Research

The following constituents have a standing AI research report. Treat each report's verdict + summary as a prior input — corroborate or challenge it with current data, but don't ignore it. The report date matters: a stale report may predate a material development.

- DLR (researched 2026-06-02) — verdict: **Buy.** With a clean Q1 FY26 beat, raised 9%-growth FFO guidance, a record $16.5B preleased pipeline, and a multiyear-low 4.7x leverage, DLR is executing directly on the fund's hyperscale-landlord thesis at a de-rated entry point — we view the sub-cost position as a multiple reset, not a broken thesis, and would add toward the 1.41% target while monitoring hyperscaler-customer concentration and capex-cycle signals.
    Digital Realty Trust (DLR) is the world's largest cloud- and carrier-neutral data center REIT, owning, developing, and operating hyperscale and interconnection-rich colocation capacity across the Americas, EMEA, and APAC. It sits in the Diamond AI Fund's Core sleeve as a Data center REIT, where it closes the most obvious sub-category gap — adding a pure-play hyperscale landlord alongside EQIX and…
- TSEM (researched 2026-06-02) — verdict: **Hold.** The fundamental thesis is not just intact but strengthening — $1.3bn of contracted, prepaid 2027 SiPho revenue and a record Q2 guide validate Tower as an upstream chokepoint to the AI optical buildout — yet at a ~$32B cap, a c.105x trailing P/E, and a fresh Nvidia-circularity tie, the name has outgrown its diamond-in-the-rough framing and carries less valuation cushion, so we would maintain a starter position at the 1.50% target and monitor rather than add aggressively here, pending a sleeve-fit review and a better entry on any capex-cycle wobble.
    Tower Semiconductor (NASDAQ/TASE: TSEM) is a specialty analog foundry that has emerged as a leading merchant Silicon Photonics (SiPho) platform supplier for AI optical interconnect — the photonic chips and wafers feeding 800G/1.6T/3.2T transceivers and co-packaged optics (CPO).
- CDP (researched 2026-06-02) — verdict: **Buy.** We view CDP as a high-conviction Diamond holding: it delivered a clean Q1 2026 beat-and-raise, trades at an ~11x FFO discount to data-center peers with an investment-grade balance sheet and a ~4% yield, and offers genuinely non-duplicative powered-shell and defense-adjacent exposure — even though we caution that its AI/data-center contribution is still a small, ramping slice and the thesis leans as much on the defense-budget cycle as on hyperscaler capex.
    COPT Defense Properties (NYSE: CDP) is a self-managed REIT that owns, operates, and develops mission-critical office and data-center-shell facilities clustered around priority U.S. defense and intelligence installations. As of March 31, 2026, the Defense/IT Portfolio of 201 properties encompassed 23.2 million square feet and was 96.4% leased, with the U.S. Government as anchor tenant.
- COHR (researched 2026-06-02) — verdict: **Buy.** We maintain the high-conviction Diamond position: COHR is executing a high-quality, accelerating datacenter ramp (Q3 revenue +21% Y/Y, datacenter +37% Y/Y, Q4 guide above consensus) with a funded Nvidia capacity backstop and a Western-supply-chain edge that the fund's networking & optical thesis is built to own — while we keep its 1.00% pinned weight modest given Nvidia-circularity concentration and stretched valuation, and would not add aggressively into further multiple expansion.
    Coherent Corp. (NYSE: COHR) is a vertically integrated photonics leader whose datacenter & communications franchise — optical transceivers (800G/1.6T), optical circuit switches (OCS), and silicon-photonics/CPO components — sits squarely on the AI networking buildout the fund is designed to capture.
- CMI (researched 2026-06-02) — verdict: **Buy.** The data-center power thesis is materially advancing — record Power Systems margins, a backlog into 2028, raised 2026 and 2030 guidance, and a funded 20 GW capacity build — and we view CMI as a high-conviction, durable expression of the fund's electrification sleeve despite a forward multiple (~22x) that sits well above its historical ~14x and warrants ongoing monitoring.
