How we pick leaders — in 30 seconds
We rank market themes by relative strength, find the strongest names inside the leading themes, and verify every quote and return against live market data at publish time. Each name gets one call: Buyable (at a buy point) · Basing (watch for a breakout) · Wait (strong, no setup yet) · Extended (overbought, await a pullback) · Too early (unconfirmed bounce) · Avoid (downtrend). Relative strength is a descriptive, backward-looking signal — not a forecast.
Market regime
SPY (~$759.55) and QQQ ($746.16) both printed fresh 52-week highs. But the cap-weighted headline badly overstates breadth.
Rank themes by measured relative strength (distance from the 52wk high, separation from the 50/200DMA) instead of YTD-return headlines, and only two groups are genuine present-tense leaders with deep MA separation. First is AI custom-silicon, connectivity and memory: SMH just hit a fresh high, up 75.5% YTD, and AVGO, MU, MRVL and LRCX are all at or near highs.
Second is copper: COMEX is at all-time highs around $6.56-6.71/lb, and FCX and TECK printed fresh 52wk highs above all their MAs. Copper is the cleanest non-AI confirmed leader, because both the metal and the equities agree.
Capital is rotating within risk assets, not buying broad beta. On the same tape, mega-cap platforms, fintech and crypto are being sold even on green-index days.
GOOGL fell 3.9% on its own $80B raise. COIN dropped 4.7% and sits 61% off its high.
IGV software is lagging. Here is where the popular narrative breaks down: the marquee 'hard-asset' and 'power-for-AI' pillars do not hold up as live leaders.
Energy owns the 2026 YTD crown (XLE ~+27%), but it sits below its 50DMA and ~8.7% off its high. That is a crude/Hormuz price artifact, not equity leadership.
Utilities and IPPs are technically broken. VST is 28% off its high and below its 200DMA.
CEG is 34% off its high after a fresh $3.1B dilutive raise. XLU is below both MAs.
Gold is in a genuine consolidation: GLD is 19% off its high, GDX is 25% off, and AEM is below both MAs. The AI-power demand thesis is real.
But it is being expressed through the names that actually trend: copper, electrical equipment (VRT/ETN) and Cameco, not the broken utility and IPP equities. The main hazard is extension.
The highest-relative-strength names are also the most fundamentally stretched, which is a classic late-stage tell. MRVL ran +32.5% in one session and MU +84% in a month.
The cleanest risk/reward sits one derivative out, in basing or just-breaking quality names, not in the parabolic spearheads.
Gold is in a genuine consolidation: GLD is 19% off its high, GDX is 25% off, and AEM is below both MAs.
Theme strength & rotation
Each theme’s leadership strength (0–100, from its leaders’ proximity to 52-wk highs, relative strength vs the proxy, and momentum), tracked across issues. ▲ strengthening · ▼ rolling over · ● holding · ✦ new. The number is the change vs the last issue this theme appeared. A theme that is rising but not yet at the top is where leadership may be emerging.
Leadership map
AI custom-silicon, connectivity & accelerators (semiconductors)
AI memory / HBM super-cycle
What’s driving it. This is a binding-contract supply shortage. On Micron's Q1 FY26 call, the company said its entire CY2026 HBM capacity is sold out.
Gross margin jumped from 56% to 74% QoQ on pricing power. Micron's FY26 capex step-up flows directly into the semicap order book, and LRCX has the highest memory mix among US large-cap semicap names.
The driver is contracted demand, not pure momentum.
Durability. The bottleneck is real and lasts multiple quarters. But memory is structurally cyclical and commodity-like.
The same sold-out tightness reverses hardest when supply catches up, with new Idaho/Taiwan/Korean capacity arriving from H2-2027. MU's price action is a blow-off: +84% in one month, roughly 3x its 200DMA.
It will mean-revert violently on any AI-memory inventory digestion. Durable theme, dangerously extended spearhead.
The picks-and-shovels name, LRCX, carries the cycle with less single-name parabola risk.
Leaders’ fundamental health. Mixed-to-strong, with valuation the swing factor. LRCX is the cleanest and rated Strong: 54% ROE, 34% ROIC, ~4.3% FCF yield, net cash, ~23x earnings.
MU is genuinely inflecting, with GM moving 22%->40%, EBITDA $8.9B->$18.5B, net cash and modest leverage. But it is Stretched on a ~$1.2T cap off trough-cyclical earnings, at ~140x trailing P/E and ~1.2% FCF yield.
TER is the highest-margin name (~59% GM, net cash) but Stretched at ~40x EV/EBITDA and ~112x P/E for only ~13% revenue growth. Aggregate: strong balance sheets, real earnings inflection, prices that fully discount the up-cycle.
Copper — AI/electrification structural leader
What’s driving it. A verified live supply shock meets structural demand. COMEX copper is around $6.56/lb, with a record $6.71 on May 13.
Add the Grasberg Block Cave force majeure, an ICSG-forecast 2026 refined deficit of ~150kt, and Chile's weakest April output in 23 years. On the demand side, AI data centers and grid electrification.
