Forced Alpha

82 posts

Forced Alpha

@forced_alpha

Supply chain knowledge graph covering AI, defense, energy, and robotics. We find the small cap gating billions in downstream value before anyone else does.

Katılım Şubat 2026

17 Takip Edilen10 Takipçiler

Sabitlenmiş Tweet

Forced Alpha@forced_alpha·2d

Over $13 trillion in market cap at the top of this chart depends on a handful of companies most people have never heard of. We've mapped the full CPO supply chain. 10 layers, 50+ companies. The red nodes are the bottlenecks. The arrows show who supplies who. A $231M laser die company. A $1.6B micro-lens maker that's the sole vendor for TSMC. A foundry with 70%+ capacity booked through 2028. The chokepoints aren't where the market is looking. Link to full data in the first comment.

English

136

Forced Alpha@forced_alpha·9h

$NVIDIA just confirmed CPO is in full production at $GTC 2026. The demand signal is public. But the critical substrate and wafer suppliers are Japanese and French companies that US investors literally cannot track through 13Fs or options flow. Public demand + invisible supply chain = mispriced bottlenecks = ForcedAlpha.

English

Forced Alpha@forced_alpha·11h

Agreed. We flagged $SIVE a few weeks ago through our supply chain graph. But, it wasn't one we really dove into until we built a bottleneck propagation layer that flows material level constraint scores to supplier companies. InP substrate bottleneck > CW DFB laser demand > SIVE as one of ~3 qualified suppliers. The asymmetry math: $300M company supplying laser sources that feed into $45B+ downstream optical value. That's the same structural pattern as AXTI before it re rated. Small substrate/component supplier gating massive downstream value with no easy substitution. The CEO buy (472K shares, March 6) was the confirmation signal too. Honestly, your content and what we're seeing here in the data alligns 100%.

English

346

Serenity@aleabitoreddit·1d

$SIVE <> $SIVEF is now up 165% this week. Valuation? ~$300M MC. However; either I’m dumb or Sivers is one of the best opportunities in photonics today. You get the laser supplier for Jabil, Ayar, Poet ( $MRVL Celestial ), O-Net, and others: That end up in $GOOGL, $MSFT, $AMZN, $META AI datacenters. At ~$300M. The EML laser suppliers today from $LITE to $COHR for reference are $45B+ This is one of the most undiscovered yet critical bottlenecks for future upstream photonics supply chains. That markets have only starting to price in today.

Serenity@aleabitoreddit

I’m long $SIVE at $140M. I believe this is the next $LITE that markets and institutions missed. $SIVE makes InP CW DFB lasers. Closest comparison is $LITE in the current EML laser bottleneck. But instead of supplying to Innolight/Eoptolink for current optical transceivers cycles. They supply the lasers to $POET Starlight, Ayar SuperNova. And others for the future CPO/silicon photonics architectures spearheaded by $NVDA. Current valuations make 0 sense to me personally.    $POET is advanced packaging for $SIVE type lasers… But $POET commands worth 11x+ more than the company making the laser itself?   It’s feels like valuing a more advanced $FN (~$20B) packaging at $400B when $LITE is valued at $40B.    So now at $130m:  - - You have a likely mini $LITE like laser supplier to Marvell Celestial + hyperscalers through $POET.    - Laser supplier to Ayar ( $NVDA, $INTC ), though they do multi source with $LITE, Sumitomo, $MTSI. And other potential up and coming suppliers potentially like Lightmatter that they’ve name dropped (eg. Q2 2023 earnings). This is unconfirmed but supply chain BOM is confidential.    On top, for revenue, they expected $453M "pipeline next few years”.   And, they have capacity expansion through WIN: “Win Semi foundry qualification in progress for volume production from Laser designs from Sivers."   Sivers feels the silicon photonics/CPO version of $LITE, with actual rapidly growing customers like Celestial through $POET, Ayar, with more to come.   I wouldn’t have liked it last year, but just 3 weeks ago, they refinanced all their debt successfully to $12M convertible loan (10.85%) and a $5M term loan (12%), which cleans up debt.   It’s $17m total, which feels like nothing to US markets when $AAOI is doing a $500m ATMs every other week. Best of all, this is their pure play inp laser segment for silicon/photonics + cpo.   Their Lidar segment is ramping up and they have $53-138M projected revenue coming in.   Downside risk:  - execution (as always)  - dilution to scale up capacity to compete with $LITE and others. - $LITE, $COHR competition on scale after $NVDA just gave them $4B - CPO ramp gets delayed.   I have no clue how, $LWLG, a pre-revenue science project with $TSEM, is valued at $1B+ MC.   Or how $POET, is worth ~9-10x more than its laser supplier.    When $SIVE, the mini $LITE equivalent for CPO/Silicon photonics, is valued at $140M. I do believe this is largely undiscovered by institutions, since this is some random company in OMX Nordic Exchange (similar to micro $AXTI before I started posting about the inp substrate bottleneck).    But I do think it will get a lot of institutional attention as Celestial and Ayar scale up. Especially if $POET and $SIVE gets qualified with other customers.    If CPO completely replaces pluggable transceivers in the next generation of hyperscaler architectures. Sivers, with possible WIN Semi qualifcation and if they become the multi-source lasers for NVIDIA, Marvell, Intel, and Broadcom architectures, can be strongly rerated. Just as how $LITE did today going from $16 -> $622. This is just my personal thesis I'm sharing, DYOR/NFI. TLDR: InP Lasers are the current bottleneck in photonics as seen with $LITE valuations.    $SIVE looks like the mini $LITE for the upcoming CPO/Silicon Photonics ramp.   I personally took long position in $SIVE, as I believe they’re a large beneficiary of the upcoming silicon photonic/CPO architectural changes by $NVDA (with GTC cataylst).    The upside here just way too compelling for me personally as the next possible $LITE.