    Cummins Inc. (NYSE: CMI) is a diversified power solutions company spanning diesel/natural-gas engines, distribution, components, the high-horsepower Power Systems segment, and the (loss-making) zero-emission Accelera unit. It earns its place in the Diamond AI Fund's Power, cooling & electrification sleeve as a Core read-through to the hyperscaler capex wave: its Power Systems segment supplies the…
- CLS (researched 2026-06-02) — verdict: **Hold.** The fundamental thesis is firmly intact — a 53% revenue print, record 8% margin, a $2 billion guidance raise, and genuine design wins (AMD Helios, hyperscaler CPO) — but with the stock having tripled over 52 weeks, trading at ~42x FY26E EPS near consensus targets, and carrying intensifying single-hyperscaler concentration plus a back-half-loaded ramp, the risk-reward no longer justifies adding; we maintain the existing Diamond weight and monitor customer mix and 2H execution closely.
    Celestica Inc. (NYSE: CLS) is a US-listed AI-hardware ODM whose Connectivity & Cloud Solutions (CCS) segment designs and manufactures custom AI server racks, 800G/1.6T Ethernet switching, and storage platforms directly for the hyperscaler/neocloud roster — precisely the non-Nvidia-circularity AI-hardware vector the fund's Servers/storage/robotics/vertical-AI sleeve is meant to capture.
- CIFR (researched 2026-06-02) — verdict: **Hold.** The pivot thesis is on track and de-risked on financing, but with HPC revenue not yet flowing, ~$5.2bn of leverage, acute three-tenant concentration, and the stock near 52-week highs at a 0.50% pinned target weight, we see balanced risk-reward that warrants maintaining — not adding to — the position while monitoring 2H26 energization milestones.
    Cipher Digital (NASDAQ: CIFR — renamed from Cipher Mining in February 2026) is a West-Texas-centric, industrial-scale data center developer mid-pivot from bitcoin self-mining to contracted hyperscale AI/HPC hosting. It sits in the fund's Neoclouds sleeve as a converted-miner name, and the thesis — AWS-anchored, contracted revenue visibility in a structurally underweight sub-category — is broadly…
- CIEN (researched 2026-06-02) — verdict: **Hold.** The AI/DCI thesis is fully intact and arguably accelerating — 33% revenue growth, 76% cloud growth to 42% of mix, and a ~$7B backlog skewing into FY27 — but with the stock up ~668% over twelve months, trading at ~77x forward earnings and ~24% above consensus fair value, the risk/reward no longer warrants adding; we maintain the 0.67% Core weight and monitor the June 4 print, trimming on any thesis crack or further multiple expansion.
    Ciena (CIEN) is a global supplier of optical networking systems, coherent transport, routing/switching, and the WaveLogic family of coherent DSP/optics — increasingly a core beneficiary of AI-driven data-center interconnect (DCI) and inter-DC traffic growth.
- CEG (researched 2026-06-02) — verdict: **Hold.** We see CEG as a high-quality, thesis-validated power-generation holding — strong Q1 beat, reaffirmed $11–$12 guidance, an ahead-of-schedule Crane restart, and a structurally favorable FERC co-location backdrop — but with the secondary-offering overhang still clearing, the FY26 multiple already re-rated to ~23x, and Street targets trimmed, we prefer to maintain the small pinned 0.83% Core position and monitor rather than add at current levels.
    Constellation Energy (CEG) is the largest US producer of carbon-free electricity, operating the country's biggest nuclear fleet alongside a now-substantial natural-gas and geothermal portfolio following the January 2026 close of its Calpine acquisition.
- CDNS (researched 2026-06-02) — verdict: **Hold.** The thesis is intact and arguably strengthening—record $8.0bn backlog, broad-based ~17–19% growth, and a leading agentic-AI/NVIDIA narrative validate the inception case—but a ~50x forward multiple and a 13% China revenue tail tied to live export-control risk leave little margin for error, warranting maintenance at the current 0.67% Core weight rather than an add.
    Cadence Design Systems (CDNS) is one of the two dominant electronic design automation (EDA) vendors globally, supplying the chip-design software, IP, hardware emulation, and—post-Hexagon—multiphysics simulation tools that every advanced semiconductor and AI-infrastructure customer relies on to tape out silicon.
