Research Coverage
AI-Era Stock Universe
Eight distinct plays on the AI revolution — from chip design software to foundry manufacturing, etch equipment, and silicon photonics. Click any card for the full research deep dive.
Role
The Architect
Cadence Design Systems
NASDAQ · EDA / Semiconductor Software
A high-quality, stable growth investment that is a core holding in the semiconductor value chain. As a market leader in the EDA duopoly, it benefits from the re...
Revenue
$4.64B
Growth
+13.5%
Mkt Cap
$81.1B
Role
The Dominant Architect
Synopsys, Inc.
NASDAQ · EDA / Semiconductor Software
A value-conscious opportunity to buy a wide-moat industry leader at a discount. The stock has underperformed recently due to concerns over its large Ansys acqui...
Revenue
$7.05B
Growth
+15.1%
Mkt Cap
$85.9B
Role
The Builder
Celestica Inc.
NYSE · Electronics Manufacturing Services
A 'picks and shovels' momentum play on the explosive build-out of AI data center infrastructure. As a key manufacturer of high-performance servers, storage, and...
Revenue
~$13.9B
Growth
+28%
Mkt Cap
~$10B
Role
The AI Engine
NVIDIA Corporation
NASDAQ · AI Semiconductors / GPU Computing
NVIDIA is the dominant supplier of the GPUs and accelerated computing platforms that power virtually all AI training and inference workloads. Its CUDA software ...
Revenue
$215.9B
Growth
+65%
Mkt Cap
~$4.4T
Role
The Enterprise OS
ServiceNow, Inc.
NYSE · Enterprise Workflow Automation
A dominant force in the enterprise software market, providing a comprehensive cloud platform to automate IT, business, and customer workflows. With strong, cons...
Revenue
$13.3B
Growth
+21%
Mkt Cap
~$200B
Role
The Manufacturer
Taiwan Semiconductor Manufacturing
NYSE · Semiconductor Foundry
TSMC is the irreplaceable manufacturer of the world's most advanced chips. It holds 71%+ of the global foundry market and a near-monopoly on sub-5nm production ...
Revenue
~$122B
Growth
+39%
Mkt Cap
~$950B
Role
The Etch Master
Lam Research Corporation
NASDAQ · Semiconductor Equipment
Lam Research dominates etch and is #2 in deposition — the two most critical steps in manufacturing advanced DRAM, NAND, and HBM. Every HBM chip inside an NVIDIA...
Revenue
$20.56B
Growth
+23.7%
Mkt Cap
~$290B
Role
The SiPho Foundry
Tower Semiconductor Ltd.
NASDAQ · Semiconductor Foundry
Tower Semiconductor is the open-foundry infrastructure layer underneath the entire optical interconnect stack. Their PH18 silicon photonics platform doubled rev...
Revenue
$1.76B
Growth
+12.8%
Mkt Cap
~$14.6B CAD
Role
The EUV Monopolist
ASML Holding N.V.
NASDAQ · Semiconductor Equipment
ASML holds a complete monopoly on Extreme Ultraviolet (EUV) lithography — the technology that physically cannot be replicated by any competitor. Every leading-e...
Revenue
€28.3B
Growth
+14%
Mkt Cap
~$520B
Market Intelligence
Essential Reading
Curated research and analysis that informs our investment thesis — the ideas worth sitting with.
The 2028 Global Intelligence Crisis
"What if our AI bullishness continues to be right... and what if that's actually bearish?"
A fictional-but-rigorously-researched macro memo written from June 2028, looking back at a financial crisis triggered by AI's overwhelming success. The thesis: AI exceeds all expectations, corporate margins explode, the S&P hits 8,000 by October 2026 — then the economy collapses anyway. Machines don't spend money. White-collar workers lose jobs and stop consuming. The entire consumer economy quietly implodes in what the authors call the "Human Intelligence Displacement Spiral."
Relevance to This Watchlist
CRDO: The AI Data Center Connectivity Play
"The biggest bottleneck in AI isn't chips anymore — it's the wiring between them."
Credo Technology (CRDO) makes high-speed Active Electrical Cables (AEC) and optical DSP chips that move data between GPUs, switches, and servers inside AI data centers. As NVDA's Blackwell GPU clusters scale to hundreds of thousands of chips, the bandwidth connecting them becomes the critical constraint — and Credo's HiWire AEC is gaining share as a lower-power, lower-cost alternative to traditional optical fiber at short distances.
