$BRICKY 🧱

$NVDA giving away the orchestration layer will make it that much harder for $PATH and $NOW to prove itself. I am invested in $PATH but that weak guide + this potential catalyst means management has to execute flawlessly this year

@highnstein_ Sounds about right from what I hear from others Most people invest in things by staring at balance sheet numbers without ever truly understanding the product.

@bricked_trades Used to work with UIPath and a few other RPAs on a process automation team. We pretty quickly figured out almost anything an RPA could do, python or other apps could do better and easier. Our spend on RPAs collapsed to next to nothing after 3 years

$PATH the biggest value trap shilled here on X Wall Street will always discount a perfect earnings if they fundamentally stop believing in the company's terminal value. No different than $TTD $ADBE where both stocks suffer from "Middleman Moat Crisis." RPA bots are mindless execution engines that require humans to map out every single edge case. AI agents can reason. $NVDA NemoClaw once announced could be final nail for $PATH

Why $ASPI Solves the Heat Bottleneck Photonics Can’t As AI compute moves toward 1nm, it has hit a "thermal wall." Transistors now generate heat densities exceeding nuclear reactor cores, threatening to kill performance scaling. While the market chases Gallium Nitride (GaN) and Photonics, ASP Isotopes ($ASPI) is quietly addressing the problem at the atomic level. Why GaN and Photonics Aren't Enough GaN (Power): Excellent for power efficiency, but its bulk thermal conductivity is mediocre (130–230 W/m·K). It reduces energy loss before it reaches the chip, but cannot effectively move heat out of a 1nm logic core. Photonics (Interconnect): Essential for moving data with light, reducing interconnect heat. However, the logic computation itself still boils. Because Co-Packaged Optics (CPO) place sensitive lasers directly next to the logic die, core heat can "detune" the optics, causing data errors. How $ASPI Is Disrupting with Silicon-28 Current chips use Natural Silicon, a "messy" isotopic mix. The 8% of atoms that aren't Silicon-28 (Si-29 and Si-30) act as atomic speed bumps, trapping heat particles (phonons) and causing the chip to throttle. $ASPI is the only commercial player scaling Silane-28 gas. By building chips with isotopically pure Si-28, the "atomic highway" is cleared. At the nanosheet level, Si-28 boosts thermal conductivity by 150%. As of March 2026, $ASPI has secured its largest Si-28 contract to date and is ramping Helium production via Renergen, the critical cooling gas required for the lithography chambers that build these chips.The Thermal Escape Hatch: Why $ASPI Wins the 1nm Era Iran War Affect The closure of the Strait of Hormuz following Iranian-US-Israeli conflict has paralyzed the global helium market. Qatar accounts for 30% of global helium supply. With QatarEnergy’s Ras Laffan facility offline and the Strait blocked, one-third of the world's helium is effectively "trapped." Why this Matters: Helium has no substitute in semiconductor manufacturing. It is essential for cooling wafers and is the lifeblood of the EUV lithography chambers used by TSML, Samsung, and Intel to print 1nm circuits. With acquisition of Renergen, $ASPI has secured the most strategic helium asset outside of the conflict zone.

Tons of quantum focused sessions for $NVDA GTC next week to look forward to. $QBTS rounded bottom with a catalyst

**Only the 17th time we've seen $VIX close down 25% from its intraday high**

The amount of fake/misleading market headlines been unreal. This year we got: - Misleading reports of rare earth price floors gone for $MP in january - $NVDA no longer wanting to invest in OpenAI - OpenAI not satisfied with $NVDA chips - Iran peace talks last week - The cuck @SylentTrade saying Iren has deal with Anthropic

