Semiconductor stocks to watch before Q2 2026 earnings

In Q1 2026, Alphabet reported a Google Cloud backlog of $460 billion. Microsoft disclosed an Azure backlog of $80 billion that it cannot fulfill because of power constraints. Meta raised its 2026 capex guide to the $125–145 billion range. NVIDIA shipped $75.2 billion of data center revenue in a single quarter, up 92% year-over-year. ASML's order book stood at €38.8 billion at year-end 2025. Vertiv's backlog hit $15 billion, up roughly 80% YoY. Eaton's electrical-Americas data center order book reached 228 GW — about twelve years of build at 2025 run rate.

Those numbers were Q1. Now Q2 results are coming — and they will either confirm the demand picture or complicate it.

This guide is for the investor reading the order books before the prints arrive.

The earnings calendar you're working with

The next six weeks determine how semiconductor multiples trade into year-end. Key dates as of mid-July 2026:

Company	Expected window	What to watch
Tesla	July 22–24	AI inference infra commentary, Dojo capex update
Alphabet	July 28–30	Google Cloud backlog update; capex intensity vs $460B prior figure
Microsoft	July 28–30	Azure backlog commentary; whether the "can't fulfill due to power" language continues
Meta	July 29–31	$125–145B capex confirm or raise; MTIA ramp status
Apple	July 30–Aug 1	TSMC N2 ramp read-through; iPhone AI silicon commentary
Amazon	July 30–Aug 1	Trainium/Inferentia deployment; AWS capex update
NVIDIA	August 20–22	Q2 FY27 results; the single most-watched print in the semi universe

NVIDIA's August 20–22 print is the terminal event for the near-term cycle read. Market consensus entering Q2 FY27 expects data center revenue sustained above the $70–80B quarterly run rate established in Q1 FY27, with continued 80–90%+ YoY growth. Any deceleration in that number — or softness in Q3 guidance — compresses multiples across the entire stack. Any beat widens the order-book-vs-multiple divergence thesis further.

The thesis going into all of these prints: by chart-based reads, semiconductor stocks look stretched. By order-book reads, the demand is contracted years out. Both pictures are real. Both will stay real regardless of what any single quarter prints.

What this guide is and isn't

It is: 30 semiconductor and semi-adjacent names organized into 6 value-chain layers, with one-paragraph verdicts on each, three model portfolios at different risk appetites, and the rupee math an Indian-resident investor needs to size positions and file taxes properly.

It is not: investment advice. Not a momentum scan. Not bullish on everything that has "AI" in the pitch.

A note on data. Every revenue, backlog, gross margin, customer concentration and segment number below is sourced from the latest filings, earnings calls, or trade-press reporting through early July 2026 — full per-name data sheets with primary source links are in the sources companion. Multiples and stock prices move daily — forward P/Es are point-in-time references for relative ordering, not current trading levels. Before acting on any name, pull the latest 10-Q and the current price. Read the trade-offs in the model portfolios before sizing anything.

The framework

Entering earnings season, you don't bet on momentum. You bet on businesses where the customer cannot postpone the order. Multiple compression hurts most when the buyer can wait a quarter. Insulation against that risk — not earnings growth — is what makes a semiconductor name "durable" into and through a results season.

Six layers, ranked by how much insulation each tier has against multiple compression:

Layer	What it is	Insulation level
1. Equipment	Machines fabs cannot operate without	Highest
2. EDA / IP	Design tools every new chip starts in	Very high
3. Foundries / IDMs	The factories	High (with geopolitical tail)
4. AI compute & networking	GPUs, custom ASICs, switching silicon	High (but bifurcated)
5. Memory, analog, specialty	DRAM, NAND, analog, mobile, edge AI	Medium
6. Infrastructure & power	Cooling, REITs, electrical, generation	High (different cycle)

Each layer below: brief thesis, names with crisp verdicts, then the Top pick of the layer.

Verdict format throughout:

Verdict — [Action]: [The reason in one line]. [The caveat in one line].

Actions: Core buy (full position), Add (build into it), Hold (own but don't add at these prices), Watch (waiting for entry), Hedge (small position as risk offset), Skip (multiple does not compensate for the risk), Avoid (the equity itself is structurally damaged).

Layer 1 — The toll booth: Equipment

Foundries cannot exist without these machines. TSMC raised 2026 capex to the high end of the $52–56 billion range. Samsung and Intel Foundry are spending alongside. Memory makers are expanding HBM capacity. When fab capex flows like this, equipment companies print — and earnings season will give us the first Q2 shipment confirmation.

ASML

The only EUV supplier on earth. Nikon and Canon publicly walked away years ago.

Q1 2026 net sales €8.8 billion at 53.0% gross margin. Year-end 2025 order book €38.8 billion, of which EUV roughly €7.4 billion. 2026 sales guide €36–40 billion — over a year of revenue locked in. High-NA EUV (Twinscan EXE) ships at approximately $380 million per machine; ASML plans 20 per year by 2028, with 10–20 already booked. TSMC has deferred High-NA deployment until 2029; Intel and Samsung are the early adopters.

What to watch in Q2 results: EUV shipment pace commentary and any update to the order book. Q1's €38.8B figure was already above a full year of guided revenue — a sustained or growing order book into Q2 confirms the insulation thesis.

Catalyst: TSMC's N2 ramp through 2026–27; Samsung and Intel High-NA adoption decisions.

Risk: China export controls keep tightening. The January 2026 BIS final rule shifted advanced-computing semiconductor licensing to case-by-case from presumption-of-denial. Successive rule rounds have removed marginal China revenue.

Verdict — Core buy: EUV monopoly with a 12-month-plus order book locked in. If you only own one semi name, this is it. Size for index-comparable weight, not for swing-trading.

Applied Materials (AMAT)

Broadest WFE portfolio — deposition, etch, CMP, ion implant.

Q1 FY26 was a revenue record at 49.1% non-GAAP gross margin. The CEO publicly guides CY26 semi equipment revenue above +20%. AI-driven leading-edge logic, HBM, and advanced packaging are the drivers.

Catalyst: TSMC capex flowing through to multi-quarter shipment visibility; HBM cycle pulling memory tool demand.

Risk: Trade tension with China is the most concentrated risk. If memory CapEx slows in H2 2026, AMAT decelerates harder than KLAC.

Verdict — Add: Broadest exposure to the foundry capex cycle. Not the cheapest WFE name; it is the most diversified.

Lam Research (LRCX)

Etch leader, deposition strong, memory-exposed.

Q3 FY26 (March quarter): $5.84 billion at 49.9% gross margin — record. HBM and advanced packaging momentum.

Catalyst: HBM3E / HBM4 capacity additions at Micron, Samsung, SK Hynix.

Risk: Memory cyclicality. LRCX rises and falls with the DRAM/NAND cycle harder than AMAT or KLAC.

