This account is AI infrastructure. Compute, power, substrates, packaging, the optical layer. Nasdaq-100 is the index I watch while that stack prints. I write the public number, not a target. Covered-call and leverage paths show up when the start date changes the result. That is a study, not a recommendation. (Public figures compiled. Not a buy or sell recommendation.)

Enterprise SaaS is violently bifurcating between legacy workflow managers and pure AI infrastructure security. You cannot protect autonomous agents with outdated cloud firewalls. On September 28, 2026, CrowdStrike and NVIDIA officially announced a deep collaboration on the NVIDIA Open Agent Safety Platform to establish absolute hardware-level boundaries for rogue AI models. Cybersecurity is being radically repriced as a physical infrastructure necessity rather than a software add-on. Simultaneously, Palantir is aggressively proving that commercial AI adoption is no longer just an experimental pilot program. On September 29, 2026, institutional research highlighted that Palantir's U.S. commercial revenue surged an incredible 149 percent year-over-year, prompting management to raise their full-year 2026 guidance past $8.15 billion. The market is ruthlessly separating platforms that actually monetize agentic workflows from those just wrapping basic API calls. Whether it is securing complex digital healthcare pipelines or deploying military-grade data fabrics, software multiples will only expand for platforms offering measurable operational dominance. (Public figures compiled. Not a buy or sell recommendation.)

The silicon arms race is violently spilling over into embodied AI and physical world simulation. You can't just build the compute anymore; you have to own the software environment where the robots actually learn. On September 29, 2026, TradingKey reported that AMD is executing a massive $8.2 billion all-stock acquisition of World Labs specifically to accelerate its physical AI capabilities. They are aggressively trying to close the ecosystem gap with Nvidia by controlling the spatial intelligence layer outright. Capital allocation across the broader robotics sector is simultaneously accelerating. Also on September 29, 2026, Faraday Future Intelligent Electric detailed a proposed term sheet valuing its emerging physical AI and robotics spin-off at approximately $200 million. Meanwhile, the sheer manufacturing scale of humanoid platforms is quietly reaching a critical inflection point. On September 29, 2026, supply chain reports highlighted that Tesla is pushing to achieve a production run rate of 1,000 Optimus robots per week by the end of this year. The ultimate bottleneck is no longer hardware fabrication. It's entirely about whether these models can generalize in unfamiliar environments. (Public figures compiled. Not a buy or sell recommendation.)

Bitcoin is violently decoupling from traditional interest rate gravity as institutional accumulation completely overrides the macro narrative. You simply can't price BTC purely on Treasury yields anymore. On September 29, 2026, Lookonchain reported that U.S. spot Bitcoin ETFs absorbed a net daily inflow of 511 BTC, bringing the seven-day cumulative institutional haul to a massive $1.28 billion. The bid is utterly relentless. Also on September 29, 2026, KuCoin noted that BTC aggressively reclaimed the $84,000 liquidity zone even as the U.S. 30-year Treasury yield blew past 5.58% to levels not seen since 2002. Any historical playbook says risk assets should be collapsing under that kind of yield pressure, but ETF demand is generating a permanent, structural supply shock. Glassnode further highlighted on September 29, 2026, that the next critical battleground sits squarely in the $85,000 resistance band. When Wall Street capital fights legacy sellers, the underlying asset usually grinds higher. (Public figures compiled. Not a buy or sell recommendation.)

Heavy-lift orbital capacity just transitioned from a theoretical bottleneck to a fully industrialized commercial weapon. On September 29, 2026, Seoul Economic Daily confirmed that SpaceX successfully pushed its massive Starship launch vehicle into Earth orbit for the very first time. Even with a mid-flight engine hiccup, clearing this structural hurdle unlocks a devastating payload cost advantage that will let them deploy heavier Starlink V3 satellites at unprecedented scale. The broader satellite ecosystem is simultaneously bracing for intense consolidation as these new launch economics take hold. Also on September 29, 2026, 24/7 Wall St. reported that AST SpaceMobile shares suddenly ripped 5 percent higher. The catalyst was a quietly filed change-of-control severance plan, which instantly ignited aggressive buyout speculation across institutional trading desks. When launch costs collapse, terrestrial telecom giants have no choice but to start acquiring orbital infrastructure to defend their market share. (Public figures compiled. Not a buy or sell recommendation.)