Ignore Overcrowded Photonics and Memory Companies Shilled by X: Why Amkor $AMKR and ASE $ASX are the Ultimate Stealth Plays Nvidia's GTC conference is scheduled for March 16, and Jensen Huang is already teasing that he will unveil a chip that will "surprise the world." The market is buzzing with anticipation that this event will highlight a massive expansion of silicon photonics and Co-Packaged Optics (CPO) technology to resolve the severe data transmission bottlenecks plaguing AI data centers. Investors are actively positioning themselves for this exact moment. They are piling into overcrowded and obvious bottleneck names like $MU and Western Digital $WDC $SNDK to play the memory shortage, or chasing component makers like Coherent $COHR and Lumentum $LITE for the optical networking hype. However, buying into these crowded trades ignores the fundamental physics of how the next generation of AI hardware is actually built. The true picks and shovels of this upcoming cycle are the Advanced Packaging OSATs like Amkor Technology $AMKR and ASE Technology $ASX. They sit right at the center of the supply chain, acting as the undisputed gatekeepers for both the memory squeeze and the photonics revolution. The Raw Math of the Current Memory Crisis To understand why advanced packaging is so critical, we have to look at the staggering memory demands of new AI chips. Nvidia's most recent Vera Rubin chip requires an unprecedented 288GB of RAM. This is an 800% increase over the memory found in a high-end PC and 2,300% more than a premium smartphone. By comparison, the H100 launched just four years ago needed only 80GB of RAM, which is 72% less than Rubin. Each new generation of Nvidia AI chips requires significantly more memory than the last. With AI giants like Alphabet and OpenAI locking up large portions of the global memory chip supply by purchasing millions of these Nvidia chips, the global memory shortage is officially out of control. Average spot prices for 16GB DDR4 RAM are up 2,352% year over year to a record $76.90, while 8GB DDR4 prices are up 1,873% year over year to an all-time high of $28.90. You cannot simply glue 288GB of RAM to a logic board and expect it to work. Fusing that much memory directly to a processor requires elite 2.5D and 3D advanced packaging to break the memory wall. This is exactly where Tier 1 OSATs step in to perform the microscopic integration that makes these chips functional. The Photonics Pivot As AI compute demand surges, the industry is accelerating a massive shift away from the conventional "optical to electrical to optical" transmission architecture toward a purely "optical to optical" structure. TSMC management has confirmed that CPO has already entered the proof-of-concept and sampling stage. The market is raising the possibility that the Nvidia Rubin Ultra generation could mark the inflection point where traditional copper interconnects finally give way to optical interconnects. Industry consensus projects that 800G to 1.6T silicon photonics modules will ramp meaningfully in 2025, with CPO and 3.2T optical engines accelerating rapidly from 2026 onward. Taiwan's supply chain is moving aggressively to capture this value. Notably, TSMC and ASE have formed a silicon photonics industry alliance specifically to target the high-speed data center interconnect market. Why OSATs are the True Choke Point Analysts note that the investment focus within the AI industry has been rotating from assembly to cooling to components and materials, and it is now firmly landing on optical networking. A trend-following approach is key here. While names like Browave, LuxNet, and APAT Optoelectronics are scaling up mass production of silicon photonics modules, they carry the risks of small market capitalizations, high share price volatility, and brutal price competition from subsidized Chinese players. The safer, higher-probability bet lies with the companies doing the actual packaging. TSMC simply cannot handle this level of advanced packaging demand alone. They are increasingly outsourcing the complex assembly to trusted OSATs. Amkor, for instance, is perfectly positioned to capture this demand overflow. Their massive advanced packaging facility in Peoria, Arizona, belongs entirely to Amkor and serves as a vital domestic hub for sovereign AI production. When Broadcom designs a custom TPU for Google or Nvidia moves to optical interconnects, they rely on Amkor and ASE to align the sub-micron lasers and ensure the final product actually yields. The Long Wave Cycle We are currently looking at a bull market that may be linked to a long-wave technology cycle analogous to the Kondratiev Wave. The AI-driven industrial revolution could very well underpin a 30 to 50 year secular economic expansion. Against this backdrop, the market is entering a phase of heightened volatility within a structurally bullish trend. Flexible portfolio management and disciplined risk control are essential. Overpaying for memory makers or chasing volatile optical networking names might work for short term trades, but the companies maximizing the efficiency of every chip are the ones building the physical infrastructure. Amkor and ASE are the critical tailwind winners that will turn Jensen Huang's shocking new chips from laboratory concepts into data center realities.

How to trade confirmation of aliens Long $AMTM Market seems to be catching on OVN session In early 2026, Tuttle Capital launched the UFO Disclosure ETF (ticker: $UFOD). This fund specifically bets on companies that could profit if the government confirms the existence of non-human intelligence. $AMTM is one of its largest holdings. $AMTM allegedly has been involved for decades in top-secret programs to develop or reverse engineer Unidentified Anomalous Phenomena (“UAP”) the company has been trusted with many of the nation’s most important core technological secrets of the last 80 years beginning with its work on the Manhattan project. The restricted data in the company’s archives is reportedly so highly classified that even sitting United States Presidents have been denied access in the past. the company’s undisclosed 70-year history at the Area 51 and Nevada Test Site complex, a highly classified military facility in the Nevada Desert, that it has operated for decades and used as a staging area for experimental “black projects”, nuclear weapons testing, reverse engineering of captured Russian planes, and its contributions to some of the most important top-secret aircraft created in the entire history of the United States

$GDS 50 C 03/20/2026 Prem: $664,288 Nutty activity on $GDS They are one of the largest datacenter operators in china Rumors DeepSeek V4 may drop over the weekend (before Lunar New Year) When DeepSeek released last January, $GDS ran 140% within a month

Hope yall followed $FSLY

$FSLY just proved the Software dead narrative is true for “cheap-applications” and not for usage and edge compute like $NET $DDOG $FROG They had a 55% RPO jump means customers are terrified of running out of edge capacity. They are pre-booking $FSLY "pipes" for 2026/2027. This confirms more AI agents = more usage = more profit. As AI agents (Bots) flood the web, Bot Management becomes the most critical piece of software a company owns. $FSLY monetizing this aggressively validates that they are a "Pick and Shovel" for the Bad Agents too (by stopping them). Acceleration to 23% revenue growth while trading at <5x Sales is a dirt cheap compared to $NET (20x+) Long $FSLY even w a 40% gap up.