Verdict — Add: Highest leverage to the memory CapEx upcycle currently underway. Buy expecting the up-cycle to continue another 18–24 months, not forever.

KLA Corporation (KLAC)

Process control and metrology — the quality gate of every fab. Near-duopoly with Onto Innovation. Highest margins in WFE (60%+ gross historically). Single-customer concentration: TSMC alone roughly 38% of revenue.

Catalyst: Advanced packaging (CoWoS) growth at TSMC drives metrology demand disproportionately. CoWoS lead times are 50+ weeks — the tightest single bottleneck in the AI supply chain.

Risk: TSMC concentration; advanced packaging capex cycle peak.

Verdict — Core buy: Cleanest margin profile in WFE, least cyclical of the three big names, structural duopoly. If you can only own one WFE name, KLAC.

Onto Innovation (ONTO)

KLAC's junior partner in process control — duopoly at the bleeding edge.

Q1 2026: revenue $291.9 million (+9.5% YoY), EPS $1.42. Top customer signed a Volume Purchase Agreement above $240 million for Dragonfly G5 + 3D bump metrology through 2027. Raised 2026 revenue outlook above $1.3 billion.

Catalyst: HBM4 ramp through 2027 — Onto's Dragonfly G5 is specifically chosen for HBM stacking inspection.

Risk: Customer concentration heavier than KLAC. Smaller scale.

Verdict — Add (advanced): Real catalyst on HBM4 ramp, but smaller name with thinner liquidity. Size one-third of a KLAC position, only for portfolios that already own KLAC.

ACM Research (ACMR)

Listing / regulation risk flag. ACMR is dual-structured — US-listed parent owns the ACM Shanghai subsidiary; the operating business is heavily China-customer-exposed (SMIC, YMTC, CXMT — historically >80% of revenue).

Q1 2026 revenue around $240 million (implied from +34% growth and FY26 guide of $1.08–1.175 billion). 46.5% gross margin — above the long-term 42–48% mid-point. FY26 reaffirmed (not raised).

Catalyst: China semi build-out continues to need wet-process / cleaning equipment.

Risk: US/China policy risk; reaffirmed (not raised) full-year guide signals back-half loaded; sentiment trades like a China stock, not an equipment stock.

Verdict — Skip (most readers): Semi exposure achievable through ASML and TSM with much less overhang. Only for portfolios that explicitly want a China cyclical leveraged play with eyes-open listing risk.

Layer 1 top pick: ASML for conviction, KLAC for cleaner risk-adjusted entry.

Layer 2 — The design tools: EDA / IP

Every new AI chip — TPU, MTIA, Trainium, Maia, Anthropic's ASIC — starts in Cadence or Synopsys. The market is a near-duopoly. Siemens EDA holds meaningful share but lags on AI workload tooling.

Cadence (CDNS)

Q1 2026: revenue $1.474 billion (+19% YoY). Record backlog $8.0 billion. $4.0 billion of remaining performance obligations expected over the next 12 months. 45% non-GAAP operating margin. 2026 revenue guide raised to +17% YoY.

Recurring revenue, sticky tools, high switching cost.

Catalyst: AI workload chip starts are roughly doubling — every hyperscaler designs its own silicon now, and every project is a CDNS deal.

Risk: China export-control overhang (roughly 7–10% of revenue at risk in late 2025; partly resolved, regulation remains live).

Verdict — Core buy: Every new AI accelerator program is a Cadence deal. Recurring backlog, duopoly structure, structural growth in chip design starts.

Synopsys (SNPS)

Cadence's twin. Backlog $11.0 billion at April 30, 2026. Greater IP-licensing mix than CDNS — the Arm-comparable royalty stream is meaningful.

Catalyst: Ansys acquisition closed; cross-sell into AI and automotive design. Continued AI chip starts.

Risk: Same China overhang. Larger acquisitions to integrate.

Verdict — Core buy: Same thesis as CDNS at a slightly cheaper multiple plus richer IP licensing mix. If you want the EDA duopoly thesis at the cheaper of the two, SNPS.

Arm Holdings (ARM)

The IP company. FY26 royalty revenue $2.61 billion (+21% YoY); datacenter royalty more than doubled. Q4 FY26 royalty $671 million (+11% YoY). FY26 total revenue $4.92 billion record.

Trades at very rich EV/Revenue (around 38×) and price-to-royalty-revenue (around 100×).

Catalyst: Royalty-rate step-up with each generation of Arm v9 architecture. Datacenter share expansion (NVDA Grace, Microsoft Cobalt, AWS Graviton).

Risk: Valuation prices in zero slip for years. RISC-V long-term substitution risk in low-end IoT/MCU.

Verdict — Watch: Beautiful business. Wait for a 25–30% pullback before sizing a real position.

Layer 2 top pick: CDNS or SNPS — pick one or barbell them.

Layer 3 — The factory: Foundries and IDMs

The companies that physically make the chips. The Taiwan tail risk concentrates here.

TSMC (TSM)

Global foundry share 70.4% in Q4 2025, projected approximately 75% in 2026. Leading-edge (3nm / 2nm) share above 90%.

Customer concentration has flipped in a way that favors durability. NVIDIA overtook Apple in 2025 as TSMC's #1 customer at roughly 19% of revenue. Apple is now #2 at 17%. Top 10 customers represent roughly 76% of revenue. Hyperscaler AI capex is contracted years out; consumer phone cycles aren't.

2026 capex raised to the high end of $52–56 billion. FY26 revenue guided above +30% YoY (around $158 billion). Gross margins recovering toward 55%+. N3 fully booked through 2026. CoWoS advanced packaging capacity ramping from ~80K/month to 115–140K/month in 2026 — still tight (50+ week lead times). June 2026 monthly revenue data confirmed continued AI-related demand tracking above the 30%+ annual guide.

Catalyst: N2 volume production started late 2025; end-2026 target ~100K wafers/month; 2027 target ~200K. Apple Arizona Fab 21 producing 4nm at par-with-Taiwan yields.

Risk: Taiwan geopolitics. CSIS's repeated wargaming work (the 2023 invasion study and the 2025 "Lights Out" blockade study) puts modal analyst probability of a major Taiwan event at roughly 10–25% by 2030. Wide dispersion; no analytical hedge.

Verdict — Add (sized down for tail risk): Second-most defensible structural bet, but carries a tail risk you can't model. Half the position you'd want at no geopolitical overhang. Express the rest of the theme via ASML.

Intel (INTC)

The turnaround narrative is dated.

18A has three confirmed external customers: Amazon, Microsoft, US DoD. NVIDIA and Broadcom tested 18A but have not committed to production. Apple has 18A-P PDK and is running simulations. The equity now trades at a triple-digit forward P/E — the turnaround is fully priced.

Catalyst: 18A external customer wins (especially NVDA / AVGO if they commit); CHIPS Act milestones.

Risk: If any testing customer walks, the multiple compresses fast.