- CCJ (researched 2026-06-02) — verdict: **Buy.** The position is small (1.33% target, pinned) and the thesis is intact and arguably strengthening — a clean Q1 2026 beat, unchanged guidance, a tightening uranium market, and the differentiated $80B Westinghouse/US-government newbuild optionality that AI data-center power demand directly underwrites — and while the ~90x near-term multiple and multi-year execution risk warrant monitoring, the risk-reward at a sub-1.5% Diamond weight justifies maintaining and adding on weakness rather than trimming.
    Cameco (NYSE/TSX: CCJ) is one of the world's largest uranium producers, operating tier-one Saskatchewan assets (McArthur River/Key Lake, Cigar Lake) plus fuel-services conversion, but its inclusion in the Diamond AI Fund rests on the *downstream* nuclear-newbuild angle: its 49% stake in Westinghouse Electric and a transformational US-government partnership to build new reactors that will, among…
- CAMT (researched 2026-06-02) — verdict: **Hold.** The thesis is intact and the order book — capped by the >$100M 2027 HBM/OSAT order on the report date — is doing exactly what a Diamond name should, but with the stock at ~45–50x non-GAAP earnings, near-flat 1H26 revenue, and ~49% China exposure, the risk-reward is balanced rather than compelling; we maintain the 1.07% target weight and monitor the 2H26 ramp and export-control backdrop for the next directional signal.
    Camtek (NASDAQ: CAMT) is an Israel-based supplier of high-end inspection and metrology equipment for the semiconductor industry, where it holds a leading position in advanced-packaging and HBM (high-bandwidth memory) inspection — precisely the chokepoints the AI buildout is straining.
- BWXT (researched 2026-06-02) — verdict: **Buy.** BWXT pairs an uncontested naval-reactor monopoly with a 50%-plus backlog surge, double-digit revenue growth, a raised FY2026 outlook, and genuine advanced-reactor optionality directly leveraged to the AI-driven nuclear-power cycle; while the ~40x forward multiple demands flawless execution and the stock trades below our cost basis, the current ~$188 level — well off the highs and below Street targets — offers a reasonable add point for a pinned, high-conviction Diamond name whose thesis is clearly on track.
    BWX Technologies (NYSE: BWXT) is the dominant US manufacturer of naval nuclear reactors and fuel — sole supplier of reactors for the Navy's submarine and carrier fleet — while building out a fast-growing commercial nuclear, microreactor, TRISO-fuel, and medical-isotope franchise.
- BE (researched 2026-06-02) — verdict: **Hold.** The thesis is executing better than at inception — backlog of ~$20bn, guidance raised to $3.4–3.8bn (~80% growth), and a deepening Oracle/Brookfield anchor validate the on-site-power play — but at ~135x forward EPS and a ~$78bn market cap the risk-reward is balanced rather than asymmetric, so we maintain the position at its 1.54% Diamond target while monitoring for either valuation reset or backlog-conversion confirmation, and flag that continued re-rating toward mega-cap scale would warrant a Diamond-outgrowth review.
    Bloom Energy (NYSE: BE) makes solid-oxide fuel cells (the "Bloom Energy Server") that generate on-site, behind-the-meter electricity from natural gas or hydrogen — increasingly the go-to "bring-your-own-power" solution for AI data centers stranded in multi-year grid-interconnection queues.
- AVGO (researched 2026-06-02) — verdict: **Buy.** AVGO is the highest-quality, most-diversified expression of the custom-AI-silicon thesis — dominant ~70% share, six-plus marquee XPU customers, accelerating revenue, and best-in-class 41%-of-revenue free cash flow — and while the valuation is full and customer concentration bears monitoring, the strengthening $100B+ AI line-of-sight more than supports its modest 0.67% Core weight.
    Broadcom (AVGO) designs custom AI accelerators (XPUs) and AI networking silicon for the largest hyperscalers, alongside a broad merchant-semiconductor franchise and a high-margin infrastructure-software business (primarily VMware). It sits in the Diamond AI Fund's Core sleeve as the anchor "picks-and-shovels" play on hyperscaler custom silicon — the dominant ASIC co-design partner controlling…
- ASML (researched 2026-06-02) — verdict: **Buy.** ASML is a high-conviction, irreplaceable EUV/High-NA monopoly delivering 13% YoY growth, 53% gross margins, and accelerating AI-driven order intake against a €38.8B backlog; while the ~44x forward multiple and China/export-control overhang temper the risk-reward, the structural moat and multi-year visibility justify holding — and selectively adding within — the fund's modest 0.67% Core weight.