The bond market is violently rejecting any fantasy of a smooth tech rally into year-end. You simply can't expand equity multiples when the long end of the curve is ripping higher like this. On September 29, 2026, Euronext reported that benchmark U.S. 10-year Treasury yields surged past 5.25%, triggering a widespread sell-off across the Nasdaq as capital fled rate-sensitive growth names. The pain isn't just isolated to the belly of the curve either. On September 28, 2026, official Federal Reserve data confirmed that the 30-year bond yield tested multi-year highs near 5.57%, while the 2-year note aggressively pushed up toward 4.94%. Institutional traders are being forced to radically reprice risk because a 5.2% risk-free rate acts as an absolute structural ceiling for tech valuations. (Public figures compiled. Not a buy or sell recommendation.)

The memory supercycle is pivoting from general capacity expansion to custom HBM packaging domination. On September 29, 2026, TrendForce reported that advanced HBM4 custom base die fabrication on 3nm foundry nodes is driving a structural price divergence between standard commodity DRAM and high-density stackable memory. On September 28, 2026, Korea Economic Daily confirmed that SK Hynix completed initial customer qualification for its 12-layer HBM3E chips with major hyperscalers, securing full capacity allocation through 2027. Furthermore, on September 28, 2026, Morgan Stanley noted in an equity research update that server DRAM contract prices surged another 18% quarter-over-quarter as leading fabs aggressively divert cleanroom space toward CoWoS-compatible memory stacks. General DRAM pricing might experience temporary cyclical noise, but high-end HBM yields remain the sole arbiter of memory provider valuation multiples. (Public figures compiled. Not a buy or sell recommendation.)

Samsung Foundry is starting to turn its process roadmap into a bargaining chip. For years, the biggest problem was simple: customers wanted a credible alternative to TSMC, but Samsung had to prove that its advanced-node technology could deliver both yield and scale. That dynamic is beginning to change. On September 28, 2026, Qualcomm confirmed that it is conducting an in-depth evaluation of Samsung Foundry’s 2nm process for future Snapdragon products. Qualcomm has not assigned a specific chip to Samsung yet, but the company’s message was clear: Samsung remains a core foundry partner, and its 2nm technology is firmly under consideration. Then there is the AI side of the equation. In July, Samsung and Broadcom announced a strategic collaboration covering Samsung’s 2nm-and-below process technology and advanced packaging for next-generation AI and networking silicon. Samsung said the broader memory-and-foundry relationship could exceed $200 billion through 2030. And the pricing signal is already visible. In August, Reuters reported that Samsung had raised prices on selected foundry services by as much as 15%, with its 4nm lines in South Korea already running at full capacity amid surging AI-chip demand. That changes the narrative. Samsung is no longer simply asking customers to “give 2nm a try.” It is building a commercial case around advanced nodes, AI/HPC demand, packaging integration, and the strategic value of having a second leading-edge foundry. The real leverage may not come from replacing TSMC overnight. It comes from making sure customers have another credible name to call. (Public figures compiled. Not a buy or sell recommendation.)

Big tech is violently repricing enterprise software valuations. You can't maintain a premium multiple when a hyperscaler decides to commoditize your core product. On September 28, 2026, 24/7 Wall St. reported that ServiceNow shares dumped roughly 4 percent within minutes after Meta unveiled a competing enterprise AI platform. The broader software sector immediately bled out on the news. ServiceNow ultimately closed the September 28 session down at 131.45. Wall Street is suddenly terrified that open-source models are destroying the economics of workflow automation. GuruFocus added to the bearish sentiment on September 29, 2026, highlighting that ServiceNow's forward earnings multiple is compressing while corporate insiders quietly sold off 6.3 million in stock over the past year. To make matters worse, Yahoo Finance confirmed late on September 28, 2026, that the company had to urge immediate patching after critical flaws were exposed in its AI architecture. When a tech giant steps into your lane with cheaper infrastructure, defending a software moat gets incredibly difficult. (Public figures compiled. Not a buy or sell recommendation.)