Revenue Growth (YoY)
+272%
Gross Margin
66.7%
Forward P/E
~25x
Analyst Target
~$185
Key Risk
Stock is down ~47% from its 52-week high of $213.80. Microsoft is believed to be a very large portion of revenue — any slowdown in MSFT's data center spending would hit CRDO hard. Monday's earnings report is the next major catalyst.
MRVL: The Custom AI Chip Architect
"While NVIDIA dominates off-the-shelf AI compute, Marvell is quietly winning the custom silicon war — building bespoke AI chips for Google, Amazon, and Microsoft."
Marvell Technology (MRVL) is a semiconductor company pivoting aggressively toward custom AI ASICs (Application-Specific Integrated Circuits) for hyperscalers. As Google, Amazon, and Microsoft seek to reduce dependence on NVDA by building their own AI chips, they turn to Marvell to design and supply them. This is a high-margin, sticky business with multi-year design win cycles — and Marvell's AI revenue is growing explosively.
Revenue Growth (YoY)
+37%
Gross Margin
50.7%
Forward P/E
~24x
Analyst Target
~$116
What to Watch — March 5 Earnings
Analysts expect EPS ~$0.79 and Revenue ~$2.2B. The key number is AI revenue guidance — any acceleration in custom ASIC design wins from hyperscalers could re-rate the stock significantly toward the $116 analyst target. Current price $81.69 implies ~41% upside to consensus.
AVGO: The AI Infrastructure Empire
"Broadcom is the quiet monopolist of AI infrastructure — every hyperscaler building custom AI chips relies on Broadcom's networking silicon, and Google's TPU is built on Broadcom's custom ASIC expertise."
Broadcom (AVGO) operates two high-margin businesses that are both accelerating with AI: custom AI ASICs (building bespoke AI chips for Google, Meta, and ByteDance) and networking silicon (Tomahawk and Jericho switches that connect GPU clusters). On top of this, the $69B VMware acquisition has transformed Broadcom into a dominant enterprise software player, adding a recurring subscription revenue stream with 70%+ gross margins.
Revenue Growth (YoY)
+16.4%
Gross Margin
77.3%
Forward P/E
~22x
Analyst Target
~$455
What to Watch — March 4 Earnings
Analysts expect EPS ~$2.02 and Revenue ~$19.1B. The key signal is AI revenue guidance — Broadcom guided for $60–90B in AI revenue by FY2027 from just three hyperscaler customers. Any update to this range, plus VMware renewal rates, will drive the stock. Current price $319.55 implies ~42% upside to the $455 analyst consensus target.
Key Risk
Stock is down ~23% from its 52-week high of $414.61. Heavy customer concentration in AI ASIC business (Google, Meta, ByteDance) means any hyperscaler capex pullback would hit disproportionately. VMware integration execution risk also remains.
MDB: The AI Data Layer Nobody Is Talking About
MongoDB just got crushed 22–25% on earnings — but the selloff is a guidance reset, not a business breakdown. Q4 revenue grew 27% to $695M, beating estimates by 4%. Atlas (the cloud product) grew 29%. The problem: Q1 guidance missed by $0.02–$0.06 EPS, and two sales leaders are departing. The market punished a great business for a small forward miss.
What MongoDB Does
The world's dominant document database — flexible JSON-style data storage that modern apps (AI, e-commerce, IoT) need. 45% NoSQL market share. 65,200+ customers. Atlas is the cloud-hosted version, usage-based pricing.
The AI Bull Case
MongoDB's vector search capability makes it the natural database for RAG (retrieval-augmented generation) AI apps — the architecture every LLM deployment uses to query private data. AI adoption accelerates MongoDB's growth, not disrupts it.
Why It Got Crushed
Q1 EPS guidance $1.15–$1.19 vs. $1.21 estimate. Atlas growth decelerating from 29% to 21–23% for FY2027. Two top sales executives departing simultaneously. Classic "sell the guidance" overreaction — 31 of 42 analysts still rate Buy.
Q4 Revenue
$695M
+27% YoY
Q4 EPS
$1.65
Beat $1.47 est.
