquires Luxtera for $660M, Feb. 2019: "Coupled with our silicon and optics innovation, Luxtera will allow our customers to build the biggest, fastest and most efficient networks in the world." * II-VI acquires Finisar for $3.2B, Nov. 2018: "Today our company is taking a giant leap forward in our scale to serve a significantly increasing addressable market." **8. Tariffs : Not a Threat** Tariffs have not materially affected Lumentum's financial performance and management has stated that they do not anticipate them to. But with our mercurial president at the helm, the risk that tariffs will be levied on optics is ever-present. **9. Short Interest : 15% SI, 5-Day D T C and 120% Institutional Ownership** While lower than the crazy 100%+ SI values we've aped on here, I consider 15% SI to be substantial, especially for a company in as strong of a position as Lumentum's. The stock has risen 200% over the past six months, during which time SI has hovered between 15% and 18%, implying they're in a neutral-to-losing position. The stock is on at least one broker's hard-to-borrow list. Its 4.7-day D T C and 120% institutional ownership imply a tight float. If Lumentum publishes a strong earning report on Nov. 4 as I expect they will, one could envision a rush to close these short positions. I have a hard time seeing the bear case here. **10. Catalysts** As mentioned, Lumentum reports 2026Q1 results Nov. 4 after market close. I expect a solid earnings beat and positive guidance around additional CPO-related orders, 200G EM L laser volumes and favorable pricing, and inflecting OCS revenue. Hearing of continuing high infrastructure CapEx from Meta, Microsoft and Amazon when they each report this week should also be viewed as beneficial to the stock. A rush to cover short positions in response to a beat would provide further upward momentum. **11. Positions** I own 2100 shares, worth $440k at today's $191 share price. I opened Nov. 7 230C for leverage to Lumentum's earnings release, and Nov. 21 250C for leverage to NVIDIA's (Nov. 19). I may also buy calls with a further expiration to remain exposed to the longer-term industry migrations detailed here. https://preview.redd.it/6ztf5n3dp3yf1.png?width=2115&format=png&auto=webp&s=cb1f9b523abac3572e3b27df6d4304e5d402db05 https://preview.redd.it/uzoo1modp3yf1.jpg?width=1179&format=pjpg&auto=webp&s=8ae21e5142142843a8c65083e1feccc29836d3ec **TLDR:** Lasers. Pew pew!

*TL;DR at the bottom for all of you with no attention span* On the 8th of October 2024, Hindenburg Research released its damning report on RBLX: [Roblox: Inflated Key Metrics For Wall Street And A Pedophile Hellscape For Kids](https://hindenburgresearch.com/roblox/). Hindenburg detailed several inflated key metrics, unconfident management, and many instances of sexual abuse, grooming, and a lack of safety measures. If you don't know, Hindenburg is famous for their excellent and detailed short theses; they are known for for exposing companies such as Carvana, Nikola, Icahn Enterprises, Singularity Future Technology, and many more. They take on short positions in the companies before posting their reports, which often crash massively following the publication. The reports feature high-quality investigative journalism, talks with insiders, and investigations from third-party consultants. Unfortunately, Hindenberg Research shut down in January 2025, with their last report being on Carvana. This post will be built upon the original report, with new information since the publication, and my own perspectives and conclusions. I really recommend reading the original article, as it goes into much more detail than I will. Surprisingly, unlike most of their reports, it had little effect on the stock price. When the article was released, the stock was trading at $40; now it's trading at $132 - a 230% gain. This rally has been fueled by Roblox reporting tremendous growth. In Q2, they reported a 51% increase YoY in Robux spend, a 41% increase YoY in active users, and a 21% increase YoY in revenue. It is a seemingly amazing growth story, with Cathie Wood on board, holding about 1.4% of outstanding stock. They currently have 110M Daily Active Users (DAU) and have a long-term vision of 1B people playing Roblox. Despite the surge in stock price, the flaws that Hindenburg pointed out are still in place, future growth is uncertain, and the path to profitability seems very unlikely. # The Financials (*One of the most important financial measures that Roblox releases in their reports is* ***bookings***\*. This is the amount of money spent on Robux, a virtual currency used for in-game transactions, and it represents almost the entire income of Roblox Corporation. The bookings number is then deferred into revenue over as many as 27 months, which means that revenue is lagging behind the bookings number.