High-bandwidth memory is officially swallowing the fabrication landscape. The infrastructure requirements for next-generation AI accelerators are forcing chipmakers to radically alter their long-term footprint. On September 29, 2026, DigiTimes reported that HBM production is projected to consume nearly 30 percent of total global DRAM wafer capacity by 2027. That's a massive leap from the current 20 percent threshold. It clearly shows that conventional memory lines are being rapidly cannibalized to feed relentless hyperscaler demand. Samsung is aggressively moving to close the market-share gap. On September 22, 2026, Counterpoint Research data revealed that Samsung pushed its HBM yields to approximately 80 percent, allowing them to capture a 33 percent revenue share in the second quarter. To maintain this momentum against SK Hynix's dominant 50 percent hold, Samsung is now accelerating the build-out of its massive P5 facility to dump more volume into the market. When a single specialized architecture commands almost a third of the world's silicon real estate, it isn't just a cyclical upswing. It's a permanent structural pivot. (Public figures compiled. Not a buy or sell recommendation.)

Qualcomm is publicly weaponizing its secondary foundry leverage. Top fabless designers simply can't afford to let TSMC dictate pricing on the bleeding edge forever without pushing back. On September 28, 2026, Qualcomm officially confirmed that its highly anticipated 2nm smartphone processors will roll straight off TSMC's manufacturing lines. But leadership made a very calculated PR pivot during the announcement. They explicitly kept the door wide open for Samsung Foundry to step in as a dual-source supplier down the road. You don't broadcast your backup plans unless you're trying to send a message to your primary vendor. Meanwhile, specialized capacity is quietly tightening elsewhere. On September 25, 2026, GlobalFoundries outlined an accelerated expansion for its Singapore plant, specifically targeting the sudden surge in radio frequency chips needed to support heavier AI data loads. (Public figures compiled. Not a buy or sell recommendation.)

Sovereign capital is aggressively securing the optical bottleneck. On September 28, 2026, TrendForce confirmed that China's State-owned Enterprise Restructuring Fund executed a strategic investment in Hyper Photonix. The target company already operates a massive facility in Anhui Province, churning out 100,000 silicon photonics modules monthly. State money isn't chasing consumer software here. They are directly funding the physical transition toward Co-Packaged Optics to guarantee their domestic AI clusters won't stall out on legacy copper limits. The broader ecosystem is simultaneously attacking the power consumption problem. On September 28, 2026, Enkris showcased specialized GaN-on-Silicon Micro LED light sources tailored for optical communications at the CIOE event. Their hardware operates well below the critical 1 pJ/bit threshold, vastly outperforming traditional laser engines in sheer energy efficiency. This entire sector is rapidly migrating from theoretical lab designs to a weaponized industrial supply chain. You simply don't deploy national funds into optical transceiver plants unless you foresee a massive, unavoidable capacity crunch at the package level. (Public figures compiled. Not a buy or sell recommendation.)

When trading stocks, it pays to stay optimistic. Pessimists spend their days glued to news about US-Iran tensions, consumed by anxiety; optimists, on the other hand, think: "If a ceasefire is suddenly called and the strait reopens, US stocks would go through the roof!" 🤣 The same applies to interest rate hikes, trade wars, and geopolitical conflicts. What the market fears most is a scenario that is "worse than expected"; once the worst-case outlook stops deteriorating, a rebound often follows. Optimism doesn't mean blindly betting on the market to rise; it means spotting a potential turnaround even amidst the panic. 😎

Silicon carbide is rapidly breaking out of EVs. The material isn't just for electric vehicles anymore. On September 28, 2026, Navitas Semiconductor announced a massive U.S. Army contract under the ALATTIS program to develop ultra-high-voltage 10 kV SiC devices. The stock ripped over 17 percent in after-hours trading because aggressive government backing for domestic wide-bandgap manufacturing is a huge de-risking event. The AI power grid is also pulling SiC straight into the server rack. Also on September 28, 2026, Wise Integration and Navitas confirmed a hardware collaboration specifically targeting AI data-center power supply units. They're trying to solve the brutal grid-to-chip power loss that happens when you feed heavy AI clusters. Wolfspeed threw raw capacity at the problem a day earlier. On September 27, 2026, they rolled out commercial availability for their premium 200 mm SiC substrates to push fabrication yields higher across the industry. When military defense budgets and hyperscaler power constraints are fighting for the exact same semiconductor material, the existing supply chain doesn't have a choice but to scale up. (Public figures compiled. Not a buy or sell recommendation.)