Verdict — Skip (or trade only with a tight stop): The story is real and the catalysts are real, but the equity already prices a successful turnaround. Wait for a meaningful disappointment to re-enter cheap.

GlobalFoundries (GFS)

Mature-node specialist (12nm and above). RF, embedded, automotive, IoT.

Q1 2026: revenue $1.634 billion (+3.1% YoY), EPS $0.40. Gross margin 27.6% GAAP / 29.0% non-IFRS. Automotive segment grew +24% YoY despite −11% QoQ. Smart-mobile dropped to roughly one-third of revenue; non-mobile and services now two-thirds.

Catalyst: Auto low-double-digit growth FY26; communications infrastructure + datacenter PSU + silicon photonics ramp; CHIPS Act-fueled US capacity 50% complete.

Risk: Not on the AI compute bandwagon — silicon photonics is the only AI-adjacent line. Mature-node margins structurally lower than TSM N3/N2.

Verdict — Hold (for diversification): Defensive mature-node exposure, off the bleeding-edge multiple cycle. Useful for a 5–10% sleeve in a balanced semi portfolio, not as a core holding.

UMC (United Microelectronics)

The Taiwanese mature-node alternative. Pure cyclical play.

Q1 2026: revenue NT$61.04 billion (~US$1.85 billion), +5.5% YoY, −1.2% QoQ. Gross margin 29.2%. 22/28nm accounts for 34% of revenue (22nm alone hit a record 14%). Utilization 79%.

Catalyst: H2 2026 price hike telegraphed; 22nm volume ramp.

Risk: Same Taiwan tail risk as TSM but without the AI accelerator upside. Squeezed between SMIC (subsidized Chinese capacity) and TSMC (technology lead).

Verdict — Skip: Taiwan risk without the AI upside. Express mature-node exposure via GFS instead, or via the equipment makers that sell to both.

Layer 3 top pick: TSM sized for tail risk; GFS as defensive mature-node diversification.

Layer 4 — AI compute and networking

This is where earnings season will be loudest. It is also where the bifurcation between "structural moat" and "expensive momentum" matters most heading into Q2 prints.

NVIDIA (NVDA)

Q1 FY27 (quarter ended April 26, 2026): Data Center revenue $75.2 billion, +92% YoY; total revenue $81.6 billion, +85% YoY. Data Center is approximately 92% of total revenue. Within DC: Hyperscale $37.9 billion + AI Clouds / Industrial / Enterprise $37.4 billion.

NVIDIA reports Q2 FY27 on August 20–22. Market consensus entering the print anticipates data center revenue sustained at the $70–80B+ quarterly run rate, with continued 80–90%+ YoY growth off a more demanding comparable base. The guidance for Q3 FY27 — and any commentary on Rubin platform ramp — will be the key read-through for the full year.

The counterintuitive read: NVIDIA's forward P/E is approximately 25× on FY27 estimates — broadly in line with the 5-year average rather than above it. Earnings have outrun price. The stock is at multi-year highs on dollar price, but the multiple itself is not extended relative to history.

Catalyst: Blackwell B200 + GB200 NVL72 systems shipping; Rubin platform in 2027; Q2 FY27 print August 20–22.

Risk: ASIC threat (Google TPU, Meta MTIA, AWS Trainium). The current forward multiple is forgiving of mild deceleration; any quarter that breaks the 70%+ DC growth pattern compresses the multiple.

Verdict — Core buy: Narrative is "expensive" but the forward multiple is roughly in line with history while earnings are at all-time highs. Concentration risk is real — size to your conviction on the ASIC-vs-merchant-silicon debate.

AMD

The second pillar of the AI compute trade.

Q1 FY26: revenue $10.3 billion (+38% YoY), EPS $1.37. Non-GAAP gross margin approximately 54%. Data Center segment $5.8 billion (~56% of revenue), +57% YoY — driven by EPYC CPUs + Instinct MI3xx/MI400 GPUs. Client + Gaming $3.6 billion (~35%), +23% YoY. Embedded (Xilinx legacy) $873 million.

Forward P/E in the 45–50× range. Microsoft, Meta, and Oracle disclosed multi-billion-dollar MI3xx commitments in 2025–26.

Catalyst: MI400 series ramp H2 2026; the OpenAI 6 GW AMD deployment commitment scaling.

Risk: AMD has historically played catch-up to NVIDIA on AI accelerator execution; CUDA + Blackwell/Rubin cadence keeps the gap structural. Custom-ASIC encroachment is a real margin pressure.

Verdict — Add: The structural "alternative to NVDA" trade at a cheaper multiple. Position size 30–50% of your NVDA position — directional bet on hyperscaler vendor diversification.

Broadcom (AVGO)

Custom AI ASICs + networking + VMware compounder.

Q1 FY26 AI revenue $8.4 billion (+106% YoY); Q2 guide $10.7 billion. AI backlog $73 billion. CEO Hock Tan public 2027 target: above $100 billion of AI revenue. April 14, 2026 Meta MTIA deal extends through 2029 with 1 GW initial deployment — described as "the industry's first 2nm AI compute accelerator." Anthropic has access to 3.5 GW via the Google–Broadcom TPU capacity. AVGO holds approximately 70% of the custom AI accelerator market.

VMware integration tracking ahead of plan. Forward P/E around 31×.

What to watch in Meta and Alphabet Q2 prints: MTIA volume commentary from Meta (July 29–31) and TPU capacity discussion from Alphabet (July 28–30) are direct read-throughs for AVGO backlog conversion pace.

Catalyst: Meta MTIA volume ramp; new ASIC customer announcement(s).

Risk: Customer concentration (top 3 are heavy share). VMware execution risk.

Verdict — Core buy: The structural hedge to NVDA — wins under both "NVDA dominates" and "hyperscalers shift to ASICs." Cleaner backlog visibility than NVDA. The barbell with NVDA is the call.

Marvell (MRVL)

Pure custom ASIC + optical networking. Trainium successor work at AWS, Microsoft custom silicon, Meta MTIA-adjacent.

Q1 FY26: revenue $1.895 billion (+63% YoY). Non-GAAP EPS $0.62. Non-GAAP gross margin ~60%. Data Center segment $1.8 billion (76% of revenue), +27% YoY. Forward P/E in the 50–55× range. Top customer (almost certainly AWS) disclosed above 25%; top 3 likely 50%+.

Catalyst: FY27 revenue guide raised to ~$11.5 billion (+40%); FY28 outlook approximately $16.5 billion (+45%). Custom silicon ramp at Amazon and Microsoft in Q3–Q4 CY26.

Risk: Custom silicon margins lower than merchant GPU. Hyperscalers can in-source if Marvell stumbles on a node transition. AWS Trainium roadmap shifts are concentration risk.

Verdict — Add: The fast-follower in custom AI silicon. Smaller, more leveraged play than AVGO. Position 30–40% of your AVGO position; partially redundant if you already own AVGO heavily.

