This week's short book is a near-monolith: 24 of 25 names are Information Technology (the lone exception, Bloom Energy, is an Industrial that trades like one), and within that, the cluster is overwhelmingly AI-infrastructure and crypto-mining hardware/semis — Astera Labs, Credo-adjacent connectivity (AAOI, NVTS, AEHR, SITM, CEVA, MXL), the Bitcoin miners pivoting to AI compute (HUT, RIOT, CORZ, WULF, CLSK, CIFR, APLD), plus richly-valued software/data names (DDOG, NBIS). The data describes a textbook short cohort: an average Oddsmaker Score of −121 and Super Multiple of −260, with valuations that are detached from fundamentals — a mean forward EV/Sales of ~24x and EV/EBITDA near 239x — sitting on negative profitability (avg ROIC −40%, with 18 of 25 outright negative), trading at 141% of the model's OM fair-value target (i.e., ~40% above where the model sees value) and screening in just the 48th-percentile Holy Trinity. The behavioral signature is extreme: average Multiple Compression Risk of 85, Behavioral Dislocation Risk of 89, an average beta above 3.0, and 11.7% short interest — these are the most crowded, most volatile, most narrative-driven names in the market. The macro read is the mirror image of the long book: where the longs say cash flow and capital discipline are mispriced cheap, the shorts say the AI/crypto-compute trade has run far ahead of earnings, with the entire downside concentrated in a single thesis — that the market is paying parabolic multiples for speculative, cash-burning, high-beta exposure to one story. Critically, the model is flagging these as expensive and fragile, not as imminently collapsing: the very high squeeze-risk scores (VSR 76, double-digit short interest) are a warning that this is a fundamentally-justified-but-dangerous side of the book — right on valuation, but where being early can be punished by momentum. In short, the framework is positioning for a regime where the speculative froth concentrated in AI/crypto hardware mean-reverts toward fundamentals, even as it cautions that the unwind is a when, not yet a now.

The story: the highest short conviction in the book — it maxes every axis with the most negative readings here: Score −209.8, Super Multiple −435, plus MCR 89.1, BDR 98.9, VSR 87.4. Why investors are excited: a re-rate from SiC/EV test toward AI/photonics burn-in. Why the model hates it: 639× forward EV/EBITDA, negative ROIC (−8.2%), price 1.64× its 200-day, beta 5.3, and revenue contracting (−26% forward) — a hyped multiple on shrinking sales. Three bull arguments: (1) an AI/photonics test-demand pivot could re-accelerate orders; (2) tiny float + beta 5.3 makes squeezes violent; (3) a single design-win headline moves the stock. Three bear arguments: (1) a 639× EBITDA multiple on falling revenue is indefensible; (2) 45% implied downside to OM Target; (3) Holy Trinity in the 2.5th percentile — bottom-of-universe composite. Verdict: top-conviction short; the cleanest valuation-vs-fundamentals gap on the board.

The story: a genuinely good business shorted purely on valuation/dislocation — EQS 90.2, Holy Trinity 99.1%ile, yet Score −126.7, SMP −309.6, BDR 99.3. Why investors are excited: a premier AI-connectivity (PCIe/CXL retimer) play with 104% revenue growth. Why the model hates it: 99.5× forward EV/EBITDA, 38× EV/sales, price 1.98× its 200-day — flawless growth at a price that prices in flawlessness. Three bull arguments: (1) real, accelerating AI-datacenter demand; (2) high earnings quality (EQS 90) — this isn't a junk short; (3) any AI-capex upside cycle lifts it further. Three bear arguments: (1) 38× sales leaves zero margin for a single soft guide; (2) 27% implied downside to OM Target; (3) lockup/insider-supply and momentum reversal risk at 2× the 200-day. Verdict: highest-quality name in the short book — a valuation/timing short, not a fundamentals short; size carefully against the strong business.

The story: ranks #3 on the highest squeeze profile in the book (VSR 92.1) plus deep model disdain (Score −88.7, SMP −237). Why investors are excited: an ex-Yandex "neocloud" renting GPUs into the AI boom, 575% revenue growth. Why the model hates it: forward FCF margin −28.5 (catastrophic cash burn), 26× EV/EBITDA, revisions in the 2nd–7th percentile. Three bull arguments: (1) hyper-growth GPU-cloud narrative; (2) 19.9% short interest = real squeeze fuel; (3) backing/relationships with marquee AI names. Three bear arguments: (1) −28.5 forward FCF margin — it consumes cash to grow; (2) GPU-cloud is becoming commoditized; (3) bottom-decile estimate revisions. Verdict: high-conviction fundamental short, but the 20% SI demands tight risk control.