FCX rose 7% on the day, the marginal price-setter rising on its own lost volume. TECK is re-rating on the QB2 copper ramp plus a coal-sale-funded buyback.
Durability. This is more sustainable than the other real-asset themes. The supply constraints are physical and multi-year, with mine depletion and 10+ year project lead times, and demand is structural.
The near-term caveat is entry, not thesis: FCX and TECK are extended into fresh highs after vertical sessions and 3-month runs. The metal-and-equity confirmation makes this higher-quality leadership than energy or gold.
Leaders’ fundamental health. Wide quality dispersion, and the price leader is the fundamentally weakest. SCCO is the quality anchor, rated Strong: 57% GM, 32% net margin, 39% ROE, 23% ROIC, net-debt/EBITDA of 0.39x.
But it is a 3-month price laggard, catching up rather than front-running. FCX is Mixed: cheap at 9.3x EV/EBITDA, but single-digit net margin, ROIC under 8%, and a thin 1.5% FCF yield on heavy capex.
TECK has the strongest price RS but is Weak on cash, with FY25 FCF negative at -C$966M, ROE of 5.6% and ROIC of 2.8%. Its thesis rests on forward QB2 and buyback, not trailing numbers.
ERO is a Mixed high-beta satellite, swinging from an FY24 loss to an FY25 turnaround with the thinnest balance sheet. Aggregate: solid balance sheets but capex-heavy, with cash generation thin-to-negative at the high-RS end.
Data-center physical buildout — cooling & electricals
What’s driving it. Hyperscaler data-center capex turned into physical gear: thermal management, power distribution and grid EPC. VRT is the tell.
It is up 106.5% YTD and ~58% above its 200DMA, on company-reported Q1 net sales +30% YoY and a backlog up sharply. The order velocity is the catalyst, and it is distinct from the broken power-generation proxies.
Durability. Sustainable as long as hyperscaler capex holds. These are the spec-locked beneficiaries of the same buildout funding the silicon.
The risk is a capex air-pocket and rich valuations that leave no error margin. Within the group, VRT, ETN and PWR are trending, digesting within an uptrend.
GEV has the weakest near tape, with the worst 1m at -9.8% and a broken 50DMA. This is more durable than the IPP/utility expression because backlog visibility is hard-contracted.
Leaders’ fundamental health. These are strong businesses, but the valuations run rich to stretched. VRT is strong: ROE 33.8%, ROIC 18.5%, positive FCF, net-debt/EBITDA 0.76x.
The catch is EV/EBITDA at ~29x. ETN is strong too, and the most reasonably valued at P/E 30x with ROE 21%.
But it's the slowest grower at +10.3%. GEV is stretched.
It carries the richest multiple at ~46x EV/EBITDA and the lowest ROIC at 6.3%. Its FY25 GAAP was flattered by a ~$2.05B tax benefit, though it sits on net cash.
PWR is stretched as well. The EPC moat is indispensable but the model is structurally thin: net margin 3.6%, ROIC 7.1%, all at 62x P/E.
Bottom line: real FCF and mostly clean balance sheets. But every name is a momentum valuation, not a deep-value entry.
Broad tech / Nasdaq mega-cap melt-up (cap-weighted index)
What’s driving it. The index sits at new highs, with QQQ up +21.5% YTD. But the alpha lives entirely in the silicon and memory sub-themes.
The relative leaders versus QQQ are MU (+251pp), MRVL (+221pp) and AVGO (+18pp). NVDA, the cap anchor at ~$5.4T, actually lags the index it dominates, at -2pp YTD.
Durability. As a breadth story, this is fragile in the short term. Leadership rests on a handful of names, and software (IGV) isn't participating.
Own the index only as a beta wrapper. The durable alpha is in the silicon, memory, copper and buildout sub-themes, not the cap-weighted index itself.
If AI-capex expectations reset, the most-owned names get hit hardest.
Leaders’ fundamental health. It's the same split as the silicon theme. The highest-quality balance sheet is the relative laggard.
That's NVDA: net cash, 30% ROE, 85% YoY revenue growth, and cheapest on PEG at ~0.63. Meanwhile the relative leaders are the most stretched.
MRVL is weak, with GAAP ROE 0.2%, net margin 1.4%, and a pure-momentum multiple. MU is strong but parabolic.
AVGO is levered. Franchise-level health is genuinely strong.
But the price leadership has detached from the quality leadership.
Energy / oil & gas (YTD-return crown, present-tense laggard)
What’s driving it. The YTD number is really a crude-price artifact, driven by the Brent risk-premium on Hormuz and Iran supply fears. The recent FANG analyst PT raise to $240 was tied explicitly to higher oil-price assumptions.
So it's a macro tailwind, not idiosyncratic leadership. That said, the two best operators, FANG and EOG, do still lead within the sector on their own RS.
Durability. This is short-term and macro-dependent. It's a YTD-return story that has lost present-tense momentum.
It holds only as long as the geopolitical crude premium does. Forecast dispersion is a late-cycle tell, with the street at ~$60 versus ~$90-100 Brent.
Don't add at the index level. The only durable expressions are the two single-name quality leaders that hold RS independent of the sector.