\*) Since its IPO in 2021, Roblox's cumulative net losses amount to almost $4B, and it has never earned a profit in a single quarter. Just in the last twelve months, they lost $952M. These losses have also partially been funded with Roblox issuing new shares. Since the IPO, they have issued more than 100M shares, increasing the amount of shares outstanding by 20% and diluting shareholders. But maybe at least management believes in this growth story they built? Unlikely. Since IPO, insiders have cashed out $1.7B, with the CEO personally cashing out $526M. Also, several key C-suites have recently left; the former CFO, Michael Guthrie, who joined in 2018, left in June this year, and the CMO and CTO, who joined in 2020, both left in 2024. This signals low trust in the long-term prospects of Roblox. Right now, Roblox is sitting at a market cap of 91B, which makes it one of the largest gaming companies, although it trails behind some giants like Nintendo, Tencent, Sony, and Microsoft. It is trading at a very high price to sales ratio of 22, which can be compared with Nintendo at 10, Take-Two at 8, and EA is being bought out by the Saudis at a price to sales of 'only' 7. Over the last twelve months, RBLX managed to make almost $1B in free cash flow, but that still places them at a price to free cash flow of nearly 90. As Hindenburg reported, the amount of Daily Active Users is likely overstated by 25-42%, due to them counting at least some alt and bot accounts. They have internal metrics that correct or 'de-alt' these numbers, but management chooses to still report these higher figures, as retroactively changing the numbers would look troubling for potential investors. There are currently 9.82B registered Roblox accounts. Yes, there are more Roblox accounts than people on this earth. What a time to be alive! Most of these accounts are likely alt accounts or bot accounts and do not represent real people. They also report that DAUs are spending an average of 2.4 hours per day on the platform. According to Hindenburg's own investigation, the real figure could be as low as **22 minutes**. This is a key metric that is important to advertisers who want to buy in-game advertisements, which is a source of income that management is really pushing. In 2024, they claimed that only 20% of hours engaged are effectively commercialized; management wants to push this number to 100%. They want to make sure that these poor kids never experience a moment of peace without being bombarded with ads or being pushed to purchase useless cosmetic items. There are also doubts as to whether companies will want to advertise on a platform infamous for child groomers and games such as 'Escape from Epstein Island'. Furthermore, there is a limited amount of money to be made in advertising to people without any income. **Roblox is already saturating its most profitable markets** The by far most lucrative market for Roblox is the US & Canada. In Q2, they earned an average of $40.69 in bookings per daily active user (ABPDAU) in the US & Canada region. This means that American and Canadian daily active users are spending on **AVERAGE** $163 per year on Robux! (Note: this does not mean that the average Roblox player is spending $163, as the total amount of paying users may be higher than the daily active user count, but it is nonetheless really high.) The next most profitable market in Q2 was Europe, where DAUs were spending on average $11.49 on bookings. Whereas for Asia that figure was around $4.95, and for the rest of the world it was only $4.29. Let's determine just how saturated the US, Canada, and Europe markets really are. 81% of DAUs are in the age range of 5-24 years old. I will assume that they also make up about 80% of the bookings. There are about 95 million 5-24-year-olds in the US & Canada, while in Q2, Roblox reported 20.6 million daily active users in the region. This means that about 17% of the 5-24 year old population are DAUs, which means that on any given day, 17% of 5-24 year olds are going to play Roblox. In reality, with the overreporting of users, that percentage is maybe closer to half, but it really seems like the market is starting to saturate. Europe has a 5-24 year old population of about 160M and 25.6M DAUs, leading to a saturation percentage of about 13%, which I believe to also be close to saturation. Also, the number of newborns in Europe is declining rapidly, leading to a smaller potential future player base. They are pushing for growth in Asia, especially, but the profitability there is much lower, and parents are stricter with how their kids spend their time, and especially money. For the sake of argument, let's say that Roblox manages 15% saturation in the 5-24-year population in Asia and the rest of the world. With the current average bookings per DAU in these markets, I have calculated that Roblox would then make 12.86B in total bookings per year, compared with the current number at 5.75B per year. So, with 15% of all kids in the **ENTIRE** world playing Roblox, they would only increase their revenue by 120%. Let's say that they somehow managed to make a 20% net profit margin on these hypothetical bookings (current gross profit margin is about 25% and net profit margin is NEGATIVE 25%), then they would still only make 2.57B in net profit. That would correspond to a PE ratio of 35! However many billions of dollars spent on Robux, it is still hard for Roblox to actually make a profit. They spend a lot of money compensating user developers for their games, they spend a lot on salaries for software engineers in San Mateo, California, and they spend a lot on infrastructure and safety. To be clear, without this spending, there would be no growth. I believe they can turn a profit eventually, but I think it will be less than most analysts have baked into their spreadsheets. # The Pedo Problem The Hindenburg report revealed an "X-rated pedophile hellscape, exposing children to grooming, pornography, violent content, and extremely abusive speech". They reported how the company prioritized growth over safety, and how, despite a massive increase in users, their spending on trust and safety expenses