English

431

142.7K

Forced Alpha@forced_alpha·13h

Both. The initial qualification is the big one. 18 months for a fab to certify a new InP substrate supplier. That covers defect density testing, device yield validation, reliability burn in aka the full stack. But it also matters for ongoing production. If your qualified supplier has a disruption and you need to switch, you're back to square one on that 18 month clock. You can't just swap in a different vendor's wafer and keep running. Each are tuned to specific substrate characteristics e.g dislocation density, surface roughness, epi ready specs. That's what makes it a chokepoint. There are only 3 merchant InP substrate suppliers globally. If one goes down today, you don't have a backup you can call on Monday.

English

Operation🅰️TM@OperationATM·15h

@forced_alpha @KawzInvests As in InP Substrates require 18 month qualification cycle - this is in the initial setup or for ongoing production? Excuse my ignorance.

English

KawzInvests 🦑@KawzInvests·2d

Building a memory fab takes 4 years. Building a photonics fab takes 9 months. There is a MASSIVE difference between the build out for Photonics vs Memory A memory fab is a precision lithography operation. You are packing billions of transistors at single-digit nanometer nodes. EUV tools alone take 12-18 months to procure and calibrate. The yield ramp after that takes years. The bottleneck is physics and it cannot be compressed. A photonics fab is an INTEGRATION PROBLEM. You are building devices that manipulate light, not electrons. Indium phosphide. Optical waveguides. Alignment tolerances measured in nanometers of coupling efficiency, not transistor density. No EUV required. The practical timeline difference: Samsung Electronics, $MU, SK Hynix 3 to 5 years from groundbreak to meaningful output. The lithography learning curve is non-negotiable. $AAOI with an existing warehouse 9 months. Not because construction is faster. Because they skip the construction problem entirely. Cleanroom retrofit, tool installation, and process bring-up run in parallel. Most companies do these sequentially. AOI does not. $AAOI has a massive automation advantage AOI runs internal testing systems at 20x the throughput of standard industry equipment. Their product platforms are standardized to the point where each new production line is not a new engineering problem it is a deployment. When they enter a new facility, they are not figuring out the process. They are executing a template they have already optimized across years of production in Taiwan. That is exactly what is happening with their new Texas facility. AOI is not building something new. They are replicating the same factory format, tooling layout, automation systems, and process templates that are already running and yielding in Taiwan. The institutional knowledge, the yield data, the calibration baselines all of it transfers. A semiconductor company standing up a new node from scratch has none of that. AOI walks in with the answer key. Vertical integration across lasers, PCBA, and final assembly means there is no external dependency introducing variance into yield. They own the entire feedback loop from wafer to finished transceiver. That matters because of what the real bottleneck actually is. Most people stop the analysis at fab timelines or InP supply. Both are real constraints. Neither is the hardest part. The hardest part is thermal qualification. A transceiver operating inside a hyperscaler switch runs continuously. These switches need to operate at full load 24 hours a day for the unit economics to justify the infrastructure spend. If the switch is down, the compute behind it is idle. At the scale hyperscalers operate, idle compute is not an inconvenience it is a direct hit to the return on billions of dollars of capex. The failure mode that defines vendor selection is thermal. Transceivers generate heat. Heat degrades the laser. A degraded laser causes signal loss. Signal loss in a switch port takes that segment of the switching fabric offline. Hyperscalers do not tolerate partial switch failures they replace the vendor. This is why qualification cycles are the longest stage of the entire ramp, not manufacturing. Hyperscalers test interoperability, sustained thermal performance, and reliability under continuous full load before committing volume. A vendor that cannot demonstrate 24/7 thermal stability does not get the contract regardless of how fast they built the factory. AOI's vertical integration is a direct solution to this problem. Because they control lasers, PCBA, and assembly in-house, they control the thermal envelope of the finished product end to end. Competitors are integrating components from separate vendors and discovering thermal variance late in qualification. AOI is designing the thermal system, not assembling one from parts. Their automated testing infrastructure means thermal issues surface during production, not during the customer's qualification cycle. That compresses the single longest stage in the entire ramp. And because the Texas facility is a copy of Taiwan, that thermal system arrives pre-validated. They are not learning how to build a thermally stable transceiver in Texas. They already know. They are just doing it closer to the customer. Memory Manufacuturing bottleneck = lithography Photonics Manufacuturing bottleneck = thermal qualification The structural thesis is sound. But there is always a layer of entropy no model accounts for. Execution risk does not disappear because the framework is good.