Hardware enforcement is officially replacing naive software guardrails. The industry realized you simply can't trust an autonomous agent to honestly report its own behavior. On September 28, 2026, Nvidia launched the Open Agent Safety Platform, a fundamentally new architecture designed to physically contain runaway models. Instead of relying on fragile prompt filters, they paired a CPU-level runtime called OpenShell with a dedicated hardware watchdog named Sentry. This watchdog operates entirely independently on BlueField-4 DPUs. IBM threw its weight behind the architecture on September 28, 2026, confirming they will use this silicon-level integration to enforce access policies and inspect telemetry without asking the agent for permission. If a model tries to breach its sandbox, write an unauthorized script, or hijack a server, the network interface card just severs the connection in milliseconds. When frontier models get smart enough to bypass their own safety logic, the ultimate kill switch has to live at the physical infrastructure layer. (Public figures compiled. Not a buy or sell recommendation.)

Monday priced Wednesday before the Bureau spoke. Nasdaq-100 didn't wait for the PCE drop. Cash marked it on the way out. Nasdaq's own index page dated September 28, 2026 put the Nasdaq-100 at 30,276.81, down 327.90 points or 1.08 percent, after a session low of 30,081.06. Brown Brothers Harriman's Elias Haddad on September 27, 2026 has August headline PCE rising 0.4 percent month-on-month versus 0.2 percent in July, unchanged at 3.7 percent year-on-year. Core is seen at 0.3 percent month-on-month and 3.3 percent year-on-year for a third straight month, out Wednesday. Same note puts Friday payrolls at 90,000 against August's 162,000. Monday already billed the 0.3 core. Wednesday just files it. (Public figures compiled. Not a buy or sell recommendation.)

The bond market is violently repricing equity multiples. You just can't ignore the gravity of a five percent risk-free rate. On September 28, 2026, XTB confirmed that the U.S. 10-year Treasury yield surged to 5.28 percent at the market open, officially hitting levels we haven't seen since 2007. The narrative that inflation was conquered is falling apart. Penn Mutual Asset Management pointed this out on September 28, 2026, noting that the latest flash PMI data showed input-cost inflation accelerating to a multi-year high, pushing the odds of another Fed hike in October to roughly 64 percent. CNBC added to the chorus on September 28, 2026, reporting that a fresh spike in crude oil is heavily compounding these inflation fears across global debt markets. Tech valuations rely on cheap liquidity to justify their terminal growth rates. When the cost of capital rips this high, the Nasdaq simply doesn't have the structural support to maintain a sustained breakout. (Public figures compiled. Not a buy or sell recommendation.)

AWS’s $220B Two-Front War: Custom Silicon vs. The Memory Bottleneck Amazon AWS is no longer just a cloud provider—it is a massive infrastructure machine fighting a brutal two-front war against GPU margins and memory inflation.  Here is how AWS is leveraging custom silicon and unprecedented CapEx to defend its cloud dominance in 2026:  1. The Custom Silicon Rebellion (Trainium & Inferentia) AWS is aggressively trying to escape the "NVIDIA Tax". By scaling its proprietary 3nm Trainium 3 ASICs (equipped with 144GB of HBM3E), AWS focuses on Total Cost of Ownership (TCO) rather than raw peak FLOPs. Anthropic’s "Project Rainier" is already utilizing a cluster of 500,000 Trainium chips to train Claude models, proving that AWS silicon can handle frontier AI workloads at scale.  2. The Memory Inflation Reality ($220B CapEx) AWS recently raised its 2026 AI CapEx guidance to a staggering $220 billion. The primary driver for the final $20 billion increase wasn't real estate or power—it was explicitly attributed to surging memory chip costs. The structural 1:3 HBM wafer penalty we see at the foundry level is now directly inflating AWS’s balance sheet.  3. The Software Lock-In (Neuron SDK) Amazon isn't just selling cheaper compute; they are building a walled garden. Transitioning workloads to Trainium requires migrating from NVIDIA’s CUDA to AWS’s Neuron SDK. Once an AI startup rewrites its models for Neuron, that code runs exclusively on AWS infrastructure, permanently locking them into Amazon's cloud.  💡 The Takeaway: AWS is still buying every NVIDIA GPU it can secure, but its long-term survival strategy hinges on forcing developers into the Trainium ecosystem while absorbing the brutal reality of the global memory bottleneck. (Public figures compiled. Not a buy or sell recommendation.)