Leaders’ fundamental health. Quality varies widely here. EOG is strong and the fundamental standout: ROIC 58%, net-debt/EBITDA 0.44x, cheapest at 5.5x EV/EBITDA / 11.4x P/E, and a 6.9% FCF yield.
Its -3.5% revenue decline is realized-price, not a franchise problem. FANG is mixed.
It has a 12% FCF yield but acquisition-inflated growth, compressed GAAP EPS on impairment, and leverage stepped to 2.0x after Endeavor. VNOM is weak, posting an FY25 GAAP net loss and negative FCF on acquisition capex.
Net: balance sheets are healthy and FCF is strong at the operators. But earnings quality and the sector tape are deteriorating.
Gold & precious metals
What’s driving it. There's a structural floor here from central-bank buying, at 863t in 2025, plus safe-haven demand against the narrow AI melt-up. But the price action is a pause within an uptrend, not current leadership.
The streamers and turnaround names lead the GDX proxy and hold their 200DMAs: FNV, WPM and NEM. The producer AEM lags badly.
Durability. This is sustainable as a multi-year structural bull, thanks to the central-bank floor. But it's not present-tense leadership.
Wait for a 50DMA reclaim before re-calling it. Own royalty and streaming for margin resilience through the consolidation.
Producers carry more downside torque if gold breaks the base lower. The theme is genuinely durable, just in a digestion window right now.
Leaders’ fundamental health. Quality runs from strong to stretched, and the cheapest name screens best. NEM is the standout value.
It's strong: cheapest at P/E ~16x / EV/EBITDA 8.3x, highest ROE at 20.9%, best FCF yield at 6.6%, and net cash after a genuine balance-sheet turnaround. AEM is fundamentally excellent, and also strong: record FY25 margins of net 37% / EBITDA 70%, net cash, ROE 18%, PEG 0.14.
But it's the price laggard, so it's the right fundamentals on the wrong tape. The streamers FNV and WPM are stretched.
They're capital-light with 82-95% EBITDA margins and fortress balance sheets. But they trade at P/E in the mid-30s, premium multiples, with thin FCF yields on upfront stream payments.
In aggregate: fortress balance sheets, record margins, premium valuations on the streamers, and value on the producers.
Uranium & nuclear / SMR (power-for-AI)
What’s driving it. This is driven by narrative and deal-flow, not spot prices. Think hyperscaler nuclear PPAs, DOE HALEU contracts and enrichment deals.
Spot U3O8 is flat at ~$83-85 while the equities ran. CCJ carries verified contract power: the India $2.6B/22M-lb deal and a 49% Westinghouse stake tied to a ~$80B US build framework.
The SMR developers and enrichers are recovery and narrative names.
Durability. The demand thesis is structurally durable. But present-tense leadership is speculative and concentrated in a single confirmed name, CCJ.
The commodity isn't confirming, so treat the broader complex as early and speculative. CCJ is the durable leader.
LEU and OKLO are turnaround and pre-commercial bets whose 'leadership' is moat and narrative, not relative strength. Both sit 55-62% below their 52wk highs, while the theme proxy is only -14%.
Leaders’ fundamental health. Mixed-to-weak. CCJ is a real franchise but stretched.
It trades at ~62x EV/EBITDA, and the equity ran while spot stayed flat. Earnings are surging, it holds net cash, and it reports in CAD.
LEU is mixed. It's profitable, with ~$690M net cash and a $900M DOE HALEU order.
But revenue is dead flat, up just +1.5% YoY, and entirely forward government-funded. OKLO is weak.
It's pre-revenue, with FCF of -$51M in Q1 and no deployment until ~2027+. It holds ~$1.6B cash, but that was freshly raised via $1.18B of Q1 dilution.
Its 14 GW pipeline is non-binding LOI grade. In aggregate, only CCJ has trailing earnings.
The rest depend on policy, contracts, and capital.
Power-for-AI generation / utilities & independent power producers
What’s driving it. The structural demand is real. Hyperscaler nuclear PPAs point to ~580 TWh of data-center demand by 2028.
But that demand is ALREADY expressed through the names that actually trend: copper, electrical-equipment (VRT/ETN), and CCJ. It is NOT expressed through the broken IPP equities.
The downtrend was fed by late-May data-center grid and state pushback, plus CEG's $3.1B dilutive secondary block sale.
Durability. Short-term broken, even though the demand thesis is durable. Only two names survive as relative-strength leaders: GEV, which sits above its 200DMA, and TLN, the one IPP above both MAs.
CEG and VST are technically broken pure-plays. Wait for the IPPs to reclaim their 200DMA before re-engaging.
The demand is durable. The current IPP price expression is not.
Leaders’ fundamental health. Stretched-to-weak. GEV is stretched.
It carries the richest multiple at ~46x EV/EBITDA on 6.3% ROIC, and FY25 net income was flattered by a ~$2.05B tax benefit. Its balance sheet is genuinely net-cash.
TLN is weak on GAAP. FY25 showed a net LOSS of -$219M, EPS of -$4.79, and ROE of -20%.