The story: the strongest of the crypto-miner shorts — Score −157.4, SMP −371.2, negative on every fundamental. Why investors are excited: a Bitcoin miner re-branding into AI/HPC datacenters and energy. Why the model hates it: negative ROIC (−18%), forward FCF margin −7.6, 68× EV/EBITDA, beta 4.3, price 1.66× its 200-day, revisions 5th percentile. Three bull arguments: (1) BTC price spikes lift the whole cohort; (2) an AI/HPC hosting contract re-rates it overnight; (3) beta 4.3 cuts both ways. Three bear arguments: (1) negative ROIC and negative forward FCF — value destruction; (2) 34% implied downside; (3) the AI-pivot is narrative, not yet numbers. Verdict: high-conviction short on a structurally cash-negative model; respect BTC-correlation risk.

The story: classic miner short — Score −80.5, SMP −259, with BDR 91.1. Why investors are excited: scaled Bitcoin mining + an AI/HPC datacenter optionality story. Why the model hates it: negative ROIC (−25%), negative forward FCF, 69× EV/EBITDA, beta 3.8, Holy Trinity 6th percentile. Three bull arguments: (1) direct BTC-price torque; (2) a Texas-power/AI-hosting pivot; (3) heavily-traded, squeeze-prone. Three bear arguments: (1) negative returns on capital with dilutive financing; (2) bottom-decile composite (Holy Trinity 0.06); (3) 17% implied downside. Verdict: conviction short tied to the miner-cohort thesis; BTC beta is the live risk.

The story: a small-cap semi ranking high on dislocation (BDR 95.4) and squeeze (VSR 88.0) despite middling fundamentals. Why investors are excited: the leading MRAM (non-volatile memory) niche supplier. Why the model hates it: price stretched (BDR 95), beta 3.4, EV/EBITDA 26×, mediocre revisions (22nd percentile), thin growth (16%). Three bull arguments: (1) MRAM design-win in aerospace/industrial; (2) tiny, illiquid float squeezes easily; (3) niche-monopoly optionality. Three bear arguments: (1) price 1.39× its 200-day with slowing revisions; (2) sub-scale economics; (3) 14% implied downside. Verdict: dislocation-driven short; moderate conviction given the niche moat.

The story: the most negative model score in the book at the deep end (Score −250.3) and highest squeeze read (VSR 97.3) — a battleground name. Why investors are excited: a GaN/SiC power-semi pure-play hyped on Nvidia's 800V datacenter-power roadmap. Why the model hates it: negative EBITDA (−102× multiple), 87× EV/sales, negative ROIC (−35%), beta 4.3, price stretched, revenue −45%. Three bull arguments: (1) the Nvidia 800V-power narrative is a powerful catalyst; (2) VSR 97 flags brutal squeeze potential; (3) any datacenter-power design win re-rates it. Three bear arguments: (1) negative EBITDA with falling revenue — the multiple is meaningless; (2) 54% implied downside to OM Target (the largest in the book); (3) GaN power is competitive and commoditizing. Verdict: strong fundamental short, but the highest squeeze risk on the board — size small, manage tightly.

The story: ranks on dislocation+squeeze (BDR 94.4, beta 5.1) plus deep negativity (Score −113.8, SMP −268). Why investors are excited: a spun-out AI-datacenter/HPC cloud-infrastructure story. Why the model hates it: forward FCF margin −5.1, negative ROIC, price 1.82× its 200-day, beta 5.1. Three bull arguments: (1) AI-cloud capacity demand; (2) small float + beta 5.1; (3) high near-term revisions (84th percentile). Three bear arguments: (1) cash-burning infrastructure build; (2) 24% implied downside; (3) unproven, narrative-driven model. Verdict: conviction short, high-volatility — treat like the neocloud cohort.

The story: deeply negative (Score −172, SMP −359) with squeeze risk (VSR 92.9). Why investors are excited: compound-semi (InP/GaAs) substrates levered to AI optical/photonics demand. Why the model hates it: 149× EV/EBITDA, 27× EV/sales, negative ROIC, flat revenue, China-export overhang. Three bull arguments: (1) AI-optics substrate demand; (2) China-listing of its subsidiary as a value-unlock; (3) cyclical-recovery torque. Three bear arguments: (1) 149× EBITDA on flat sales; (2) export-control/China risk; (3) 37% implied downside. Verdict: high-conviction valuation short with a China-policy kicker.