Atlas Growth
29%
Q4 YoY
Current Price
~$253
Down 22% post-earnings
COHR: The Most Vertically Integrated Photonics IDM
GTC Update (Mar 17, 2026): COHR spiked ~10% on the $2B NVIDIA supply deal (through 2030), then gave back gains after Jensen Huang's GTC keynote confirmed NVIDIA will continue using copper cables (ACC) in near-term Blackwell and Vera Rubin systems. Co-packaged optics (CPO) — where COHR has its deepest moat — is now a 2027–2028 revenue story, not 2026. The bull case is intact; the timeline is longer. This pullback is a better entry point than last week's spike.
Coherent Corp. is the highest-conviction name in the photonics supply chain because it owns both the laser IP and the fab — the exact combination the author of the Photonics Boom article identifies as the source of durable value creation. Coherent operates the world's largest 6-inch Indium Phosphide (InP) wafer fab, giving it a 4x manufacturing cost advantage over competitors still on 2-inch or 4-inch wafers. That fab is not replicable in under 5 years. NVIDIA's $2B investment and supply agreement through 2030 provides a durable revenue floor.
The IDM Moat: Coherent designs and manufactures its own InP lasers — the active light-emitting components inside every 800G and 1.6T transceiver. Competitors that buy lasers from third parties are at Coherent's mercy on pricing and supply. As AI cluster density increases (800G → 1.6T → 3.2T), the demand for InP lasers grows exponentially, and Coherent's 6-inch fab is the only facility in the world that can scale to meet it at competitive cost.
The Risk: COHR trades at ~251x trailing P/E — expensive by any traditional measure. Jensen Huang's GTC comments (Mar 17) confirmed CPO revenue ramp is 2027–2028, not 2026 — the market is repricing the timeline. The bull case requires continued hyperscaler capex growth and a successful 800G→1.6T→CPO transition. If AI capex slows or the CPO transition is further delayed, the multiple compresses sharply. The $2B NVIDIA deal through 2030 provides a floor. Size accordingly and be patient on the CPO thesis.
AXTI: The InP Substrate Chokepoint — The CRDO of Raw Materials
Every 800G and 1.6T transceiver that COHR, LITE, and ALAB make requires Indium Phosphide (InP) wafers as the raw substrate. AXT Inc. is one of the only companies in the world that manufactures these wafers at commercial scale. With a 70% supply gap in InP substrates — orders booked through 2026, 18–24 month equipment lead times to add capacity — AXTI sits at the most acute bottleneck in the entire photonics supply chain. This is the CRDO thesis applied one layer deeper: not the transceiver, but the raw material the transceiver is built from.
The Chokepoint Thesis: InP is not silicon. It cannot be grown in standard semiconductor fabs. It requires specialized MOCVD (metal-organic chemical vapor deposition) reactors that take 18–24 months to procure and qualify. AXT has spent 30+ years perfecting InP crystal growth. The photonics boom is creating demand that the entire industry cannot fulfill — and AXT is one of the only companies positioned to supply it. As COHR and LITE ramp their 6-inch InP fabs, they need more InP substrates. AXTI is the upstream supplier they cannot replace quickly.
The Risk: AXTI is a ~$300M market cap company with lumpy revenue and significant customer concentration. It is not profitable on a consistent basis. If the photonics buildout slows or COHR/LITE internalize more substrate production, AXTI loses its leverage. This is a speculative position — size it as 1–2% of portfolio maximum.
Data Industry Primer
An overview of one of the most fragmented industries in tech
"Storage is at the center of the modern data estate. Most of the companies in the limelight — Snowflake, MongoDB, Databricks, Azure, AWS, GCP — make this a pillar of their data strategy and then have an ecosystem around it."
A comprehensive map of the entire data industry — from the 1970 IBM relational model through Hadoop, Spark, Snowflake, and the modern cloud data lakehouse. Covers every layer of the stack: ingestion, processing, storage (data warehouse vs. lake vs. NoSQL), observability, governance, and AI/ML workloads. Essential context for understanding where MongoDB, Snowflake, Databricks, and Oracle actually compete — and why the lines between them are blurring fast.
Looking for the Next 800% Tech Stock Winner?