The thesis leans entirely on adjusted EBITDA/FCF and forward FCF guidance, so own it as torque, not quality. CEG is mixed.
The nuclear franchise is quality, but FY25 net income fell -38% YoY, EPS dropped from $11.90 to $7.40, the FCF yield is a thin 1.2%, and there's a fresh dilution overhang. In aggregate, the franchises and backlogs are real, but GAAP earnings quality is poor and valuations rich.
Healthcare & financials (persistent laggards)
What’s driving it. The sector is weak, but a few single names are winning on their own stories. UNH is on a 2026 margin-repair turnaround, with Q1 adj EPS of $7.23, an MCR of 83.9%, and FY26 guidance raised to >$18.25.
LLY won FDA approval for orforglipron/Foundayo on Apr 1, 2026. The popular 'healthcare laggard-to-leader' biotech-rotation narrative is NOT yet price-validated.
XBI is weak and VKTX is broken.
Durability. The sector is a laggard for now, with both below their 200DMAs. The leadership is name-specific and uneven.
LLY's RS leadership is the cleanest and most durable. It sits above both MAs with healthy growth.
UNH's recovery is already extended, up +28% in 3 months. JPM's 'leadership' is overstated.
It's below both MAs and lagging XLF YTD. VKTX is a broken-chart binary, not a leader.
Leaders’ fundamental health. These span the full range. LLY is stretched.
It has best-in-class growth, with revenue up +44.7%, ROE of 77.8%, and ROIC of 30.2%. But EV/S is 15.3x and the FCF yield is <1%, so the price prices perfection.
JPM is strong. ROE is 15.7%, the P/E is 15.8x, CET1 is a fortress, and it's the most diversified bank.
The quality is genuine, but the leadership claim is weak. UNH is mixed.
FY25 fundamentals DETERIORATED: net income fell, net margin was 2.7%, and ROE and ROIC were both down YoY. The bull case rests entirely on forward quarterly repair.
VKTX is weak. It's pre-revenue, the FY25 loss widened to -$359.6M, and ~$165.8M of cash gives only ~0.6yr of runway, with dilution and binary Phase 3 risk.
In aggregate, quality is real at LLY/JPM, recovering at UNH, and absent at VKTX.
Cybersecurity
carried forward
What’s driving it. Fortinet's blowout Q1-2026 sparked a sector-wide AI-security re-rating. Billings rose +31%, product +41%, and OT billings +70%.
AI is both the threat and the tailwind, since security budgets are rising as AI lowers the cost of attack. CRWD and PANW report this week, and a soft print is the proximate risk.
Durability. The short-term move is parabolic and sympathy-correlated. CRWD and PANW each roughly DOUBLED in 3mo, and FTNT is up +86%.
So the IMMEDIATE leg is stretched and pullback-prone. The underlying spend cycle is sustainable.
But the specific 06-01 melt-up was one earnings-driven tide, not four independent breakouts. Durability of the THEME is high.
Durability of the current PRICE is low.
Leaders’ fundamental health. Mixed in aggregate, with a clear quality leader. FTNT is the standout.
It's GAAP-profitable, with FY25 net income of $1.85B, ROIC of ~29%, an 81% gross margin, a ~3.8% FCF yield, and the lowest EV/S at ~8.5x, though its appliance-led revenue is cyclical. PANW is the other profitable name.
ROE is ~14.5%, the FCF yield is ~3%, and EV/S is ~12x, but it carries a ~$25B CyberArk plus $3.35B Chronosphere integration/dilution overhang. CRWD is still GAAP-loss-making at -$162.5M, with the loss WIDENED YoY, on a 22x EV/S nosebleed multiple.
ZS is the broken laggard, down -30.8% YTD, -53.8% below its high, below its 200DMA, and GAAP-loss-making. It is NOT a leader.
All four are net-cash.
Neocloud / AI Datacenter Buildout (GPU-rental + servers)
carried forward
What’s driving it. Vera Rubin NVL72 validation (CRWV, 06-01), HPE's record Q2 FY26 beat ($10.68B rev +40% YoY, biggest beat since 2018), and multi-year hyperscaler capacity contracts (NBIS Microsoft $17B + Meta $27B; CRWV $99.4B backlog; IREN Microsoft $9.7B + NVIDIA $3.4B).
Durability. Short-term momentum is real but fragile — three of four names carry Weak/Stretched balance sheets and the group is a laggard catching up to the extended parent (CRWV still ~33% below its $187 high). The contracts are multi-year, but the leverage (CRWV current ratio 0.31, working capital -$12.2B, Q1 FCF -$4.7B) means this is the highest tail-risk expression of the AI buildout; durability is contract-backed but financially precarious.
Leaders’ fundamental health. Weak-to-mixed in aggregate — the lowest-quality theme in the AI complex. CRWV is Weak (current ratio 0.31, WC -$12.2B, Q1 FCF -$4.7B, D/E 3.7x, EV/S ~27x).
NBIS is Stretched (EV/S ~68x, Q1 GAAP 'profit' is ~$743M non-operating other income masking a -$128M operating loss) but has the best balance sheet (current ratio 8.3x, near-net-cash). IREN is Weak (FCF -$0.92B/qtr, beta 4.18, revenue DECLINING sequentially $240M→$185M→$145M as the mining base winds down faster than AI ramps).