has actually declined. There have been many instances of child groomers using Roblox to find vulnerable victims on the platform. The National Center on Sexual Exploitation in 2024 labeled Roblox “a tool for sexual predators, a threat for childrens’ safety”. There have even been criminal indictments for kidnapping and rape related to the platform. Still, there continue to be new criminal cases brought against Roblox for the lack of safety for minors on the platform. The Hindenburg reported how there were open pedophile rings on Roblox where users were trading child pornography, grooming children, and discussing pedophilia. There has been some improvement in safety since the Hindenburg report. But it is still not enough, and there has been permanent damage to the reputation of Roblox and parents' trust in the game. Since the Hindenburg report, they have rolled out age estimation technology using AI facial scans (don't think about the privacy concerns of scanning millions of minors' faces pls). They have also apparently improved chat and voice filters and added more parental controls. However, there are still issues. For instance, take the National Center on Sexual Exploitation article that was released in August: [Roblox’s Predator Problem and Why Federal Inaction Lets Them Get Away With It ](https://endsexualexploitation.org/articles/robloxs-predator-problem-and-why-federal-inaction-lets-them-get-away-with-it) They recently banned the YouTuber Michael Schlep, who is known for exposing sexual content and child predators on the platform after he himself was groomed by a Roblox developer as a teen. Roblox banned him on the basis that they do not permit 'vigilante' activity. Rather than focusing on banning child predators, they are banning users who expose them. Roblox has also lobbied and continues to lobby against new online child protection laws. Republican representative Laurie Schlegel tweeted in August: [For the last couple of years, I have passed a number of child online protection bills. And every year, Roblox hires lobbyists to oppose and actively work to stall my bills. This session, a Roblox executive personally came down to Louisiana to work against my Duty of Care bill](https://x.com/RepSchlegel/status/1956106631173038251) Several high-profile cases have been brought against Roblox, and new ones are being started all the time. Here are a few noteworthy recent examples: * A 15-year-old Californian boy was groomed via Roblox and Discord, and eventually took his own life in April 2024. His mother is now suing both companies. [link](https://www.nbcnews.com/news/us-news/roblox-discord-sued-15-year-old-boy-was-allegedly-groomed-online-died-rcna231049) * A Wisconsin 5-year-old was sexually exploited on Roblox. They filed a civil lawsuit against Roblox on October 7th. [link](https://eu.jsonline.com/story/news/crime/2025/10/08/wisconsin-5-year-old-girl-sexually-exploited-on-roblox-lawsuit-says/86585932007/) * Michigan lawsuit claims 10-year-old was sexually exploited on Roblox [link](https://www.clickondetroit.com/news/local/2025/08/16/michigan-lawsuit-claims-10-year-old-girl-was-sexually-exploited-on-roblox-app/) * State Attorney Generals from Florida, Louisiana, and Kentucky have all announced legal action against Roblox. [link](https://www.bitdefender.com/en-us/blog/hotforsecurity/florida-roblox-attorney-general-investigation), [link](https://www.bitdefender.com/en-us/blog/hotforsecurity/roblox-child-protection-lawsuit-safety-policies), [link](https://www.cbsnews.com/news/kentucky-attorney-general-russell-coleman-sues-roblox/) * California opened a class action lawsuit against the company in May,  [link](https://www.bitdefender.com/en-us/blog/hotforsecurity/roblox-class-action-lawsuit-data-kids) In August, WIRED also wrote an article called: [Is Roblox Getting Worse?](https://www.wired.com/story/is-roblox-getting-worse/). Matt Dolman from Dolman Law Group told WIRED that “I would assume by the end of September there should be about 100 to hundreds of these \[lawsuits\] pending, and I would assume by this time next year you'll probably be looking at over 1,000 of these filed”. He also says that “We alone already have about 300 of these cases.” Dolman says the vast majority of his clients are under the age of 16 and estimates around 60 percent of the cases involve girls. If anything, it seems like the problem is getting worse, not better. # Conclusion RBLX is a pricy stock. Although the growth seems great, they are already approaching saturation in their most lucrative markets. They have been found to fudge their numbers, but will continue to do so since no one is stopping them. RBLX has never made a profit, and it will likely be very long until they do (if ever). Despite implementing new safety measures, sexual abuse on the platform seems to be more rampant than ever. The legal pressure of thousands of lawsuits is mounting on them. **Catalysts** * Missing growth targets * Dilution of stock over time * Financial misaccounting * Lawsuits **Position** * RBLX Jun18'26 120 Put **TL;DR** Hindenburg exposed RBLX fudging numbers in 2024. They have never made a profit; meanwhile, they are diluting their stock, and insiders are cashing out. Despite the current high growth, the room for future growth is very uncertain. The problem of sexual abuse and grooming on Roblox seems to be growing despite Roblox implementing new 'Safety Measures'. They are facing hundreds of lawsuits while trying to attract advertisers to advertise in-game.