    ASML Holding (ASML) is the sole global supplier of extreme-ultraviolet (EUV) lithography systems — the machines required to print every leading-edge AI logic and HBM memory chip — which makes it the second chokepoint in the AI supply chain behind TSMC and the rationale for its place in the fund's Semi-cap equipment sleeve.
- ARM (researched 2026-06-02) — verdict: **Hold.** The architectural thesis is squarely intact and arguably strengthening — data center royalty doubling, Armv9/CSS mix uplift, and a credible first-party silicon optionality via the Meta-partnered AGI CPU — but a ~165–190x forward multiple leaves no cushion for the FTC investigation, the Qualcomm dispute, or any licensing-driven quarterly air-pocket, so we maintain the Core 0.67% position and monitor rather than add at current levels.
    Arm Holdings plc (ARM) licenses the CPU instruction-set architecture and increasingly the ready-to-integrate compute subsystems that underpin essentially all Arm-based silicon — and, as of March 2026, sells its own data center chip for the first time.
- APLD (researched 2026-06-02) — verdict: **Hold.** The backlog thesis is intact and arguably strengthening — the Delta Forge 1 lease lifted contracted revenue above $23bn, the CoreWeave credit enhancement de-risks the anchor tenant, and the AI-factory segment is inflecting at +139% YoY — but persistent GAAP losses, heavy debt-funded capex, and ~69% CoreWeave concentration argue for maintaining the pinned 2.21% weight and monitoring tenant diversification and funding execution rather than adding here.
    Applied Digital Corporation (APLD) is a Dallas-based designer, builder, and operator of purpose-built AI-factory data centers, leasing critical IT capacity to hyperscalers and AI-cloud tenants under long-dated, build-to-suit colocation contracts — the "AI landlord" model the Diamond sleeve holds for visible, contracted supplier backlog that diversifies away from CIFR/IREN/NBIS/CRWV.
- APH (researched 2026-06-02) — verdict: **Hold.** The thesis is executing ahead of plan — record AI-led organic growth, a 1.24 book-to-bill, margin expansion, and a raised Q2 guide all validate APH as the fund's copper-interconnect dollar-content vehicle — but with the stock up ~94% over the trailing year and trading ~30% above its 10-year average P/E, we would maintain the Core position at its 0.67% target and monitor for a capacity-unlock or hyperscaler-digestion signal rather than add at current levels.
    Amphenol (NYSE: APH) is the world's second-largest interconnect manufacturer — high-speed copper and fiber connectors, cable assemblies, busbars, and power-distribution systems — that has rapidly repositioned as a tier-one AI infrastructure supplier.
- ANET (researched 2026-06-02) — verdict: **Hold.** The AI Ethernet thesis is on track and the Q1 beat plus raised $3.5B AI target validate the fund's rationale, but a ~43x forward multiple, structural margin pressure, ~one-third revenue concentration in Microsoft and Meta, and accelerating Spectrum-X share encroachment argue for maintaining — not adding to — the current 0.67% Core weight while monitoring hyperscaler capex and Nvidia competitive dynamics.
    Arista Networks (ANET) designs high-performance Ethernet switching and routing for cloud and AI data centers, built around its merchant-silicon-plus-EOS software stack — the purest large-cap play on AI *back-end* networking and the reason it anchors the fund's Networking & optical sleeve.
- AMZN (researched 2026-06-02) — verdict: **Hold.** AMZN's thesis is firmly on track — re-accelerating 28% AWS growth, record margins, a $364B backlog, and a validated Trainium franchise — but its mega-cap status and the fund's anti-concentration design correctly cap it at a modest Core weight, so we maintain the pinned 1.10% target and simply monitor the negative-FCF capex cycle and the deepening Anthropic/OpenAI circularity cluster.
    Amazon (AMZN) is the world's largest cloud-infrastructure provider via AWS, the dominant Western e-commerce platform, and an increasingly important custom-silicon designer through its Trainium accelerators — the full-stack profile that earns it a Core seat in the fund's Hyperscalers & cloud sub-category.