HPE is the only FCF-positive name (Mixed: EV/S ~5x, P/E ~16x, ~5.2% FCF yield) but thin AI-server margins.
Enterprise Software (AI monetization)
carried forward
What’s driving it. Agentic-AI monetization (ServiceNow Now Assist ACV target raised to $1.5B from $1B; PLTR revenue +84.7% YoY; Salesforce Agentforce $1.2B ARR +205%) plus a sector-wide software rally (IGV +5.9% on 06-01). The 'software is a laggard' read was a GICS-bucket artifact — software lives in Tech (+1.59%), not the -1.53% Comm Services bucket.
Durability. Most sustainable RUNWAY of the AI complex because YTD is still low (IGV only +1.9%) and the basket is a fresh base-breakout, the OPPOSITE of an extended trend — the deep discount-to-high leaves room. But on the live tape the individual leaders are recovery-bounce names (NONE beats IGV over 3mo or YTD; only NOW leads on 1mo, a parabolic +53% bounce off a 52-week low), so durability favors the BASKET/ETF over chasing single parabolic recoveries.
Leaders’ fundamental health. Mixed with a wide quality and valuation spread. NOW is Strong (22% durable subscription growth, 75% GM, 40%+ FCF margin) but extended +53% in a month.
PLTR is Stretched — elite/accelerating fundamentals (rev +85%, 46% op margin, 55% FCF margin, net cash) at the most extreme valuation in software (~213x TTM EV/S; analyst avg PT only ~2.3% above spot). CRM is Mixed and the deepest laggard (below both MAs, -20.9% YTD, growth decelerated to ~13%) with a NEW balance-sheet flag — a $24.8B debt issuance funded a $27.2B buyback, pushing net-debt/EBITDA to ~13.6x from ~2.6x.
Quantum Computing — laggard-to-leader reversal into the Quantinuum IPO
carried forward
What’s driving it. US Commerce/NIST $2.013B CHIPS incentives + minority equity stakes across nine quantum names (May 21, verified) plus the imminent upsized Quantinuum IPO (~$1.43B, up to $14.3B valuation) — a consensus-confirming event still AHEAD, meaning broad recognition has not arrived.
Durability. Short-term and event-driven — high-beta basket sized as a TRADE, not a core holding. The reversal is recent (basket bottomed late March; IONQ ~+132% since), the policy catalyst is discrete, and the IPO is the proximate event.
All names pulled back/consolidated into the IPO (IONQ -3.9% on 06-01) rather than blowing off, which keeps it early; but this is the least fundamentally-grounded theme and durability depends entirely on the IPO landing well and government endorsement holding.
Leaders’ fundamental health. Weak across the pure-plays — selection here is PURELY relative-strength, not fundamentals. IONQ is the genuine RS leader (only pure-play beating the QTUM ETF YTD, real revenue $64.7M) but Weak fundamentally (EV/S ~150x, FCF -$159M/qtr, cash fell $1.04B→$0.5B in a quarter; Q1 GAAP 'profit' is a non-cash warrant gain masking a -$271.5M operating loss; EXCLUDED from the May 21 government program).
QBTS and RGTI are mis-tagged laggards (QBTS YTD +8.3%, -37.6% off high, EV/S ~1753x; RGTI -55.9% off high, NEGATIVE gross margin). HON is the only investment-grade expression but a heavily DILUTED proxy (Quantinuum is a small slice of a $150B industrial).
Core watchlist — the fundamentally-strong names we track
The persistent core. Leaders scored 0–100 (deterministically, from a point-in-time fundamentals cache) on a CANSLIM + SEPA growth-leader quality fingerprint — durable revenue growth, earnings inflection, margin expansion, high ROIC, and free-cash-flow generation (O’Neil + Minervini + the quality-factor literature). Winner-grade names (≥60) are carried forward run-to-run and tracked here. The RS leaders that fail the profile still appear on the theme cards as current momentum, but are NOT tracked (momentum is transient). This measures “a great, improving business now” on names the RS engine already flagged — it does not encode the 100-bagger size/runway/cheap-entry asymmetry (those describe finding a multibagger before it moves, which this pipeline can’t do), so a high score on a mega-cap is a quality read, not a forward-multibagger one. Descriptive, not a forecast.
Proven compounderswinner-grade · tracked core · mature / large-cap17
Emerging candidateswinner-grade · tracked core · small-enough base to still compound5
Momentum — not trackedRS leaders shown on the cards; fundamentals don't match the winner fingerprint (transient)18
Top actionable ideas
Entry. buyable-near-pivot — breaking out to a fresh 52wk high TODAY (+5.5%) rather than already-vertical like MU; the pivot is still actionable. Stop-reference below the breakout level / rising 50DMA.
Thesis. This is the cleanest actionable entry in the two strongest themes. It has the highest fundamental quality in the memory complex: 54% ROE, 34% ROIC, a net-cash balance sheet, and ~23x earnings.