The story: the worst earnings-quality in the book (EQS 11.9) plus deep model disdain (Score −104.6, SMP −301). Why investors are excited: a "clean-energy" Bitcoin miner pivoting to AI/HPC hosting. Why the model hates it: forward FCF margin −22.0, negative ROIC (−35%), 61× EV/EBITDA, 23% short interest, revisions bottom-1%. Three bull arguments: (1) BTC torque + an AI-hosting contract narrative; (2) 22% SI = squeeze fuel; (3) zero-carbon power positioning. Three bear arguments: (1) −22 forward FCF margin is among the worst here; (2) bottom-percentile revisions and EQS 12; (3) dilutive capital structure. Verdict: strong fundamental short; the 22% SI is the offsetting risk.

The story: the single worst ROIC in the book (−640%) and extreme dislocation (BDR 98.3). Why investors are excited: network-on-chip (NoC) IP licensing into every AI/automotive SoC. Why the model hates it: 4,326× EV/EBITDA, ROIC −6.4×, price 1.89× its 200-day, Score −156.9. Three bull arguments: (1) IP-royalty leverage to the AI-chip boom; (2) recurring licensing model with design-win momentum; (3) acquisition-target optionality. Three bear arguments: (1) catastrophic returns on capital (−640%) and a 4,300× multiple; (2) 34% implied downside; (3) price stretched ~1.9× its trend. Verdict: high-conviction valuation short; the IP-royalty story is the bull risk.

The story: the lone non-IT name, ranking on strong fundamentals-vs-price tension — FRM 89.9, Holy Trinity 83%ile, yet Score −88.7, SMP −250, BDR 90.6. Why investors are excited: solid-oxide fuel cells positioned as on-site power for AI datacenters. Why the model hates it: 105× EV/EBITDA, 22× EV/sales, ROIC ~0, beta 4.3, price stretched, momentum way ahead of cash. Three bull arguments: (1) the AI-datacenter-power thesis is the hottest narrative in energy; (2) 89.9 FRM — genuine growth acceleration; (3) beta 4.3 + datacenter contract headlines. Three bear arguments: (1) 105× EBITDA / 22× sales with ~zero ROIC; (2) 19% implied downside; (3) a long history of cash burn and dilution. Verdict: valuation/timing short on a real narrative — the strongest bull case in the cohort; size against it.

The story: deep negativity (Score −154.3, SMP −300.5) and 23.5% short interest. Why investors are excited: a post-bankruptcy Bitcoin miner pivoting to AI/HPC hosting (CoreWeave relationship). Why the model hates it: ROIC −130%, forward FCF margin −1.4, 29× EV/EBITDA, beta 2.6. Three bull arguments: (1) a marquee AI-hosting contract underpins the bull thesis; (2) 23.5% SI = explosive squeeze risk; (3) BTC torque. Three bear arguments: (1) ROIC −130%; (2) negative forward FCF; (3) 33% implied downside. Verdict: conviction short on economics, but the 23.5% SI plus a real customer make this the squeeze-iest name here — manage tightly.

The story: negative across fundamentals (Score −83.8, SMP −232.8) with squeeze (VSR 89.6). Why investors are excited: an optical-transceiver supplier riding AI-datacenter interconnect demand. Why the model hates it: 55× EV/EBITDA, negative ROIC, negative forward FCF, price 1.28× its 200-day. Three bull arguments: (1) 800G/1.6T datacenter-optics demand; (2) hyperscaler design-win optionality; (3) high-beta squeeze potential. Three bear arguments: (1) chronically negative FCF and margins; (2) lumpy, customer-concentrated revenue; (3) 18% implied downside. Verdict: conviction short with binary design-win risk to the upside.

The story: the single most price-stretched name in the book (2.39× its 200-day, BDR 99.4). Why investors are excited: a broadband/connectivity/infrastructure semi positioned for a cyclical recovery. Why the model hates it: 46× EV/EBITDA, negative ROIC (−21%), Score −125.6, price nearly 2.4× its trend. Three bull arguments: (1) a broadband/datacenter cyclical upturn; (2) optical-DSP and infra-design wins; (3) M&A/consolidation optionality. Three bear arguments: (1) price 2.39× the 200-day — the most extended dislocation here; (2) negative ROIC; (3) 27% implied downside. Verdict: high-conviction mean-reversion short on the dislocation alone.