$ACLS is the "High-Energy Refueler" for the AI Memory Squeeze If you missed $SNDK $MU its time to look at their suppliers Samsung, SK Hynix, $MU are one of $ACLS biggest customers Thesis is simple Next-gen memory (HBM4) requires specialized ion implanters to make these chips work. Axcelis Technologies $ACLS are the "Molecular Surgeons" of the memory chips. Their Purion machines use high-energy beams to "tune" the silicon with surgical precision. They are the only major player besides $AMAT capable of handling the high-energy needs for HBM4. While other equipment names are at all-time-highs, $ACLS has been trading relatively flat during a temporary slowdown. Recently: * They launched the Purion H6 on Feb 4, 2026, specifically to solve the HBM4 stacking problem. * Big memory makers are gearing up for mass production in early 2026. * The Value: Trading at ~20x P/E—a big discount to peers like $LRCX (~32x)—with a massive $590M cash pile. As of late 2025, memory applications made up a relatively small portion of shipped system revenue (~2%), but bookings for memory are now rapidly improving. Management projects that increased investments in DRAM and HBM (High Bandwidth Memory) will be the main revenue driver starting in early 2026. This would be a major surpriser for earnings next week

Strange day today Emerging market and international stocks are ripping ATH $DFAE $AAXJ $EEM $XCEM $CGIC $SPEM

$FROG is the Toll Booth of the AI Agent Economy That fat $IGV selloff last week from Claude Code, OpenClawd / Moltbot, and OpenAI Codex got everyone mindlessly buying anything Software company but I think this is the time to pick the real winners of the new AI Agent economy. New agentic coding tools can do 1,000 commits per hour. Compare that to a human developer who might do 10 a day. SemiAnalysis reports that AI already writes 4% of all code on GitHub and predicts that will hit 20%+ by 2026. Basically companies could perform 100x the workload with a fraction of the headcount. Even non technical people like Sales and Marketing I see are building tools themselves with Claud Code. The Problem: Humans Are Too Slow Current storage tools like GitHub act like filing cabinets great for humans, but they overflow when AI dumps 1,000 files an hour. The bottleneck isn't writing code anymore; it’s storing and checking it to make sure the AI didn't hallucinate a security flaw. I found $FROG to be a direct winner of this new reality. They are the "Digital Customs Office". This is where $FROG steps in. It’s not just storage; it’s an automated vault. Every time an AI agent tries to ship a "binary" (a finished piece of software), $FROG security tool, Xray, scans it instantly. If the code is bad, JFrog blocks it. It’s the only way to manage the flood of "industrialized" software without breaking the company. Their financials back this up. Unlike GitHub and other SaaS models like $ADBE $ASAN $MNDY $CRM $HUBS and more charges per "seat" (human), JFrog gets paid more as AI generates more data. Their cloud revenue grew 50% year-over-year, and they flex 84% gross margins, proving they can scale this "AI flood" profitably. Broader market research supports this "flight to quality." Goldman Sachs said there is a thematic shift in the market where smart money is rotating out of generic software and into "mission-critical infrastructure" supporting the AI agent economy. If you are looking for quality Software names to dip buy, avoid seat-based billing and look at symmetric workload/usage based billed SaaS companies.

$HPQ - Traded multiple inside sessions holding steady while rest of tech sold. Market punishing them with higher memory prices affecting margins on device sales but they will report for sure the reality of the effect this month. Could very well Vshape hard if the selling was FUD

$VKTX - With $NVO and $HIMS both getting the heat from the FDA, $VKTX looking for a momentum trade higher at this local bottom at $26 with earnings this week. Ph3 subq trial update as well as oral drug developements can see a reversal

$NBIS - Respected this IHS within the MSFT deal gap news level. $NBIS reports this week and well positioned to capture that huge big tech Capex spend

$FROG - #1 beneficiary of the AI agent trade no one on FinX has realized (thank god). Earnings coming up this week trading lower while the AI agentic workload + commits have accelerated parabolically last month. I took some lottos at this 50.00 psyche. x.com/bricked_trades…

Took large stake in $FROG 1 dev with 100 agents = 100x the consumption vs traditional SaaS seat billing like $ADBE $FROG Artifactory is the only "Vault" designed for a CTO to audit the code base of thousands of lines of code from hundreds of agents