It also has the highest memory mix among US large-cap semicap names. So Micron's FY26 HBM capex step-up flows straight into its order book.
You get the sold-out HBM bottleneck one derivative removed from MU's parabola.
Risk. Being one derivative removed from the bottleneck means high-beta to an AI-capex pause in both directions. It also carries China/export-control and NAND-cycle exposure.
A memory-capex digestion hits semicap orders before it hits the memory makers.
Entry. buyable-near-pivot — above both its 50DMA ($137.89) and 200DMA ($119.71), only 8.8% off its high. Note the sector's recent leg is partly a crude/Hormuz tailwind, so size for macro reversal; this is a quality-at-reasonable-price hold, not a momentum chase.
Thesis. This is the highest-quality balance sheet in the energy cohort. It's also a confirmed RS leader inside the sector, holding its trend while the majors and the XLE index roll below it.
It earns best-in-class returns at the cheapest multiple in the group, with ROIC of 58%. The -3.5% revenue decline is driven by realized prices, not a franchise or volume problem.
Risk. Energy is a YTD-relic theme on present-tense RS, with XLE below its 50DMA. Part of the recent strength is an Iran/Hormuz crude-price premium that unwinds if the geopolitical risk fades.
The single-name RS holds. But sector beta is the headwind.
Entry. buyable-near-pivot — trades above its 200DMA ($100.42), sits marginally below its 50DMA ($111.02); shallowest 3m drawdown of the cohort. Buying a high-quality producer near its base while the theme digests, not chasing a breakout.
Thesis. This is the cleanest value entry into the gold-consolidation digestion. It's the cheapest of the major precious-metals names on every multiple, with the highest ROE and best FCF yield after a genuine turnaround to net cash.
As the largest, most-liquid US producer, it has the most torque if gold resumes from the base. It also has a self-help FCF tailwind that's independent of the metal.
Risk. Gold is consolidating, down -19% off its high, not leading right now. Wait for a 50DMA reclaim to confirm.
As a producer, its margins move directly with gold, so it has more downside torque than the streamers WPM/FNV if the base breaks lower. It's also still working through a 2026 production trough with a sprawling, recently restructured asset base.
Entry. constructive-basing — above its 50DMA ($308) and well above its 200DMA ($212), but ~12% off its high (not pressed against it like ETN). Better risk/reward than chasing ETN at its high; add on consolidation toward the rising 50DMA.
Thesis. This is the highest-quality, least-speculative way to own the AI buildout. It's the thermal/power franchise leader, on a clean multi-month trend backed by hard order velocity rather than a one-day spike.
It's the cleanest leader in the group. And it's the only name with both strong RS and genuine fundamental quality.
Risk. The valuation leaves no room for error, with EV/EBITDA around 29x and FCF yield of only about 3%. A hyperscaler capex air-pocket or cooling-share loss would compress estimates and the multiple together.
The backlog and order figures are company-reported and not independently re-verified.
Entry. constructive-basing — printed a fresh 52wk high but at roughly half the index's pace (a relative laggard that is participating, not spearheading), so it has not gone vertical. Accessed via the legitimate US-listed NYSE ADR (ISIN US8740391003); reasonable valuation gives a lower-risk entry than the extended leaders.
Thesis. This is the lowest-variance, best risk-adjusted way to own the #1 theme. It has the best fundamental profile in the semi cohort: 60% GM, 45% net margin, 32% ROE, net cash, and a P/E of only 28x.
It rides the silicon up-cycle through every customer, since AVGO, NVDA, MRVL, and AAPL all depend on its leading-edge and CoWoS capacity. And it does so without the single-name parabola risk of MRVL or MU.
Risk. It's foreign-domiciled in Taiwan, which brings geopolitical and concentration risk. Add high capex intensity, over 30% of revenue, plus US export-control and tariff exposure.
It's a relative laggard, so it lags hardest on the upside if the theme runs vertically. It's the defensive expression here, not the spearhead.
Entry. buyable-near-pivot — above both its 50DMA ($957.76) and 200DMA ($939.55), only 7.3% off its high, with the trend structure intact in a sector where almost nothing else is leading. The only large-cap healthcare name with both clean RS and clean fundamentals.
Thesis. This is the cleanest single-name RS leader inside a broadly red sector. It leads XLV by about 12pp over 3 months, with a healthy, accelerating franchise: GM 83.8%, ROE 77.8%, ROIC 30.2%, and rev +44.7%.
It also has a fresh, verified catalyst in the orforglipron/Foundayo FDA approval on Apr 1, 2026. That's genuine RS plus genuine fundamental health, unlike UNH (a recovery with deteriorating fundamentals) or JPM (below both MAs).
Risk. The valuation prices perfection, with EV/S of 15.3x and FCF yield under 1% as FCF lags net income on heavy incretin-capex and inventory build, with DIO around 475 days. Any orforglipron commercial-ramp or manufacturing-supply miss would compress that premium fast.
The caution is about price, not the business.