A peek inside the AI datacenter optical interconnect revolution
"As AI models grew exponentially, datacenters were hitting a critical bottleneck. They needed to move massive amounts of data between servers and switches, but the infrastructure couldn't keep up. That's where Credo came in — and that's where optical is going next."
McHugh maps the three-stage evolution of AI datacenter connectivity: copper (CRDO, done — 800%+ gains), optical interconnects (LITE, COHR, CIEN, ALAB, CLS, GLW — running now), and co-packaged optics / silicon photonics (POET, LWLG — speculative next frontier). His key warning: the optical plays are largely priced in after massive runs. The best risk/reward is ALAB, down 29% YTD despite 115% revenue growth. COHR is the best-valued optical name with a direct Nvidia partnership. POET and LWLG are high-risk binary bets on next-gen photonic integration.
The Photonics Boom: A Simple Guide to the Biggest Investment Theme in 2026
Yiannis Zourmpanos · YIAZOU · Mar 3, 2026
"The question is no longer whether optical interconnects will dominate data center architectures, but how quickly the ecosystem can mature."
Big Tech is committing $660–690B to AI infrastructure in 2026 — nearly double 2025. The AI cluster optics market alone hits $26B in 2026 (+60% YoY). The key insight: downstream transceiver makers (LITE at 202x P/E, COHR at 251x) are already priced in. The real opportunity is the upstream supply chain — Indium Phosphide (InP) substrate makers, MOCVD equipment, and testing companies that most investors don’t know exist. A 70% supply gap in InP with orders booked through 2026 and 18–24 month equipment lead times creates a structural bottleneck. Co-packaged optics (CPO) deployment in 2028–2030 extends the boom into the next decade.
The Soitec Series: Photonics SOI — The Engineered Wafer Monopoly Hiding in Plain Sight
"Soitec holds a near-monopoly on Photonics-SOI — the engineered silicon-on-insulator wafer that every silicon photonics chip is built on. Their two nominal competitors are not yet at commercial scale. This is the Soitec thesis in one sentence."
A 7-part deep-dive into Soitec SA (SLOIF), the French specialty materials company that makes the engineered wafers silicon photonics chips are built on. Covers the Smart Cut technology moat, the Photonics-SOI monopoly, the mobile SOI trough and recovery thesis, competitive landscape (Shin-Etsu, GlobalWafers), and the investment case at 2x EV/revenue. The author calls it his "career-defining bet" and largest position.
Market Structure · AI Data Center & Networking
The Dawn of a $73 Billion Market
Every bucket of the AI Data Center & Networking TAM by CY2030E — mapped to the publicly traded companies that own each layer, their revenue exposure, and which ones are in this watchlist.
Photonics · AI Datacenter Infrastructure
Photonics Supply Chain Map
The full stack from raw wafer to AI datacenter rack — every layer, every publicly traded company, and where the value is being created.
AXTI / SLOIF → COHR / LITE / IPGP → ALAB / CIEN / AAOI → GLW → AI Datacenter Rack
Raw InP/SOI wafers → Laser chips → Transceivers & connectivity → Fiber backbone → NVIDIA GB200 cluster
Photonics · Architecture Evolution
Light Source Architecture: Gen 1 → Gen 4
The EML shortage is accelerating the transition from pluggable optics to co-packaged silicon photonics. Understanding where each generation stands — and which companies own each stage — is the key to sizing positions correctly.
Laser and modulator are integrated on a single InP chip. Dominant architecture for 400G/800G transceivers today. The EML shortage — driven by limited InP fab capacity globally — is the defining supply constraint of 2025–2026. Lead times are 18–24 months. This shortage is the forcing function accelerating Gen 2 adoption.
Signal Path
InP substrate → MOCVD epitaxy → EML chip → pluggable transceiver module → switch port
Timeline
Dominant today through 2026–2027
Key Bottleneck
InP fab capacity — only ~5 companies globally can produce at scale
Portfolio Positioning
World's leading EML maker. 100G EML at record shipments. 200G EML ramping for 800G.
CORE BENEFICIARY6-inch InP fab — 4x cost advantage. NVIDIA GB200 EML supplier through 2030.
CORE BENEFICIARYIDM laser manufacturer. Industrial + datacenter. Deep InP manufacturing moat.
BENEFICIARYInP substrate supplier. Gen 1 raw material. Structural headwind if Gen 2/3 accelerates.