KawzInvests 🦑@KawzInvests

Applied Optoelectronics $AAOI released their capacity ramp plan. We are in Q2 2026, which means this timeline has already started. Current monthly revenue: $64 million. By Q4 2026, guidance puts that at $303 million per month. By Q4 2027, $701 million per month. These are monthly figures, not annual. Here is why those numbers are achievable. Transceiver pricing does not compress as speeds scale. It expands. 100G sells at $20 per unit. 400G at $64. 800G at $320. 1.6T at $640. Each speed generation commands a price that far outpaces the throughput increase, because the engineering complexity at each step grows significantly. The revenue mix tells the real story. Today, 800G contributes $44 million per month and 1.6T contributes $6.4 million per month. By Q4 2026, 800G reaches $134 million and 1.6T reaches $147 million per month. The ELSFP module, which is not in the revenue mix today, scales from $3.2 million per month in Q4 2026 to $256 million per month by Q4 2027. This is ASP mix shift, not volume growth. Each product cycle that ships carries materially more revenue per unit than the one before it. The execution risk is the manufacturing facility ramping on schedule. Even at 70% of guidance, the revenue trajectory is not priced into the current valuation $AAOI $COHR $LITE

English

358

90.8K

Forced Alpha@forced_alpha·21h

ZXX

Forced Alpha@forced_alpha·21h

The same facility also feeds Qatar's entire fertilizer chain. That cascade is already live Urea +32% in a week, 150+ tankers stuck outside the Gulf, planting season happening now. One facility. Two cascades. One hits semiconductors. The other hits food. This is what supply chain intelligence looks like. Not after the headline. Before it.

English

Forced Alpha@forced_alpha·21h

The Killer shot. Iran hit Qatar's $raslaffan gas facility. The real cascade is $helium not oil. Ras Laffan produces ~30% of global helium. There is no public spot market. There is no strategic reserve. The US sold theirs off in 2023.

English

Forced Alpha@forced_alpha·21h

We mapped this with a supply chain knowledge graph. Here's what it shows: - Helium sits upstream of 83 downstream nodes. Remove it and the graph fragments. - Industry buffer: 44 days. Restock time: 16 weeks. Demand growing 3x faster than supply. - The crossover was already projected for Q3 2027. Today just pulled it forward.

English

Forced Alpha@forced_alpha·21h

The cascade: → ~30% of helium supply offline → Spot prices up ~50% in weeks → Smaller fabs hit buffer limits within weeks. Even TSMC and SK Hynix have months, not quarters. → HBM production constrained — already in deficit, sold out through 2026 → GPU shipments slip → AI infrastructure buildout slows

English

Forced Alpha@forced_alpha·1d

@siflower @aleabitoreddit Fair point. CW DFB is the CPO architecture, EML is pluggable. Mixed up the two. SIVE supplies both but the CPO ramp is on the CW side. Qualification moat still applies on CW DFB. Substrate and epitaxy sourcing is just as concentrated. Thanks for the correction.

English

siflower@siflower·1d

@forced_alpha @aleabitoreddit CW (Continuous-Wave) lasers are the primary type used for mainstream CPO (Co-Packaged Optics) architectures in 2026, not EML.