The story: a composite short (Score only −29.5) ranking on dislocation (BDR 96.3) rather than deep negativity. Why investors are excited: precision sensors/load-cells/resistors with industrial-recovery exposure. Why the model hates it: price 1.96× its 200-day, 42× EV/EBITDA, near-zero ROIC, soft growth. Three bull arguments: (1) industrial/test-and-measurement recovery; (2) defensive niche with stable demand; (3) modest valuation vs. the rest of the cohort. Three bear arguments: (1) price ~2× its trend with thin growth; (2) low ROIC; (3) 6% implied downside (smallest here — weakest gap). Verdict: lower-conviction dislocation short; the thinnest downside in the book.

The story: a composite short on dislocation (BDR 95.6) with the worst forward-revenue trend in the cohort (FRM 11.0, revenue −7.7%). Why investors are excited: semiconductor-capital-equipment (MOCVD/deposition) levered to advanced-node and AI buildouts. Why the model hates it: declining revenue, price 1.66× its 200-day, revisions 1st–2nd percentile. Three bull arguments: (1) a semicap up-cycle lifts all equipment names; (2) advanced-packaging/AI tool demand; (3) low beta (1.4) limits downside velocity. Three bear arguments: (1) contracting revenue with bottom-percentile revisions; (2) price extended vs. trend; (3) 7% implied downside. Verdict: moderate-conviction short — weak fundamentals but a small valuation gap.

The story: ranks on the highest multiple-compression risk in the book (MCR 94.4) plus decent quality (EQS 69.3). Why investors are excited: a high-margin low-power FPGA leader with secular content growth. Why the model hates it: 67× EV/EBITDA, 25× EV/sales, Score −57, price above trend — premium multiple, decelerating growth. Three bull arguments: (1) genuine FPGA franchise with 57.9 CAS (best in the short book); (2) an inventory-correction recovery; (3) M&A target appeal. Three bear arguments: (1) 67× EBITDA on slowing growth (MCR 94); (2) revisions soft (22nd percentile); (3) 12% implied downside. Verdict: valuation/multiple-compression short on a good company; lower conviction, watch for a growth re-accel.

The story: a composite short with the second-highest multiple-compression and squeeze reads (MCR 95.7, VSR 95.0) and 26% short interest. Why investors are excited: AI-datacenter/HPC hosting (plus legacy crypto) with hyperscaler-lease optionality. Why the model hates it: forward FCF margin −2.9, negative ROIC, 57× EV/EBITDA, 26% SI. Three bull arguments: (1) a large hyperscaler AI-lease can re-rate it instantly; (2) 26% SI = severe squeeze risk; (3) datacenter-capacity scarcity. Three bear arguments: (1) cash-burning build-out with negative ROIC; (2) dilutive financing needs; (3) leased-datacenter economics are thinner than the hype. Verdict: fundamental short, but the 26% SI + lease-headline risk make it among the most dangerous to be short — small size only.

The story: deeply negative model score (−146.8, SMP −296.4) on a micro-cap. Why investors are excited: embedded-FPGA/eFPGA IP with defense and AI-edge design-win potential. Why the model hates it: 44× EV/EBITDA, negative ROIC (−41%), revenue −21%, revisions 6th percentile. Three bull arguments: (1) lumpy defense/eFPGA contracts can spike revenue; (2) tiny float squeezes hard; (3) IP-licensing optionality. Three bear arguments: (1) shrinking revenue with negative ROIC; (2) 32% implied downside; (3) sub-scale and cash-constrained. Verdict: high-conviction micro-cap short; contract lumpiness is the upside risk.

The story: ranks on the highest short interest in the entire book (33%) and bottom-of-universe composite (Holy Trinity 0.7%ile). Why investors are excited: a pure-play Bitcoin miner scaling hash rate efficiently. Why the model hates it: forward FCF margin −4.8, negative ROIC, beta 3.9, 33% SI, Holy Trinity 0.7%ile. Three bull arguments: (1) the most direct BTC-price torque in the cohort; (2) 33% SI = the most explosive squeeze setup here; (3) operational-efficiency leadership among miners. Three bear arguments: (1) negative FCF and ROIC; (2) pure BTC-cost-curve exposure; (3) 9% implied downside (small — the squeeze risk has compressed the gap). Verdict: the model likes the fundamentals-short, but 33% SI makes this the highest squeeze-risk name on the board — arguably express via puts, not borrow.