AT RISK LONG-TERMInstead of an integrated EML chip, a continuous-wave (CW) laser provides a steady light source, and a separate silicon photonics chip handles the modulation. This decouples the laser from the modulator — allowing silicon fabs (TSMC, Intel Foundry) to manufacture the modulator at scale, while the CW laser is sourced separately. The EML shortage is directly accelerating this transition as hyperscalers seek supply chain diversification.
Signal Path
CW laser (InP or other) → silicon photonics modulator chip (TSMC fab) → pluggable or near-package module
Timeline
Ramping now through 2027. CW laser demand projected to 2x in 2026 vs 2025.
Key Bottleneck
CW laser coupling precision and thermal management at high power
Portfolio Positioning
Developing CW laser sources for Gen 2 alongside Gen 1 EML. Multi-generation platform.
POSITIONEDCW laser R&D underway. Risk: Gen 1 moat partially disrupted if CW+SiPho scales fast.
TRANSITIONINGPCIe/CXL connectivity fabric benefits regardless of which optical generation wins.
AGNOSTIC BENEFICIARYWaveLogic coherent optics integrates SiPho. Hyperscaler backbone interconnect.
BENEFICIARYThe optical engine is co-packaged directly on the switch ASIC package — eliminating the pluggable transceiver entirely. Light never has to travel through a connector. An External Laser Source (ELS) module provides CW light to the co-packaged engine. This eliminates ~30% of power consumption vs. pluggable optics. Jensen Huang confirmed at GTC 2026 (March 17) that CPO revenue ramp is 2027–2028, not 2026. Coherent demonstrated a 6.4T socketed CPO at OFC 2026 (March 17–19, LA).
Signal Path
ELS module → optical engine co-packaged on switch ASIC → eliminates pluggable connector entirely
Timeline
2027–2028 per Jensen Huang at GTC. Coherent demonstrating at OFC 2026 now.
Key Bottleneck
Complex packaging yield, ELS thermal management, ecosystem standardization
Portfolio Positioning
Demonstrating 6.4T socketed CPO + ELS at OFC 2026 right now. Deepest CPO moat.
LEAD POSITIONCopperEdge DSP + HieFo InP integration. CPO signal integrity play. Timeline: 2027.
2027 CATALYSTCXL/PCIe fabric becomes more critical as CPO integrates into the switch package.
INFRASTRUCTUREPrimary CPO customer. Vera Rubin (2027) likely first NVIDIA platform with CPO.
DEMAND DRIVERThe ultimate end state: laser, modulator, detector, and electronic driver all fabricated on a single chip in a single manufacturing process. Eliminates all coupling losses between components. Requires either a new fab process that can handle both III-V materials (for lasers) and silicon (for electronics), or a breakthrough in silicon-based light emission. Still largely in academic research. The companies that crack this will own the next decade of optical interconnect.
Signal Path
Single chip: laser + modulator + detector + driver → zero coupling loss → maximum density and efficiency
Timeline
2029+ — still in research. No commercial products yet.
Key Bottleneck
III-V / silicon process integration. Silicon light emission. Thermal co-design.
Portfolio Positioning
Deepest III-V + SiPho integration expertise. Best positioned for monolithic transition.
LONG-TERM CONTENDERSilicon photonics DSP integration. Investing heavily in monolithic research.
RESEARCH STAGEIntel Photonics Solutions Group. Long-running silicon photonics program.
RESEARCH STAGEStrategic Takeaway
The EML shortage is a double-edged catalyst
Near-term (2025–2026): Extremely bullish for LITE and COHR — they make EMLs and demand exceeds supply. But the shortage is also the forcing function accelerating the Gen 2 transition — hyperscalers do not want to be held hostage to InP fab capacity, so they are funding CW+SiPho alternatives aggressively. COHR is the most durable long-term position because it is the only company demonstrating products across Gen 1, Gen 2, and Gen 3 simultaneously at OFC 2026 right now. SMTC is a 2027 story that got priced as a 2026 story — the thesis is intact, the timeline is longer. AXTI has a structural headwind if Gen 2/3 adoption accelerates beyond InP substrates.
GTC 2026 · March 17 · San Jose
The Fourth Piece Ships
NVIDIA's Groq 3 LPX changes the inference business model — and the investment thesis for every name on this watchlist.