- AMT (researched 2026-06-02) — verdict: **Hold.** The CoreSite interconnection inflection validates the fund's AI/hybrid-cloud rationale and the five-year-low valuation offers a favorable risk-reward, but with DISH churn still compressing 2026 AFFO growth to ~2% and CoreSite only ~11% of property revenue, we see no case to add to a sub-1% Core position — maintain at target weight and monitor the post-DISH 2027 reacceleration.
    American Tower (NYSE: AMT) is one of the largest global communications REITs, operating nearly 150,000 communications sites alongside CoreSite, its wholly owned, interconnection-dense US data-center platform. The fund holds it in the Core sleeve (Data Center REITs band, 0.78% target) not as a tower bet but as the cleanest non-overlapping fourth REIT exposure to the AI/hybrid-cloud interconnection…
- AMD (researched 2026-06-02) — verdict: **Hold.** The accelerator thesis is intact and arguably ahead of inception expectations — Data Center +57% YoY, raised full-year guidance, and anchored OpenAI/Meta GPU commitments — but with the shares at ~59–69x forward earnings, trading above the average analyst target, and carrying circular-financing dilution risk, we see the risk-reward as balanced and would maintain the modest 0.67% Core weight rather than add at current levels, monitoring H2 2026 MI450 delivery milestones and the Nvidia-circularity cluster.
    Advanced Micro Devices (AMD) is a merchant-silicon leader designing data-center CPUs (EPYC), AI accelerators (Instinct GPUs), client/gaming chips (Ryzen, Radeon) and embedded/adaptive products, and sits in the Diamond AI Fund's Core sleeve as the principal GPU-led challenger to Nvidia's accelerator dominance.
- AMAT (researched 2026-06-02) — verdict: **Buy.** A high-conviction Core holding: AMAT delivered a clean record quarter (revenue +11% YoY, EPS beats, margins expanding to 35.1% at Semiconductor Systems), raised its CY26 equipment-growth outlook to >30%, and sits squarely on the GAA/HBM4 chokepoint the thesis was built around — even with the China export-control drag now quantified and largely de-risked via the BIS settlement, the supplier-side AI-capex read-through remains compelling at a forward multiple that is full but justified by accelerating WFE spend.
    Applied Materials (AMAT) is the broadest-line wafer-fabrication-equipment (WFE) vendor, supplying deposition, etch, CMP, ion implant, metrology/inspection, and advanced-packaging tools across the full materials-engineering flow. It sits in the fund's Core sleeve (Semi-cap equipment) at a 0.67% target weight precisely because it is a 2nm gate-all-around (GAA) and HBM/advanced-packaging chokepoint…
- ALAB (researched 2026-06-02) — verdict: **Hold.** The operational thesis is intact and arguably strengthening into the Scorpio scale-up ramp, but with the stock trading above the Street's high-end target at ~110x forward earnings and ~90% of revenue concentrated in five hyperscalers, the risk-reward is balanced rather than compelling — we maintain the pinned 1.94% Diamond weight and monitor for either a multiple reset (add) or a hyperscaler capex wobble (trim).
    Astera Labs designs purpose-built connectivity silicon — PCIe/CXL retimers, smart fabric switches, and memory controllers — that links GPUs, CPUs, and memory inside AI server racks, positioning it as a high-margin "picks-and-shovels" beneficiary of the hyperscaler capex wave.
- AGX (researched 2026-06-02) — verdict: **Hold.** The thesis is intact and arguably strengthening — record backlog, expanding margins, a fortress balance sheet, and a multi-year turbine-scarcity tailwind — but with the shares re-rated to ~66x trailing earnings against a lumpy revenue base and a consensus price target well below spot, we see the risk-reward as balanced rather than compelling, and would maintain the 1.76% position while monitoring backlog conversion and the Q1 FY2027 print closely rather than adding at current levels.
    Argan, Inc. (AGX) is the pure-play engineering, procurement and construction (EPC) contractor building the gas-fired and renewable power plants that increasingly underpin AI data-center load growth — a "picks-and-shovels" play on the electrification bottleneck rather than on compute itself. It sits in the Diamond sleeve (Power, cooling & electrification;
- AEHR (researched 2026-06-02) — verdict: **Hold.** The fundamental thesis is intact and arguably strengthening — record bookings, the validated silicon-photonics win, and a history-making $41M hyperscale ASIC order give multi-year FY27 visibility — but the stock has re-rated to roughly $2.9B against a sub-$50M revenue base and trades 30–70% above virtually every published price target, so at a 1.07% target weight we would maintain the position to retain the differentiated WLBI/silicon-photonics exposure while declining to add at current valuation, monitoring closely for the FY27 order-to-revenue conversion and any sign of hyperscaler program slippage.