English

Serenity@aleabitoreddit·2d

The Photonics Supercycle is here. $NVDA is spearheading the next leap into CPO & Silicon Photonics. And we’re only near the inflection point with chokepoints in the supply chains like Soitec ( $SOI ) or Sivers ( $SIVE ). “NVIDIA’s update on the Spectrum-X switch with co-packaged optics is an important moment, confirming that silicon photonics is central to next-generation AI infrastructure.” Despite a long-standing reliance on copper-based interconnects for scale-up systems, the company is now placing photonics at the core of its future platforms, including Vera Rubin Ultra. This transition is expected to support increasingly complex configurations, such as NVL576 and future architectures like Kyber NVL1152.” “Nvidia is already in production with Spectrum-X Photonics, which is co-packaged optics (CPO) Ethernet switch. The company also announced the Quantum-X Photonics InfiniBand switch, which delivers up to 800 Tb per second of scale-out throughput using its proprietary scale-out interconnect” Although copper is important, it can no longer alone can no longer handle AI-scale demands. NVLink8 CPO is probably the biggest signal with $NVDA also bringing silicon photonics into its scale-up NVLink interconnect, not just scale-out networking. CPO for scale-out is shipping now/2026, CPO for NVLink scale-up arrives soon. The paradigm has shifted, and the bottleneck of AI infrastructure is now officially being solved by light. It’s only a matter of time before markets find these chokepoints in the supply chains. Then price them in.

Serenity@aleabitoreddit

The upcoming CPO / Silicon Photonics Bottleneck Cheat Sheet: $SIVE, Sumitomo, $LITE, $COHR, $AVGO, $MTSI, $AAOI - Light Source (CW DFB Lasers) $TSEM, $GFS, $UMC, $TSM, $INTC - SiPh foundry $NOK, $CIEN, $CSCO, $COHR - DCO $HIMX, FOCI (3363.TWO) - Micro-lens + Fiber Arrays $POET - Optical Interposers $SOI, $AXTI, Shin-Etsu - Substrates $FN, $ASX, Innolight, Eoptolink - Optical Packaging and Assembly $MTSI, $SMTC, $MRVL, $MXL - Analog/Mixed-Signal ICs $LWLG - Speculative Modulator Materials. $GLW, $APH, $TEL, $FIT, Fujikura - Connectors and Fibers $FORM, $KEYS, $VIAV, $AEHR- Test & Measurement $BESI, $SMHN, $ONTO, $CAMT - Advanced Packaging & Hybrid Bonding Many are private companies from Lightmatter, Ayar, Ranovus and others. Now... Everyone is asking... How do you profit? If you look at the forecast for CPO TAM, it's a straight line up, and next year is inflection point for CPO mass deployment. The alpha is capturing the rotation: From the current EML bottlenecks ( $LITE, $COHR type) to SiPh / CW DFB architectural winners for CPO. Highest upside potential are the ones that aren't included in current cycles. But that are in the next. Companies like $SOI, $SIVE, or $AEHR are perfect examples. Ride the current pluggable bottleneck like $AAOI. But the alpha is frontrunning institutions with the next CPO bottleneck. The capital rotation is inevitable.

English

577

96.5K

Forced Alpha@forced_alpha·2d

Full interactive version where you can click any node and trace the dependencies: forcedalpha.com/research/photo…

English

Forced Alpha@forced_alpha·2d

English

136

Forced Alpha@forced_alpha·2d

Aha! We've had it on our radar for a bit. The TFLN angle is the one people are sleeping on. Same China concentration problem as rare earths but for photonics modulators. Raytheon pulling them in for domestic production tells you where DoD thinks the vulnerability is. Plus the dual use into AI datacenter CPO is the $$.

English

Karel Capital@KarelCapital·3d

Photonics is the trade everyone on X is watching right now. $AAOI is up +173% YTD, $LITE was added to the S&P 500 and is up a lot, $COHR got a $2B Nvidia investment alongside $LITE, $AXTI is still the InP bottleneck play that started this whole rotation. But the name nobody is talking about right now: $GHH (Gooch & Housego,~£230M market cap) Here's why it matters: TFLN (thin-film lithium niobate) is the material that makes co-packaged optics (CPO) actually work at scale. It enables ultra-high-speed, low-loss modulators that push 1.6T+ optical interconnects. Without it, the next generation of AI data center switching stalls. Right now the entire TFLN supply chain is dominated by one Chinese manufacturer, according to Raytheon. G&H just got pulled in by Raytheon on an Air Force Research Lab contract to build the first domestic US TFLN wafer production line. Production transfer to G&H's Ohio facility is already underway in 2026. This isn't just a defense angle. The same wafers go into AI data center photonics, telecom, and quantum. Small cap. Vertically integrated crystal grower. Raytheon's chosen partner. Trading at a fraction of its US photonics peers. The big names get all the headlines. G&H gets the contract.