Arista Networks (ANET)

Switching silicon for AI data center fabrics. EOS network OS is chosen by hyperscalers as the merchant alternative to NVIDIA Spectrum-X for back-end AI fabrics.

Q1 2026: revenue $2.71 billion (+35.1% YoY). Non-GAAP operating margin 47.8%, operating income $1.29 billion. Microsoft + Meta together historically ~35–40% of revenue; top 5 around 55%. Forward P/E in the 43–48× range.

Catalyst: FY26 revenue growth target raised to +27.7% (~$11.5 billion); AI target raised to $3.5 billion. 1.6T production scale planned for 2027. XPO optics form factor with 100+ vendor endorsements.

Risk: Spectrum-X (NVIDIA proprietary) and emerging custom-silicon scale-up (NVLink Fusion, AMD UALink) compress merchant scale-out share. Meta / Microsoft hardware mix shift could be sudden.

Verdict — Add: The clean networking pick for AI data centers. Customer concentration is the risk — size accordingly. Sub-30% position weight in any AI-heavy semi portfolio.

Layer 4 top picks: NVDA + AVGO as the barbell. ANET for networking. AMD if you want hyperscaler-diversification leverage.

Layer 5 — Memory, analog, specialty

The cyclical tier mixed with niche AI-adjacent power and connectivity plays.

Micron (MU)

HBM is the entire story. Q3 FY26 guide: revenue $33.5 billion ± $750 million at approximately 81% gross margin — peak-cycle levels. HBM3E sold out through most of 2026. FY26 capex raised above $25 billion. SK Hynix and Samsung have telegraphed an approximate 20% HBM3E price hike for 2026.

Catalyst: HBM4 ramp begins Q3 2026; AI accelerator HBM demand visibility through 2027.

Risk: 81% gross margin is historically the cycle top. Memory cycles compress fast when supply catches demand. Capex expansion now seeds the next over-supply.

Verdict — Hold (if already own); Skip (if just buying): The cycle is real but the upside is priced. Buying here is buying at peak margin expansion. Existing holders can stay — new positions face asymmetric downside.

Western Digital (WDC)

Post-split (February 2025), WDC is the HDD-pure entity. NAND business is now standalone SanDisk (SNDK). The thesis is AI nearline HDDs.

Q3 FY26 (CQ1 2026): revenue $3.34 billion (+45% YoY). Gross margin 50.2% GAAP / 50.5% non-GAAP — all-time high. Free cash flow $978 million (29% margin). Cloud (data center nearline HDD) $3.0 billion (89% of revenue), +48% YoY. 222 exabytes shipped (+34% YoY). 4.1 million ePMR drives at up to 32TB. Forward P/E in the 30–35× range — stock has more than doubled YTD 2026.

Catalyst: Q4 FY26 revenue guided +36–44% YoY; gross margin 51–52%; HAMR transition 2026–27; 40TB drives in qualification.

Risk: HDDs are still HDDs — flash continues to encroach on hot-tier. Multiple has re-rated meaningfully; the "deep value" framing no longer applies after the 2026 run.

Verdict — Hold / Watch: The AI nearline thesis worked. After 100%+ YTD it is no longer a contrarian-cheap setup. Existing holders stay; new positions wait for a 15–20% pullback.

Seagate (STX)

The other HDD name. HAMR leader.

Q3 FY26 (CQ1 2026): revenue $3.11 billion (+44% YoY). Adjusted EPS $4.10 (+116% YoY). Non-GAAP gross margin 47% — all-time high. Operating margin around 30%. Data center 88% of exabytes, 80% of revenue. Forward P/E in the 14–17× range — cheaper than WDC.

Catalyst: Mozaic 4 ramp Q4 FY26; raised annual revenue growth target to "at least 20%"; nearline contract visibility through 2027.

Risk: Same as WDC — HDD is the wrong technology long-term. HAMR delivers margin once, then commoditizes.

Verdict — Add: STX is the cheaper of the HDD pair and the HAMR-transition leader. If picking one HDD name, STX.

SanDisk (SNDK)

WDC's NAND spin-out. Lower margin than HDDs; more cyclical exposure to the memory cycle.

Verdict — Skip: No clear advantage versus Micron. If you want NAND exposure, do it through MU.

Texas Instruments (TXN)

Analog incumbent. Capex cycle (committed above $30 billion over 2023–26) now winding down; FCF inflection ahead.

Q1 2026: revenue $4.83 billion (+19% YoY), EPS $1.68. Analog operating margin 41.7%. Analog segment $3.92 billion (81% of revenue), +22% YoY. Embedded Processing $723 million (15%), +12% YoY. End-market signal: Industrial +30% YoY; Data Center +~90% YoY — TI is now AI-exposed through power management, signal chain, and sensors.

Catalyst: Q2 guide $5.0–5.4 billion (above consensus); $7.5 billion Silicon Labs acquisition closes H1 2027.

Risk: Mature industrial / auto cycle — margin compresses as utilization underwhelms on heavy new fab capacity. AI exposure is real but only ~10% of revenue.

Verdict — Add (defensive sleeve): Slow grower, steady payer, low correlation to AI cycle, with a quiet AI-data-center revenue line that the headlines don't capture. 3–6% sleeve in a diversified semi portfolio.

Analog Devices (ADI)

The premium analog name.

Q2 FY26 (CQ1 2026): revenue $3.62 billion (+37% YoY) record. Gross margin 73.0% non-GAAP (+360 bps YoY). Operating margin ~42% non-GAAP. Industrial $1.80 billion (50% of revenue), +56% YoY — aerospace, defense, automation. Automotive $872 million (24%), +2% YoY with BMS up double-digit for the first time in two years.

Catalyst: Industrial cycle re-acceleration; auto BMS recovery; defense electronics spend rising.

Risk: Industrial cycle could roll over; defense spend politically sensitive; consensus already reflects strong recovery.

Verdict — Hold: Slightly better quality than TXN. Same general thesis. Pick one for the analog sleeve, not both.

Onsemi (ON)

Auto-power semi specialist. SiC market leader for EV powertrain.

Q1 2026: revenue $1.513 billion (above guidance mid), non-GAAP EPS $0.64. Non-GAAP gross margin ~38.5% (expanding sequentially). Automotive $797 million (53% of revenue), +5% YoY — first YoY gain after 7 declining quarters. AI data center power +30% QoQ. SiC Treo platform revenue +2.5× QoQ.

Catalyst: Q2 guide $1.535–1.635 billion; H2 above H1 visibility; AI data center power becoming material.

Risk: SiC pricing under pressure from China entrants. Sony winning back CIS image-sensor share.

Verdict — Watch / Add small: The bottom in auto is in (first YoY gain after seven declining quarters), but the recovery is fragile. Add 1–2% only if you want EV/SiC cycle leverage; otherwise wait for two more quarters of YoY gains.

Microchip (MCHP)

Embedded MCU + analog. Distribution-heavy.