Emerging / early-rotation watch
- SCCO (copper)the QUALITY leader of the copper complex (ROE 39%, ROIC 23%, 57% GM, fortress balance sheet net-debt/EBITDA 0.39x) but a 3-month PRICE laggard (-7.9% vs flat COPX proxy), basing ~9% below its high. Trigger to upgrade from watch to buy: a 50DMA reclaim that confirms the catch-up move is underway — you get the cleanest balance sheet in the theme without chasing TECK/FCX's vertical breakouts.
- GOOGL (mega-cap platform)best mega-cap relative strength (+18% 3m, +15% YTD, above both MAs) being sold purely on its OWN capital-allocation event (the $80B equity raise funding $180-190B of 2026 AI capex), not broken business momentum (ROE 31.8%, ROIC 21.8%, near net cash). Trigger: stabilization above the rising 50DMA ($349) once the $40B ATM supply overhang is digested and capex visibly converts to cloud/ads ROIC — a fundamentally-Strong name temporarily on the wrong side of the rotation.
- TLN (IPP)the ONLY independent power producer above BOTH its 50DMA and 200DMA, holding its uptrend while CEG/VST broke; the cleanest technical expression of the nuclear-for-AI thesis. Trigger to engage: a fresh-high breakout that confirms the structural demand is re-asserting through the IPPs — but size small given the FY25 GAAP net loss (-$219M) and single-plant Susquehanna concentration; this is torque, not quality.
- Biotech / oral-obesity rotation (XBI, IBB)the popular healthcare laggard-to-leader narrative is NOT yet price-validated (XBI weak, below its 50DMA; VKTX broken -32% off high). Trigger: XBI reclaiming its 50DMA on expanding breadth would be the first confirmation of a genuine sub-sector rotation — until then it is a thesis, not a trade.
- OKLO (SMR)leads the SMR sub-group on 1m/3m momentum with a real regulatory de-risking event (NRC Aurora PDC approval, 2026-05-06) and a verified 14 GW pipeline, but lags the broad theme YTD and is -62% off its high, pre-revenue, FCF-negative. Trigger: a 200DMA reclaim ($85.7) — until then it is a small-sized high-beta call-option on power-for-AI, not a position.
Avoid / fading
- MRVLDANGEROUSLY EXTENDED: +32.5% in a SINGLE session (its largest one-day gain ever) to a fresh ATH, ~95% above its 50DMA, on a Jensen Huang verbal endorsement ('next trillion-dollar company') NOT a fresh earnings release. CORRECTION TO THE NARRATIVE: fundamentals are MIXED not strong — GAAP net margin ~1.4%, ROIC ~6%, and the FY26 GAAP profit was flattered by a ~$1.9B one-time gain; the move is momentum + sentiment, not an earnings re-rate.Chasing this print is buying a vertical blow-off with maximal gap-fill risk.
- MUgenuine leader but PARABOLIC: +84% in one month, +273% YTD, ~3x its 200DMA. Cheap on earnings but the chart is a blow-off, and memory is cyclical/commodity — the same sold-out tightness reverses violently when supply catches up (new capacity H2-2027).Do NOT initiate at the high; own the bottleneck via LRCX instead.
- COINCORRECTION: mis-framed as a 'pick-and-shovel leader' but on the live tape it is the deepest LAGGARD of the fintech/crypto cohort — a broken chart -61% off its high, -26% YTD, below both MAs, with profitability collapsed (net income -51% YoY, ROE 25.1%->8.5%) yet still ~36x P/E. Broken-avoid; it is funding the rotation, not leading it.
- CEG / VST / XLU (utilities & IPPs)CORRECTION TO THE 'power-for-AI' NARRATIVE: technically broken, not leading. CEG -34% off high after a fresh $3.1B dilutive secondary (6/1) with FY25 net income -38% YoY; VST -28% off high and below its 200DMA; XLU below both MAs.The demand thesis is real but is being expressed through copper / VRT/ETN / CCJ, NOT these equities. Wait for a 200DMA reclaim before re-engaging.
- Energy at the index level (XLE)CORRECTION: owns the YTD crown (~+27%) but is a present-tense laggard (below its 50DMA, ~8.7% off its high). The YTD number is a crude/Hormuz price artifact, not equity leadership; the recent pop is an oversold bounce off a 50DMA undercut.Not a place to add at the index — only the single-name quality leaders (EOG, FANG) hold independent RS.
- AEMCORRECTION: proposed as a co-'dominant-leader' of gold but is actually the relative LAGGARD — worst on 1m/3m/YTD vs both GDX and all three peers, the only name below BOTH its 50DMA and 200DMA, deepest drawdown (-30%). Fundamentals are excellent (record margins, net cash, PEG 0.14) but it is 'right fundamentals, wrong tape' — too-early, no confirmation.Own NEM (cheaper, leading) instead.
- VKTXbroken-chart binary: -32% off its high, below both MAs, pre-revenue with a FY25 loss widened to -$359.6M and only ~0.6yr cash runway, into a binary VANQUISH-2 Phase 3 readout (Q3 2026). Never a leadership name; a sized-down satellite at most.
- VNOM / LEUpresent-tense laggards mis-framed as leaders. VNOM (energy royalty) trails XLE and both operators, below its 50DMA, with a FY25 GAAP net loss and negative FCF.LEU (enrichment) is -57% off its high, below its 200DMA, YTD -17.8% vs theme +25% — a turnaround/basing name, not a leader, despite real DOE-contract fundamentals.