The Four-Piece Convergence Thesis
BEP Research's thesis: NVIDIA's AI infrastructure platform has four convergent pieces. The first three — stacked memory (HBM), backside power delivery, and the compiler (CUDA/Dynamo) — were already in place. The fourth piece — optically-synchronized dataflow inference — shipped at GTC 2026 as the Groq 3 LPX.
The Thresher vs. The Speed Demon
AI inference has two fundamentally different phases. The GPU dominates one. The LPU owns the other. NVIDIA now sells both.
Processing the input prompt. Compute-intensive, parallelizable, HBM-memory-bound. GPUs excel here. Handles training, prefill, and high-concurrency batch inference.
Generating tokens one by one. Memory-bandwidth-bound, latency-sensitive, sequential. LPUs with 500MB on-chip SRAM per chip are dramatically better. 256 LP30 chips per rack, 640 TB/s scale-up bandwidth.
The Revenue Tier Model — Jensen's Slide That Changes Everything
NVIDIA's keynote slide showed that pairing LPX racks with NVL72 racks unlocks dramatically higher revenue per megawatt. This is not a replacement cycle — it is an additive revenue layer. Jensen said ~25% of every AI cluster should be LPX-type inference hardware.
Source: NVIDIA GTC 2026 keynote · Jensen Huang · March 17, 2026 · SAP Center, San Jose
Groq 3 LPX — Key Specifications
Portfolio Implications — How LPX Changes Each Name
The street still models NVIDIA primarily as a training hardware company. LPX changes the revenue model: NVIDIA now sells a GPU rack for training/prefill AND an LPU rack for decode. The combination unlocks 10x more revenue per megawatt for hyperscalers. This is additive revenue, not a replacement cycle. The $20B Groq acquihire is already looking cheap.
If 25% of every AI cluster is LPX-type inference hardware, and LPX racks communicate optically at 640 TB/s, the optical transceiver and CPO market just got a structural demand uplift. The article's title references 'optically-synchronized dataflow inference' — the LPX architecture is exactly the system CPO was designed for. COHR's OFC 2026 CPO demonstration is directly relevant.
The optical interconnect fabric tying 256 LPUs together at 640 TB/s requires silicon photonics. Tower's PH18 platform is the leading open-foundry SiPho process. This is a direct demand driver for TSEM that the market has not fully priced. The LPX architecture is a proof-of-concept for exactly the kind of rack-scale optical interconnect Tower is building capacity to serve.
Astera Labs makes the PCIe/CXL connectivity fabric that connects LPX racks to NVL72 racks and to the host CPU. In a heterogeneous inference architecture with GPU racks and LPU racks working together, the connectivity fabric between them becomes critical infrastructure. ALAB is the toll road. This is the strongest argument yet for ALAB's valuation — currently 51% off its 52-week high.
The LPU uses SRAM, not HBM. The Groq architecture was specifically designed to avoid HBM because HBM has too much latency for decode-phase inference. If 25% of AI cluster compute shifts to LPX-type SRAM architectures, that is 25% of the compute budget that does not buy HBM. Mild headwind at the margin — not a thesis-killer, since NVL72 still uses massive HBM4. But worth noting going into Wednesday's earnings (Mar 19).
Same optical demand uplift as COHR. LPX rack-scale optical interconnect at 640 TB/s is exactly the kind of system that drives 800G and 1.6T transceiver demand. The CPO 'headwind' narrative for LITE is backwards — CPO is the enabling technology for the LPX architecture to work at scale. LITE's EML lasers feed the CW+SiPho engines that power Gen 2 and Gen 3 optical interconnects.
⚠ Key Risk — The One Thing the Bull Case Requires
The LPX ships in Q3 2026 — Jensen said "very likely Q3." That is 6 months away. Samsung has to manufacture 256-chip racks at scale, the optical interconnect fabric has to work at 640 TB/s, and the Dynamo inference software stack has to orchestrate the whole thing. NVIDIA has never shipped a product like this before. The probability of a delay is non-trivial. The market will price this in before it ships — but execution risk is real. Watch for any Q3 2026 guidance language in upcoming earnings calls from NVDA, MU, and ALAB.