Q4 FY26: revenue $1.311 billion (+35.1% YoY, +10.6% QoQ). FY26 full year $4.713 billion (+7.1% YoY). Non-GAAP gross margin 61.6%. Operating margin 30.6%. EPS $0.57. Distribution inventory below normal; restocking orders accelerating. CY26 H2 growth visibility strong.

Catalyst: Inventory normalization compounding into broad industrial / auto demand recovery.

Risk: Inventory normalization is one-time; post-restock unit growth depends on the underlying cycle.

Verdict — Add (cycle play): The recovery is now visible in the print — +35% YoY is hard to ignore. 2–3% sleeve for embedded / industrial cycle exposure.

Qualcomm (QCOM)

Smartphone modems + auto + edge AI. Apple modem revenue rolling off.

Q2 FY26: revenue $10.6 billion. Non-GAAP EPS $2.65. Non-GAAP gross margin ~56%, operating margin ~30%. Handsets $6.02 billion (57%), −13% YoY. Automotive $1.33 billion (12.5%), +38% YoY — record quarter. IoT $1.73 billion (16%), +9% YoY. Licensing ~$1.3 billion, flat. Forward P/E in the 18–22× range — the cheapest large-cap semi in the universe. ByteDance edge-AI deal signed May 2026 ("biggest AI deal in QCOM history" per management).

Catalyst: Auto record continues; ByteDance edge-AI deal scales H2 2026; Snapdragon X PC traction.

Risk: Apple modem fade is permanent (now ~10% and shrinking — last 5G iPhone modem socket lost). China handset cycle uncertain.

Verdict — Hold (cheap defensive): Real diversification value at a single-digit-plus multiple. The Apple cliff is the cost. 3–5% position for value-oriented portfolios.

Ambarella (AMBA)

Edge computer vision and robotics SoC.

Q1 FY27: revenue $100.4 million (+16.9% YoY). 58.4% GAAP gross margin. Edge AI SoCs = 80% of revenue. Auto record quarter, driven by commercial vehicle AI penetration. Not yet GAAP profitable; EV/sales around 10–12×.

Catalyst: Q2 FY27 guide $105–111 million; long-term customer agreements being signed — "new phase of market development." Commercial vehicle AI ramp continues.

Risk: Subscale versus NVIDIA Jetson and Qualcomm Auto. Lumpy quarterly print. Cyclical consumer security camera exposure.

Verdict — Add (speculative sleeve only): The cleanest public edge-AI / robotics CV pick. Small position only — 1–2% of an aggressive portfolio.

Monolithic Power Systems (MPWR)

Power management ICs for AI servers.

Q1 2026: revenue $804 million (+26% YoY). Adjusted EPS $5.10 (+26% YoY). Enterprise Data segment +97.7% YoY — AI accelerator power, server VRMs, optical module power. Enterprise Data growth floor raised from +50% to +85% YoY; manufacturing capacity target raised to $6 billion. Q2 guide $890–910 million. Gross margin showed a step-down in Q1 — possibly mix, possibly competitive share-back from Renesas / Infineon.

Catalyst: NVDA Blackwell / Rubin power-stage content.

Risk: NVDA platform concentration is now extreme. The Q1 GM step-down is the early sign that competitors are pushing back.

Verdict — Add: Picks-and-shovels for AI server power. Direct correlation with NVDA platform shipments. Watch the gross margin trajectory closely.

Astera Labs (ALAB)

Connectivity / retimers for AI accelerator interconnect.

Q1 2026: revenue $308.4 million (+93% YoY, +14% QoQ). Non-GAAP gross margin around 74%, operating margin 36.2%. PCIe Gen 6 contributed above one-third of Q1 revenue. Top 3 customers likely 60–70% of revenue. NVLink Fusion design partner — locked at silicon level on Blackwell / Rubin platforms. Forward multiple 80–100× — high-growth specialty connectivity premium.

Catalyst: Scorpio X-Series 320-lane fabric switch ramp; NVLink Fusion scale-up content per system; PCIe 6 broader adoption.

Risk: Single-thesis stock — if AI fabric topology shifts (e.g., direct optical from chip), ALAB is the displaced layer. Valuation leaves no margin for stumbles.

Verdict — Watch: Real catalyst, real growth, real moat — but priced for perfection. Wait for a 30%+ pullback before sizing meaningfully.

Credo Technology (CRDO)

Active electrical cables and SerDes for AI data center interconnect.

Q4 FY26: revenue $437 million (+157% YoY). FY26 full year $1.3 billion (+206% YoY). Non-GAAP gross margin 68.3% — at the top end of fabless semi. Top customer (Microsoft AEC) reportedly above 40% of revenue; top 3 around 70%. Optical DSPs + SiPho + Zero-Flap Optics expected to deliver above $600 million in FY27.

Catalyst: Optical DSP ramp + new SiPho transceivers diversify away from AEC concentration. 1.6T product cycle 2026–27.

Risk: Hyperscaler customer concentration is extreme. If Microsoft AEC ramp plateaus, FY27 optical contribution must come through cleanly. Margins are at the peak.

Verdict — Add (speculative sleeve): Pure-play AI interconnect at the cable layer with optical optionality. Smaller position than ALAB. 1–2% of an aggressive portfolio.

Layer 5 top picks: TXN as the analog defensive (with a hidden AI-data-center line); STX for HDD AI nearline at a cheap multiple; MPWR for AI server power; AMBA for the speculative robotics sleeve.

Layer 6 — Infrastructure and power

Every NVIDIA Blackwell rack needs power, cooling, switchgear, and the grid behind it. Microsoft disclosed an $80 billion Azure backlog it cannot fulfill — because of power. Power is now the binding constraint, not chips. This layer has the strongest read on that. Meta's $125–145B capex and Alphabet's $460B cloud backlog are demand anchors that will either be confirmed or raised in Q2 earnings.

Vertiv (VRT)

Power and cooling for AI data centers.

Q1 2026: net sales $2.65 billion (+30% YoY, 23% organic). Backlog $15 billion (approximately 80% YoY growth). Book-to-bill approximately 2.9× — nearly three dollars of orders for every dollar shipped. Americas revenue $1.81 billion (~70% of total), growing 44% organic. Adjusted operating margin 20.8% (+430 bps). 2026 guide raised: organic sales +29–31%, adjusted EPS $6.30–$6.40, net sales $13.5–14.0 billion. Forward P/E around 54×.

Catalyst: Liquid cooling penetration accelerating (Blackwell systems require it).

Risk: Forward multiple has re-rated meaningfully. Margin compression at scale. Competition from Schneider, ABB, Eaton.

Verdict — Core buy: The cleanest "AI capex happens, this happens" play that isn't a chip. Don't pay chip-multiples for it — size like a quality industrial with AI optionality.

Equinix (EQIX)

The premium data center REIT — network neutrality + interconnection.