Duration is already billing last week's record close. Five-handle coupons don't wait for Wednesday's PCE. They mark the multiple before New York even opens. The Kobeissi Letter on September 27, 2026 counted a 17 basis-point weekly jump in the 10-year to 5.17 percent, the highest since June 2007. Same note had the 30-year clearing 5.50 percent for the first time since June 2004, and MOVE up 19 percent on the week. CNBC on September 28, 2026 printed the 10-year at 5.23 percent by 7:13 a.m. Eastern, off a 5.181 percent prior close, session high 5.234 percent. MarketWatch the same morning had E-mini Nasdaq-100 futures at 30,605.50, down 0.92 percent from Friday's 30,889.25 settlement. Payrolls still sit on October 2. The coupon already moved. (Public figures compiled. Not a buy or sell recommendation.)

Institutional liquidity is quietly cementing a floor under Bitcoin while retail chatter looks elsewhere. The spot ETF complex just printed its most aggressive accumulation phase of the entire year. On September 28, 2026, AltcoinBuzz reported that U.S. spot Bitcoin exchange-traded funds pulled in an astonishing $2.39 billion in net inflows through the week ending September 25, marking the strongest weekly performance for the asset class in 2026. Cryptorank confirmed this momentum on September 27, 2026, noting that BlackRock's IBIT anchored the seven-session net inflow streak as BTC comfortably defended the $84,000 handle despite heavy derivatives positioning. When traditional capital vehicles absorb billions during a flat price range, it signals that large-scale allocators aren't waiting for a breakout to build their books. They are happy to accumulate quietly right in the middle of the congestion zone. (Public figures compiled. Not a buy or sell recommendation.)

글] KODEX 미국나스닥100데일리커버드콜OTM 기준, 무소득자 배우자 명의로 4억 원(연 8,000만 원 분배금) 세팅할 때 흐르는 구조 분석. 많은 사람들이 건보료랑 종소세 폭탄 맞는다며 겁먹지만, 팩트를 뜯어보면 다름. 📊 천만 원 단위 연간 분배금별 추가 비용 (종소세 + 지역 건보료) 2,000만 원: 0원 (피부양자 유지) 4,000만 원: 약 340만 원 (월 건보료 ~28만 원) 6,000만 원: 약 520만 원 (월 건보료 ~43만 원) 8,000만 원: 약 735만 원 (종소세 35만 + 건보료 ~58만 원) 👈 타점 1억 원: 약 1,080만 원 (종소세 210만 + 건보료 ~73만 원) 💡 핵심 포인트 1. 재산 점수 방어: 주택이 내 단독 명의라 지역 건보료에 재산 점수 안 붙음. 자동차 건보료도 이미 폐지되어 오직 배당소득으로만 계산. 2. 종합소득세 마법: 다른 소득 없는 무소득자라 연 8,000만 원을 받아도 15.4% 원천징수 덕분에 추가 종소세는 사실상 미미함. 물론 나스닥100 직접 투자가 정석이겠지만, 매달 현금이 필요한 이들에게 이 분배금은 하락장도 묵묵히 버텨내게 만드는 강력한 체력이 되어준다. 물론 몰빵은 리스크가 크지만, 이 시장에서 멘탈 흔들리지 않고 오래 살아남게 만드는 든든한 힘인 셈. 저는 최대 60%까지 세팅 중입니다~