Investment Framework · Vernor Vinge (1993) × Jason's Chips (2026)
Why This Cycle Is Different
Three principles for investing in the age of AI — derived from a 1993 paper that predicted almost everything about today's AI boom, and why the dot-com analogy collapses under scrutiny.
⚠ KEY RISK NOT IN THE ARTICLE: The staircase demand curve assumes scaling laws hold — that each new frontier model requires meaningfully more compute than the last. If scaling laws break down before AGI (capability plateau), the demand curve flattens and the infrastructure thesis weakens. Additionally, ASML and TSMC face geopolitical concentration risk that the "atoms beat bits" framing amplifies rather than reduces. These risks do not invalidate the thesis but must be sized accordingly.
Options Strategy · Weekly Income
Covered Call Screener
~4,057 NASDAQ stocks · Live data updated hourly · Nearest Friday expiry · Sorted by weekly yield
Geopolitical Trade · Energy Sector
Oil Sector Stock Picks
US-Iran conflict driving Brent crude +8–9%. Strait of Hormuz handles ~20% of global oil supply. These are the US energy names positioned to benefit.
2028 Global Intelligence Crisis: Energy Implications
Trigger
Supreme Leader Khamenei killed in targeted strike. IRGC activates Hormuz contingency plans. Iran threatens to mine the strait.
Market Impact
Brent crude +8–9% to ~$90/bbl. WTI +7.5%. Natural gas +12%. US energy sector ETF (XLE) +5.8% on open.
Hormuz Risk
~20% of global oil supply transits Hormuz daily. Full closure would spike Brent to $120–150/bbl within 72 hours.
Fwd P/E
13.2x
ExxonMobil
Largest US oil major. Permian Basin growth + Pioneer acquisition adds 1.3M boe/d. Fortress balance sheet, $17.5B buyback program.
Dividend
3.5%
Analyst Target
$135
Fwd P/E
14.1x
Chevron
Hess acquisition adds Guyana deepwater exposure — one of the world's lowest-cost oil fields. Strong free cash flow yield.
Dividend
4.2%
Analyst Target
$185
Fwd P/E
12.8x
ConocoPhillips
Pure-play E&P with the best cost structure in the industry. Marathon Oil acquisition adds 2B boe of low-cost inventory.
Dividend
1.9%
Analyst Target
$140
Fwd P/E
11.5x
EOG Resources
The tech-forward E&P. Uses proprietary data analytics to find premium-return wells. Best-in-class Permian and Eagle Ford operator.
Dividend
2.8%
Analyst Target
$160
Fwd P/E
15.3x
Occidental Petroleum
Warren Buffett's oil bet — Berkshire owns 28%. Carbon capture technology (DAC) provides long-term optionality. Permian-heavy portfolio.
Dividend
1.8%
Analyst Target
$72
Fwd P/E
11.2x
Halliburton
Oilfield services giant. As oil prices rise, E&P capex surges and Halliburton captures the spending on drilling, completion, and production.
Dividend
2.1%
Analyst Target
$38
Forward P/E Comparison — Oil Picks
⚠ Geopolitical Premium Risk
Oil prices are currently elevated by a geopolitical risk premium — not fundamental supply/demand. If the US-Iran conflict de-escalates, Hormuz risk fades, or a ceasefire is announced, Brent could rapidly retrace 50–70% of its gains. These picks are event-driven trades, not long-term value investments. Size positions accordingly and set stop-losses.
Geopolitical Trade · Defense Sector
Defense Sector Stock Picks
US-Iran conflict and elevated global tensions are driving defense budgets higher. These four prime contractors hold the largest backlogs in the industry.
Why Defense Wins in a Hot War
Immediate Demand
Patriot missile batteries, PAC-3 interceptors, and THAAD systems are being rushed to the region. RTX and LMT are the primary suppliers.
Budget Tailwind
US defense budget likely to see emergency supplemental spending. NATO allies accelerating procurement. Multi-year backlog expansion for all four names.
Long-Cycle Revenue
Unlike oil, defense contracts span 10–30 years. A conflict today creates procurement cycles that persist long after the shooting stops.
Fwd P/E
17.8x
Lockheed Martin
Largest US defense contractor. F-35 program generates decades of recurring revenue. Missile defense systems (THAAD, PAC-3) are direct Iran conflict beneficiaries.