Q1 2026: revenue $2.444 billion (+10% YoY), EPS $4.20. Adjusted EBITDA margin 51%. Record $378 million annualized gross bookings. FY26 guide $10.12–10.22 billion — first $10 billion+ year. Customer concentration is exceptionally diversified — top 10 customers around 17% of revenue, no single customer above 5%. 8 of top 10 AI model providers + 4 of top 5 neo-clouds are expanding with EQIX. Forward P/AFFO ~25–28×.

Catalyst: AI inference workloads spreading across more enterprise customers; Fabric interconnect ecosystem.

Risk: Hyperscaler self-build for above-100MW campuses; cap-rate sensitive; the interconnect moat doesn't help on pure compute build leases. REIT tax treatment for Indian residents differs from equities — consult your CA before sizing.

Verdict — Add (REIT sleeve): Real beneficiary of the data center build, dividend income, and unusually diversified customer mix. Indian tax treatment is the asterisk.

Digital Realty (DLR)

Larger, more wholesale data center REIT than EQIX.

Q1 2026: revenue $1.64 billion (+16% YoY). Core FFO $2.04/share (+15.3%). Record $707 million bookings, including the largest-ever single MW lease. Top 20 customers above 50% of revenue (Microsoft, Oracle, Meta among publicly disclosed). FY26 guide $6.65–6.75 billion.

Catalyst: AI deal pipeline robust; HD Hyperscale joint venture monetization.

Risk: Higher hyperscale exposure than EQIX = more concentration risk; cap-rate sensitive; large new builds dilute FFO/share near-term.

Verdict — Hold: Faster topline growth than EQIX but more concentration. Pick one for REIT exposure; EQIX edges on diversification, DLR on growth.

Eaton (ETN)

Electrical equipment — switchgear, UPS, transformers — for AI infrastructure. Plus diversified industrial portfolio.

Q1 2026: revenue $7.45 billion record (+17% YoY, +10% organic). Adjusted EPS $2.81. Electrical Americas rolling 12-month orders +42%; AI data center orders +240%. Data center backlog 228 GW — about twelve years of build at 2025 run rate. Aerospace orders +13%. eMobility to be spun off Q1 2027. FY26 organic growth raised +200 bps to ~10% midpoint; EPS guide raised to $13.05–13.50.

Catalyst: $11 billion of Q1 acquisitions (Boyd Thermal + Ultra PCS); Mobility spin-off Q1 2027.

Risk: Multiple already prices a long AI cycle; aerospace cycle dependent on commercial OEM; integration risk on Q1 acquisitions.

Verdict — Add (core for non-chip AI exposure): AI infrastructure beneficiary with diversification cushion. Lower volatility than VRT for similar theme. The 228 GW backlog is the cleanest single data point in the entire layer.

Quanta Services (PWR)

Utility-grade electrical construction — grid build-out, transmission, renewable connectivity.

Q1 2026: revenue $7.9 billion. Adjusted EPS $2.68. Adjusted EBITDA $686 million. Record backlog $48.5 billion (vs $44.0 billion at YE25). Raised FY26 revenue $34.7–35.2 billion, EPS $13.55–14.25.

Catalyst: US grid interconnect projects; data center transmission line construction; substation pipeline scaling.

Risk: Construction-services margins limited; project execution risk; ROIC structurally lower than chip names; cyclical to utility capex.

Verdict — Add: The grid behind the data center. Highest leverage to "data centers require grid" thesis. 2–4% sleeve.

Vistra Energy (VST)

Independent power producer with nuclear (Comanche Peak) plus the Cogentrix acquisition.

Q1 2026: revenue $5.64 billion (+43.4% YoY). Net income $1.03 billion (turnaround from −$268 million Q1 2025). Adjusted EBITDA $1.494 billion (+20% YoY). Meta signed a roughly 2,600 MW PPA at PJM nuclear; AWS PPAs in place. Forward P/E around 18×. FY26 EBITDA guide $6.8–7.6 billion reaffirmed (excludes new deals).

Catalyst: Cogentrix close (5,500 MW gas); Meta + AWS PPAs incremental to base case.

Risk: Power demand case "not yet priced in" per bulls — but stock has already moved sharply. Capacity-market pricing peaks. If PPA signings pause, sentiment can turn.

Verdict — Add: Power has become the binding constraint on AI buildout. VST is leveraged to power prices and nuclear monetization. Real catalyst, real risk — 2–4% sleeve.

Constellation Energy (CEG)

Largest US nuclear fleet (~22 GW pre-Calpine) plus Calpine's 23 GW of gas added in 2026.

Q1 2026: revenue $11.12 billion (+64% YoY). Adjusted EPS $2.74 (vs $2.14 PY). GAAP EPS $4.49 (vs $0.38 PY). Long-term PPAs with Microsoft (Three Mile Island Unit 1 restart 2028), Meta, and others — above 5,650 MW of long-term clean energy hyperscaler contracts. Forward P/E around 24–25×. 20% EPS growth target FY26.

Catalyst: Calpine integration synergies; hyperscaler PPA pipeline continues; restart timeline visibility (Crane / TMI).

Risk: Nuclear capex inflation; PPA contracted volumes already material — incremental signings get harder. Calpine integration brings gas/dispatch risk into the "clean" thesis.

Verdict — Add: Less leveraged than VST but cleaner exposure to nuclear baseload for data centers. 2–4% sleeve.

GE Vernova (GEV)

Spun-off GE power generation. Gas turbines, wind, grid.

Q1 2026: revenue $9.3 billion (+16%, +7% organic). Orders $18.3 billion (+71% organic). Backlog $163 billion. Electrification (grid + power conversion): data center equipment orders $2.4 billion in Q1 alone — more than all of 2025. Power segment gas slot reservations at 100 GW (target 110 GW by year-end 2026). Forward P/E in the 58–69× range — the most extreme multiple in this layer. FY26 revenue guided $44.5–45.5 billion; FCF $6.5–7.5 billion.

Catalyst: Gas turbine demand for AI data center peaking power; grid orders compounding; "$200B backlog by 2027" — a year ahead of prior plan.

Risk: Multiple is the most extreme in the AI power triumvirate. Wind drag is real if offshore continues to slip. Gas turbine supply tightness is a 2027–28 story — until then revenue is order-recognized vs delivery-recognized.

Verdict — Hold: Real backlog story but the multiple has re-rated meaningfully. Don't chase. Existing holders stay.

Coherent (COHR)

Vertically integrated photonics — InP epi, EUV laser source (sole supplier to ASML for High-NA), 800G/1.6T transceivers, optical circuit switches.

Q3 FY26 (CQ1 2026): revenue $1.81 billion (+21% YoY, +27% pro forma). Non-GAAP gross margin 39.6%. Non-GAAP EPS $1.41 (+55% YoY). Datacenter + Communications $1.4 billion (77% of revenue), up from $1.0 billion. NVIDIA disclosed a multi-year supply agreement plus a $2 billion equity investment in Coherent — now a structural anchor customer. Q4 FY26 guide $1.91–2.05 billion.