The story: a broad analog/discrete name ranking on dislocation (BDR 98.4, price 2.02× its 200-day). Why investors are excited: a cyclical-recovery bet on industrial/auto semis and passives. Why the model hates it: price 2× its trend, negative forward FCF, weak revisions (22nd percentile), Score −75.7. Three bull arguments: (1) an industrial/auto-semi inventory-restock cycle; (2) broad diversified end-markets; (3) capacity-expansion payoff. Three bear arguments: (1) price 2.02× the 200-day ahead of fundamentals; (2) negative forward FCF; (3) 16% implied downside. Verdict: mean-reversion short on the dislocation; moderate conviction.

The story: a high-quality composite short — EQS 72.1, Holy Trinity 88.9%ile, the highest MCR in the book (95.7) — shorted on valuation, not weakness (Score only −43.5). Why investors are excited: a best-in-class cloud-observability SaaS with strong revenue growth and AI-monitoring tailwinds. Why the model hates it: 65× EV/EBITDA, 16.5× EV/sales, price above trend — premium multiple vulnerable to compression. Three bull arguments: (1) durable, high-retention SaaS with EQS 72 — a quality name; (2) AI-observability is a real growth vector; (3) low beta (1.5) limits squeeze velocity. Three bear arguments: (1) 65× EBITDA leaves no room for deceleration (MCR 95.7); (2) usage-based revenue is macro-sensitive; (3) 9% implied downside. Verdict: the highest-quality short here — a multiple-compression trade on a great company; lowest conviction, smallest gap.

The story: ranks on dislocation (BDR 95.6) with weak fundamentals (FRM 19.7, revenue −3%, negative ROIC). Why investors are excited: a semiconductor-capital-equipment subsystem/supply-chain play levered to a WFE up-cycle. Why the model hates it: price 1.82× its 200-day, declining revenue, negative ROIC, Score −50.2. Three bull arguments: (1) a wafer-fab-equipment spending recovery; (2) advanced-node/AI-tool content gains; (3) operating leverage on a rebound. Three bear arguments: (1) contracting revenue with negative ROIC; (2) thin, cyclical margins; (3) 11% implied downside. Verdict: moderate-conviction cyclical short; a WFE up-cycle is the bull risk.

The story: the mega-cap anchor — $668B name with strong dislocation (BDR 98.2, price 1.91× its 200-day) and decent growth optics (FRM 69.6). Why investors are excited: a turnaround/foundry-pivot narrative with government and partner support, plus AI-PC hopes. Why the model hates it: negative ROIC, negative trailing FCF, 34× EV/EBITDA, Score −76.6, price ~1.9× its trend on a still-unproven turnaround. Three bull arguments: (1) a foundry/18A-node milestone or marquee customer re-rates it; (2) policy/subsidy and strategic-investment support; (3) deep value if the turnaround takes hold. Three bear arguments: (1) negative ROIC and FCF while spending heavily on fabs; (2) price 1.91× its 200-day has run ahead of execution; (3) 16% implied downside. Verdict: valuation/dislocation short on a hope-driven rally; the turnaround-headline risk is the live offset.

Disclosure & Disclaimer

The Oddsmaker is a financial media and research publication provided for informational and educational purposes only. Nothing contained herein constitutes investment advice, a recommendation to buy or sell any security, or legal, tax, or accounting advice. The Oddsmaker, its affiliates, employees, contributors, related parties, and associated accounts may hold long, short, or other positions in securities discussed and may buy or sell such securities without notice. Any scores, rankings, ratings, probabilities, expected returns, forecasts, analytics, models, simulations, or backtested results are hypothetical analytical opinions based on assumptions and methodologies that may prove incorrect. They are not guarantees of future performance or outcomes. Information is obtained from sources believed to be reliable; however, The Oddsmaker makes no representation or warranty as to its accuracy, completeness, or timeliness. Past performance is not indicative of future results. Investing involves risk, including the possible loss of principal. Readers are solely responsible for conducting their own due diligence and consulting qualified financial, legal, tax, and accounting professionals before making investment decisions.

© The Oddsmaker. All rights reserved.