Dividend
2.8%
Analyst Target
$580
Fwd P/E
21.4x
RTX Corp. (Raytheon)
Patriot missile systems are the primary air defense against Iranian ballistic missiles. Pratt & Whitney jet engines add commercial aviation diversification.
Dividend
2.1%
Analyst Target
$155
Fwd P/E
16.2x
Northrop Grumman
B-21 Raider next-gen stealth bomber program. Space and cyber defense leader. Long-cycle programs provide revenue visibility through 2040+.
Dividend
1.7%
Analyst Target
$590
Fwd P/E
18.5x
General Dynamics
Diversified defense: Abrams tanks, nuclear submarines (Virginia class), Gulfstream jets. Starliner contract adds space exposure. Strong backlog visibility.
Dividend
2.0%
Analyst Target
$320
Forward P/E Comparison — Defense Picks
⚠ Valuation & Conflict Risk
Defense primes are not cheap — they trade at 16–21x forward earnings, reflecting already-elevated expectations for budget growth. A rapid ceasefire or diplomatic resolution could cause a short-term pullback. The long-term thesis (rising global defense budgets, NATO 2% GDP commitments, US re-arming) remains intact regardless of near-term conflict outcomes.
Side-by-Side Analysis
Comparative Overview
| Metric | CDNS | SNPS | CLS | NVDA | NOW | TSM |
|---|---|---|---|---|---|---|
| Latest Revenue | $4.64B | $7.05B | ~$13.9B | $215.9B | $13.3B | ~$122B |
| Revenue Growth | +13.5% | +15.1% | +28% | +65% | +21% | +39% |
| P/E Ratio | ~65–70x | ~65–75x | ~39x | ~35x | ~85x | ~28x |
| Market Cap | ~$81B | ~$86B | ~$10B | ~$4.4T | ~$200B | ~$950B |
| Business Type | EDA Software | EDA Software | Manufacturing | AI Chips / GPU | Enterprise SaaS | Chip Foundry |
| AI Role | Chip Design Tools | Chip Design Tools | Hardware Builder | AI Compute Engine | Workflow Automation | Chip Manufacturer |
| Risk Profile | Medium | Medium | High | Medium-High | Medium-Low | Medium-High |
| Investment Style | Quality & Momentum | Quality at Discount | Aggressive Growth | Highest Conviction | Stable Compounder | Highest Conviction |
Revenue Comparison — Latest Fiscal Year (USD Billions)
Portfolio Tracker
Portfolios & Watch Lists
Track positions, gain/loss, and allocation across multiple portfolios.
Prices as of Feb 28, 2026 close* For informational purposes only. Not financial advice. Click any Notes cell to add your own comments.
Initial Watch List
Core AI-era research positions — EDA software, chip foundry, hardware manufacturing, enterprise software, and GPU computing.
| Ticker | Company | Price | Shares | Avg Cost | Value | Gain / Loss | Notes |
|---|---|---|---|---|---|---|---|
CDNS | Cadence Design Systems | $301.40 | — | — | — | — | + add note |
SNPS | Synopsys, Inc. | $414.00 | — | — | — | — | + add note |
CLS | Celestica Inc. | $277.63 | — | — | — | — | + add note |
NVDA | NVIDIA Corporation | $177.19 | — | — | — | — | + add note |
NOW | ServiceNow, Inc. | $108.01 | — | — | — | — | + add note |
TSM | Taiwan Semiconductor Manufacturing | $183.50 | — | — | — | — | + add note |
MDB | MongoDB Inc. | $253.37 | — | — | — | — | + add note |
CRUS | Cirrus Logic, Inc. | $95.00 | — | — | — | — | + add note |
TXN | Texas Instruments Inc. | $172.00 | — | — | — | — | + add note |
ALAB | Astera Labs, Inc. | $68.40 | — | — | — | — | + add note |
COHR | Coherent Corp. | $82.50 | — | — | — | — | + add note |
CIEN | Ciena Corporation | $91.20 | — | — | — | — | + add note |
GLW | Corning Incorporated | $47.80 | — | — | — | — | + add note |
IPGP | IPG Photonics Corp. | $72.15 | — | — | — | — | + add note |
This is a research watchlist — share counts are not tracked here. Use Holly's Portfolio or Tracie's Portfolio tabs to track actual positions.