Catalyst: NVDA multi-year supply deal scales; OCS (optical circuit switch) volume.

Risk: Industrial laser business shrinking; transceiver pricing eventually compresses; Lumentum is a direct competitor with the same NVDA investment structure.

Verdict — Add: Broadest exposure of the optical names — EUV laser business + datacom + photonics. NVDA equity stake is structural. Pick this if picking one optical name.

Lumentum (LITE)

Optical transceivers, OCS, laser chips. The other side of the NVDA optical-investment double-down.

Q3 FY26 (CQ1 2026): revenue $808.4 million record (+90% YoY). Non-GAAP gross margin ~37%. GAAP EPS $1.50; non-GAAP EPS approximately $2.50. Cloud + AI above 70% of revenue. NVIDIA $2 billion equity investment plus multi-year supply agreement signed March 2026. Q4 FY26 guide $960 million – $1.01 billion (+85% YoY). OCS backlog above $400 million; CPO order H1 CY27 in multi-hundred-million dollar range.

Catalyst: OCS backlog converting; CPO (co-packaged optics) transition; 1.6T product cycle.

Risk: NVIDIA dependency now extreme. Transceiver pricing power has expiry. CPO transition could disintermediate transceivers entirely.

Verdict — Add (speculative): Higher AI growth than COHR but more concentration. If you want pure-optical AI exposure with the cleanest growth rate, LITE.

Ciena (CIEN)

Optical networking systems integrator. WaveLogic 6e coherent DSP — leadership at 800ZR+ / 1.6T over distance.

Q1 FY26: revenue $1.43 billion record. Q2 FY26 results already reported (June 2026) — the near-term binary has resolved. Watch the Q3 guide for DCI and AI training cross-DC connectivity scaling visibility.

Catalyst: DCI and AI training cross-DC connectivity scaling; enterprise optical ramp.

Risk: Carrier capex remains lumpy; cloud orders concentrated in a handful of contracts; consolidation pressure from Cisco/Acacia and Infinera-Nokia merger.

Verdict — Add (post-print): Q2 print has resolved the near-term binary. If Q2 beat and the guide is constructive, this transitions from Watch to Add. Pull the June print before sizing.

Layer 6 top picks: VRT as the core; ETN for non-chip AI exposure with a 228 GW backlog; EQIX for the REIT sleeve; VST + CEG for power; COHR as the optical pick.

Three model portfolios

Sketches, not advice. Position weights are within the semiconductor allocation — not your full portfolio. Sizing semi exposure inside your overall asset mix is the question above this one. A widely used institutional heuristic caps single-sector ETF / thematic exposure at 15% of an equity portfolio, single names at 5%, combined sector exposure at 20%. For a retail Indian-resident portfolio, semi-only exposure typically maxes at 15–25% of the equity sleeve; within semis, no single name above 30–40% of the sleeve.

Historical drawdown context (SOXX and SMH):

Period	SMH	SOXX	Recovery
2022 (full year)	−33.5%	−35.1%	~14 months
2018	~−13%	~−16%	~12 months
2008–09 GFC	—	−52%	~6 years
2000–02 dot-com	−85%	−70%	Multi-year+

Single names draw down harder — NVDA −65% in 2022, AMD −60%, MU −50%.

Defensive: capital preservation, structural moats only

For investors who want AI infrastructure exposure but cannot tolerate a 50% drawdown in a single year. Expected max drawdown roughly −25 to −35% in a serious correction based on the framework precedents (4–6 names, top weight 30–35%, no name under 10%, all names with 15-year track record of margin discipline or above-50% market share in a moat layer).

Layer	Name	Weight
Equipment	ASML	22%
Equipment	KLAC	10%
EDA	CDNS or SNPS	12%
Foundry	TSM	12%
AI compute	NVDA	14%
AI compute	AVGO	12%
Analog defensive	TXN	8%
AI power	CEG	5%
Cash buffer	—	5%

Logic: equipment + EDA + analog = 52% of the sleeve, in the most insulated tiers. TSM sized down for Taiwan tail risk. NVDA + AVGO as the barbell at moderate weights. CEG as the non-chip AI infra hedge. 5 cash buffer to deploy on drawdown.

On Rs 25 lakh deployed: roughly Rs 5.5L ASML, Rs 3L TSM, Rs 3.5L NVDA, Rs 3L AVGO, Rs 2.5L KLAC, Rs 3L CDNS/SNPS, with the rest spread. LRS budget: well within the ₹10L investment threshold per FY in a 25–50 lakh annual deployment cadence.

Balanced: mix of core compounders and opportunistic exposure

For investors who want broader participation and can tolerate roughly −35 to −45% drawdown (8–12 names total, largest position 15–20%, smallest 3–5%, mix of 60% core blue-chip + 25% mid-cap opportunistic + 15% thematic).

Layer	Name	Weight
Equipment	ASML	14%
Equipment	KLAC	5%
Equipment	AMAT	4%
EDA	CDNS	7%
EDA	SNPS	5%
Foundry	TSM	10%
AI compute	NVDA	12%
AI compute	AVGO	11%
AI compute	AMD	5%
AI compute	MRVL	4%
Networking	ANET	5%
Memory	STX (HDD cheap)	3%
Infrastructure	VRT	5%
Infrastructure	ETN	4%
Infrastructure	EQIX	3%
AI power	CEG	3%

Logic: still ~30% equipment + EDA. AI compute broadens to include AMD + MRVL (hyperscaler diversification) and ANET (networking). Memory exposure capped via STX (cheaper than MU). ETN added for the 228 GW backlog signal.

Aggressive: maximum AI capex leverage

For investors with conviction in continued AI infrastructure build through 2028 and tolerance for −50 to −65% drawdowns (5–8 names, largest position 25–35%, all triple-digit YoY growth or prime AI-capex beneficiary).

Layer	Name	Weight
AI compute	NVDA	22%
AI compute	AVGO	16%
AI compute	AMD	8%
AI compute	MRVL	6%
Networking	ANET	8%
Equipment	ASML	8%
EDA	CDNS	4%
Foundry	TSM	6%
Infrastructure	VRT	8%
AI power	VST	4%
AI power	CEG	3%
Speculative	AMBA / ALAB / CRDO	3% combined
Optical (speculative)	COHR	4%

Logic: 60% in AI compute + networking. 18% in equipment + EDA + foundry as the only "insulated" allocations. Power leg via VST + CEG. Speculative sleeve for asymmetric upside.

What not to chase at these levels

Categories that look like AI plays but carry more risk than the multiples compensate for:

Pure-play AI software without semi exposure — multiples expanded with no structural silicon moat.
China-listed semis ADRs (SMIC, etc.) — SMIC's ADRs were withdrawn from OTCQX on January 31, 2026 following a November 2025 executive order. Listing risk is real.
Intel at a triple-digit forward multiple — turnaround fully priced.
Micron at peak-cycle 81% gross margin — memory cycles invert fast.
Anything trading on a momentum chart without earnings to back the move — the run-up heading into earnings compressed many names. Those without contracted backlogs will compress first on any sentiment turn.