English

9.9K

Forced Alpha@forced_alpha·2d

More moneyshots! Capacity is the bottleneck right now in this market. All EML lines sold out years forward. AAOI's sugarland expansion is basically pre sold. TBH the question isn't demand it's whether they can actually ramp fast enough. InP wafer supply upstream is the real bottleneck.

English

1.1K

Serenity@aleabitoreddit·2d

$AAOI looks very undervalued at $6.49B. If we model ASP and their newest capacity projections today: Revenue from Capacity: Q2 2026: ~$312.1M Q4 2026: ~$1.41B Q2-2027: ~$1.53B Q4-2027: ~$1.97B This is absurd ramp (off ~34-40% est. gross margins). ASP modeled off (LightCounting, Dell'Oro Group & Yole, pricing for ELSFP modules is the most speculative). And some sell-side models (from firms like Raymond James, B. Riley, Northland Capital, and Goldman Sachs). Exact contract pricing for massive volume orders is not known, so this is speculative. But the Q2 volume * ASP estimates actually align with their $378M/month target Q2-2027. Again, you might be wondering? This is capacity, doesn't translate into revenue right? Hyperscalers from $AMZN to $MSFT are buying any capacity any of these companies from $LITE to $COHR can make, years out. This includes $AAOI from their former earnings call.

English

826

116K

Forced Alpha@forced_alpha·2d

Great breakdown. What gets me is what this means upstream. Coherent pushing inward = exponential demand for InP lasers and photonics. Companies making EMLs and CW lasers for all of this are a handful of suppliers globally. Everyone's mapping the coherent vs IM/DD transition but few are looking at who actually makes the light sources.

English

Nutty@NuttyCLD·4d

x.com/i/article/2033…

ZXX

62.1K

Forced Alpha@forced_alpha·2d

@aleabitoreddit Moneyshot right here! SIVE at $200M looks like the same pattern as AXTI, just one layer down. Once you're qualified into these CPO programs you're locked in. Time to ramp up. We're mapping this and others with knowledge graphs.

English

1.5K

Serenity@aleabitoreddit·2d

$SIVE is the upstream laser supplier for CPO and Silicon Photonics. They're the likely $COHR / $LITE type future light source for: - $AMZN Trainium Clusters - $MSFT Maia Clusters and possibly other hyperscalers like $META MTAI and $GOOGL TPU clusters. At a ~$200M MC. Relational Mapping (speculative): $SIVE (light source) -> $POET (optical interposers) -> $MRVL (Likely Celestial Captive) -> $MSFT Maia + $AMZN Trainium. $SIVE (light source) -> Ayar -> AiChip -> $AMZN Inferentia/Trainium $SIVE (light source) -> Enablence -> O-Net -> ? Asia Hyperscalers _ Ongoing: $SIVE (light source) -> Ayar -> GUC -> ? (Google $TPU) $SIVE (light source) -> Ayar (TeraPHY/SuperNova)-> Wiwynn (captive CPO) -> ? ( $MSFT, $META historically Wiwynn's largest clients). Because of captive models like $MRVL Celestial, they get a free ride. However, they do compete multi-source ELS against Lumentum, Coherent, and $MTSI with Ayar and win anyway in merchant models. But they win either way. For high-volume production ramp up, a large part of it depends on the ongoing Win semi qualification, but this will likely be a large indicator. Again supply chain BOM is extremely confidential. $AMZN will never tell anyone "Hey, we use $SIVE ". But if you put 1+1+1+1+1 together, you can piece together the likely suppliers. Most people see "Poet Starlight" uses $SIVE. Or Ayar uses $SIVE. But don't map all the multi-hop relations to see where they end up. I do think $SIVE is an extremely undiscovered opportunity as the next possible mini $LITE for Silicon Photonics at $200m MC. As they're the likely upstream laser supplier for hyperscaler supply chains for future CPO/Silicon Photonics scale up with cw dfb lasers and scale out with laser arrays.

English

432

308.5K

Keşfet

@KawzInvests @siflower @aleabitoreddit @elonmusk @BarackObama @taylorswift13 @cristiano @BillGates