Caveats
- DESCRIPTIVE, NOT A FORECAST: This is a snapshot of which themes and names lead on measured relative strength as of the data date. It describes the current tape. It does not predict future returns. Relative-strength leadership is a momentum observation. It can persist or reverse without notice. Nothing here is a price target or a probability statement.
- DATA AS-OF: Prices and returns are live FMP data from the 2026-06-01 and 2026-06-02 sessions. The session varies by theme. Silicon, memory, gold, uranium, and mega-cap were verified 06-02. Copper, energy, buildout, utilities, and healthcare were verified 06-01. Several leaders printed fresh 52wk highs intraday on the as-of date. So entry levels are stale by the time you read this. Re-check the live tape and DMAs before acting.
- ROTATION-REVERSAL RISK: The whole regime assumes capital is rotating INTO AI-hardware and copper and OUT of platforms, fintech, crypto, and utilities. That rotation is the thesis. It is also the main risk. If AI-capex expectations reset, the most-extended leaders air-pocket hardest: MRVL, MU, AVGO. And the 'funding-source' names being sold, GOOGL and META, could become the relative winners. The leaders and the laggards can swap on a single capex-digestion headline.
- CORRELATION HAZARD: The top ideas are NOT independent bets. LRCX, TSM, and the broader silicon/memory complex are all single-factor exposures to AI-data-center capex. A hyperscaler capex pause hits all of them at once. It also hits VRT (buildout), CCJ (power), and even copper (data-center demand leg). A portfolio built from this readout is heavily concentrated in one macro driver, despite spanning multiple 'themes'. Size for the shared factor, not the apparent diversification.
- VERIFICATION CORRECTIONS CARRIED FORWARD: Several original theses were wrong. We corrected them against live data. Energy and utilities were YTD-return relics, not live leaders, with XLE/XLU/VST/CEG below key MAs. AEM was a laggard, not a co-leader. COIN was the deepest laggard, not a pick-and-shovel leader. The LEU/VNOM/OKLO 'leadership' was moat and narrative, not relative strength. MRVL's spike was a CEO soundbite, not an earnings re-rate, with MIXED, not strong, fundamentals. The relative-strength picks survived these corrections. Treat any single-day pop or YTD-headline claim with suspicion until MA structure confirms it.
- EXTENSION / ENTRY RISK: The strongest-RS names are mostly extended-chasing entries: MRVL, MU, AVGO, FCX, TECK, CCJ, ETN. The cleanest setups deliberately prioritize basing or just-breaking quality instead: LRCX, NEM, EOG, VRT, TSM, LLY. That accepts lower beta for better risk/reward. There is no buyable-near-pivot entry in the most extended leaders today.
- FUNDAMENTAL DATA LIMITATIONS: FMP key-metrics market caps are stale versus current prices for the fastest-moving names. This is worst for MU, where FY-end mcap understates EV/EBITDA badly. So some reported valuation multiples understate current richness. P/E was recomputed on live price where flagged, but treat all multiples as approximate. Company-reported backlog and order figures were not independently re-verified, including VRT, GEV, and pipeline GW for OKLO and CCJ.
- NOT PERSONALIZED ADVICE: This is research output, not investment advice. It is not tailored to your objectives, risk tolerance, time horizon, tax situation, or existing positions. It is a top-down theme-leadership screen for a sophisticated reader to use as one input among many. Do your own due diligence and position-sizing.
What’s driving it. The live catalyst is hyperscaler custom-XPU/ASIC programs plus scale-out Ethernet and optical connectivity. SMH leads on every window: +24% 1m, +62% 3m, +76% YTD.
Three names (MU/NVDA/GOOGL) drive 40%+ of 2026 S&P EPS revisions. The most recent leg came when Nvidia's CEO endorsed Marvell, layered on top of record backlogs.
Durability. The buildout is sustainable as a multi-year structural story. But the current leadership tape is increasingly froth-driven and extended.
MRVL gapped +32.5% in a single session on a CEO soundbite, not earnings. AVGO and MU sit within ~1.5% of fresh highs.
The structural demand lasts. The present price tape is vulnerable to a sharp AI-capex-digestion air-pocket.
Own the theme, not the top tick.
Leaders’ fundamental health. Bifurcated. The fundamentally strongest businesses are the relative-strength laggards.
NVDA runs 71% GM, 76% ROE, 63% ROIC, $96.7B FCF, net cash, and a reasonable 21x EV/S. TSM runs 60% GM, 45% net margin, 32% ROE, net cash, and 28x P/E.
Meanwhile the relative-strength leaders are the most stretched. AVGO has real FCF but EV/S of 27x, net-debt/EBITDA of 4.79x post-VMware, and negative tangible book.
MRVL is the cheapest on EV/S at 8.5x but has thin GAAP profitability, with ~1.4% net margin and ROIC ~6%. Aggregate health: strong cash generation and balance sheets at the quality end, momentum-stretched valuations at the leadership end.