The risk scenarios you should price in

Four scenarios that materially change the picture. Each has different exposure consequences.

Taiwan crisis (semi-supply disruption). Highest tail risk in the stack. Direct revenue concentrated at TSM (above 70% global foundry, above 90% leading-edge). CSIS's wargames (2023 invasion study, 2025 "Lights Out" blockade study with 26 wargames — two of which spiraled to general war) frame the scenario but not the probability. Modal analyst probability of a major Taiwan event by 2030 is roughly 10–25%; dispersion is wide. De-risking: under-weight TSM, over-weight ASML (sells to all foundries), over-weight equipment makers (post-event, capex accelerates in non-Taiwan capacity). Add domestic-fab plays (GFS, INTC). Power and infrastructure layers largely unaffected by Taiwan supply (they are downstream demand stories).

China escalation (export controls tighten). The January 2026 BIS final rule shifted advanced-computing semiconductor licensing to case-by-case from presumption-of-denial. After April 13, 2026 the 180-day authorized-IC-designer designation became operational — effectively forcing recertification of every China-customer designer/integrator. Likely next moves (Q3/Q4 2026): closing Malaysia / Singapore / UAE re-export gateways; lower H20/H200-equivalent specs ceiling; deeper FDPR. De-risking: avoid ACMR, prefer US-domiciled equipment and AI-compute names with diversified customer bases.

AI capex pause (hyperscaler ROI questions). The scenario where Microsoft, Google, or Meta pulls back 2027 capex guidance. The Q2 earnings calls in late July are the first real read — watch for any language shift from "compute constrained" to "balanced supply and demand." Six leading indicators analysts track, organized from earliest to latest:

ASML net bookings (12–18 months ahead) — currently €38.8B / €7.4B EUV, no slowdown signal.
TSMC monthly revenue / leading-edge ramp commentary — June 2026 monthly data tracked above +30% YoY guide; no slowdown.
Hyperscaler language shift from "compute constrained" to "balanced supply and demand" — all Big-4 still saying compute constrained going into Q2 prints (MSFT $80B Azure backlog the strongest data point).
Capex/revenue intensity — currently 25–35% at the Big-4, peak conviction.
NVDA / AMD data center revenue QoQ deceleration — neither showing two consecutive QoQ declines; Q2 FY27 print August 20–22 is the next data point.
ROI per AI revenue dollar — Jefferies put the 2025 ratio at roughly 10 cents of AI revenue per dollar of capex. Needs to improve through 2027.

De-risking if signals turn: trim NVDA + AVGO; rotate to power, data center REITs, and equipment (already-contracted multi-year deliveries). The chips are most cyclical at the moment of pause; the picks-and-shovels keep shipping through it.

Memory cycle inversion. When DRAM/NAND supply catches demand. Direct exposure: MU. Indirect: LRCX (memory equipment). Signals to watch: DRAM inventory weeks (currently 3.3 weeks — extreme tightness; flip-trigger is above 8 weeks); HBM contract pricing rolling over (currently HBM3E lifted 20% YoY for 2026 — opposite direction); hyperscaler memory order push-out commentary (none in Q1 2026 prints); PC + smartphone unit growth combined +5% YoY (both currently flat-to-down). All three preconditions for cycle peak — peak YoY pricing, trough inventory, stock-price inflection — are not met as of Q2 2026.

How an Indian-resident investor actually executes this

Practical layer:

Access. All names listed on NYSE or NASDAQ are accessible through Vested, INDmoney, IBKR, and Rovia. The structural cost differences are covered in Vested vs INDmoney vs IBKR vs Rovia — for a portfolio deploying Rs 25 lakh+ per year, IBKR's FX cost edge becomes material; for under Rs 10 lakh per year the Indian-aware platforms cost less in absolute terms.

LRS limits and TCS. LRS limit is USD 250,000 per FY per individual. TCS on LRS for investment purposes: 20% above ₹10 lakh per FY (education / medical / tour purposes use ₹7 lakh thresholds with different rate schedules). For a Rs 25 lakh annual deployment, plan for roughly Rs 3 lakh of TCS that you'll reconcile against your tax liability at year-end. Form 12BAA (Finance Act 2024, effective October 2024) lets salaried employees declare TCS to their employer for reduced monthly TDS, compressing the recovery timeline from 12–18 months to within the same FY.

Tax on gains. Foreign equity LTCG = 12.5% under Section 112 with a 24-month holding period. STCG = slab. Indexation removed for foreign equity post-Budget 2024. The 24-month holding period is real — selling at day 720 vs day 740 can cost ₹50k–₹2 lakh on a meaningful gain.

Schedule FA. Any of these names held at any point during the calendar year (January–December — Schedule FA is the one ITR schedule that doesn't use the financial year) triggers a Schedule FA disclosure obligation. Peak value, closing value (December 31), dividends, gross sale proceeds all need to be reported. Penalty for non-disclosure under the Black Money Act 2015: 30% tax + 3× penalty + 3–10 year prosecution risk. See the Schedule FA disclosure guide and the Schedule FA helper tool for the field-by-field walkthrough.

Dividend treatment. Most names here pay token dividends or none. AVGO, TXN, ADI, QCOM, EQIX, DLR are the meaningful payers. US dividend withholding is 25% under the India-US DTAA with a valid W-8BEN; the credit flows through Form 67 in your Indian ITR (renumbered Form 44 from Tax Year 2026-27).

Currency exposure. When you buy any of these you're long USD assets against the rupee. The 24-month holding period for LTCG also locks in USD-INR exposure for that window. Build that into position sizing — USD-INR is not a free option.

The closing

Combined Big 5 hyperscaler capex tracking roughly $660–690 billion in 2026 — about 75% AI-specific. Google Cloud backlog $460 billion. Microsoft Azure unfilled backlog $80 billion. AVGO AI backlog $73 billion. ETN data center backlog 228 GW. PWR backlog $48.5 billion. GEV backlog $163 billion. ASML order book €38.8 billion. NVIDIA shipped $75.2 billion of data center revenue in a single quarter.

These aren't forecasts. They're contracts.

Earnings season will tell us the rate of conversion — not whether the contracts exist. The order books were real in Q1. The Q2 prints arriving July 22 through August 22 will either confirm that pace or introduce the first wobble. Watch the hyperscaler capex language in late July before NVIDIA's own print on August 20–22. That sequence — Alphabet and Microsoft on July 28–30, Meta on July 29–31, then NVIDIA six weeks later — is the read-through chain for the rest of 2026.

Buying before an earnings print is timing the chart. Understanding what the order books say before the print is understanding the business.

The forums see the chart. The order books are already in 2028.