    Aehr Test Systems (NASDAQ: AEHR) is a Fremont-based micro/small-cap supplier of wafer-level and package-level test and burn-in systems — its FOX-XP/FOX-NP wafer-level platforms and high-power Sonoma package-level systems — for semiconductors used in AI processors, silicon photonics, data center, automotive, and industrial applications.
- AAOI (researched 2026-06-02) — verdict: **Hold.** The thesis is operationally on track — datacenter revenue more than doubled, FY26 guidance was raised above USD 1.1 billion, and the capacity/backlog story is real — but with the stock up ~441% YTD, trading above the average analyst target at ~14–15x EV/sales while still loss-making and ~98% customer-concentrated, the risk-reward no longer supports adding; we maintain the existing ~1.07% Diamond weight and monitor 1.6T qualification milestones and customer diversification closely, with any guide-down or Digicomm/Microsoft wobble a trim trigger.
    Applied Optoelectronics (AAOI) designs and manufactures vertically integrated optical transceivers — increasingly 800G and 1.6T modules for AI data-center interconnect, alongside a legacy CATV/HFC broadband-access business. It sits in the fund's Networking & optical sleeve as a Diamond name on a clean read on the hyperscaler-capex wave: AOI completed its first volume shipment of 800G products to…
- ZS (researched 2026-06-02) — verdict: **Hold.** We maintain the pinned 0.85% Core position but step back to monitor: the valuation has compressed to a historically cheap ~7x sales and ~38x FY26E EPS, yet the combination of a 16–17% FY27 growth guide, sales-leadership turnover, and a lowered FCF margin removes the near-term conviction needed to add — we want one to two quarters of evidence that organic net-new ARR has stabilized before re-rating our view either way.
    Zscaler (ZS) is the leading cloud-native zero-trust security platform — its Zero Trust Exchange brokers user-to-application traffic in the cloud, displacing the legacy VPN/firewall stack — and sits in the Diamond AI Fund's Core sleeve under AI software & cyber on the thesis that it is the cybersecurity layer for the AI age, with an expanding agentic-SecOps offering post the Red Canary…
- WULF (researched 2026-06-02) — verdict: **Hold.** The pivot thesis is intact and the catalyst path is strong, but with WULF re-rated >100% YTD, trading on backlog NPV rather than earnings, concentrated in two tenants, and counting against the fund's converted-miner sub-limit, we maintain the 1.00% pinned weight and monitor lease commencement and customer credit rather than add at current levels.
    TeraWulf (WULF) is a vertically integrated, energy-advantaged digital-infrastructure operator that is pivoting from bitcoin mining to contracted HPC/AI colocation, anchoring its thesis in the Diamond sleeve's Neocloud / AI-HPC transition sub-category.
- WDC (researched 2026-06-02) — verdict: **Hold.** The thesis is intact and arguably strengthening — sold-out capacity through 2026, contracted visibility to 2028, 50%+ gross margins, and a clean pure-play structure all validate the supply-tight nearline rationale — but after a 900%-plus run the stock now trades through the analyst-target average at ~55x FY26E / ~39x FY27E, leaving little cushion if capacity unlocks, HAMR timing slips behind Seagate, or hyperscaler capex digests; we maintain the position at its 0.67% Core weight and monitor rather than add into strength.
    Western Digital (WDC) is now a pure-play hard-disk-drive (HDD) maker — Western Digital completed the spinoff of SanDisk last year, and has been streamlining its business to become a pure-play HDD firm — selling high-capacity nearline drives to hyperscalers and cloud providers building AI training-data lakes.
- VST (researched 2026-06-02) — verdict: **Buy.** We view VST as a high-conviction hold within the Power generation / nuclear sleeve: the inception thesis is intact and strengthening — record Q1 EBITDA, reaffirmed guidance, fresh investment-grade ratings, and 3.8+ GW of 20-year hyperscaler nuclear PPAs whose 2027+ contribution and the Cogentrix accretion are not yet in numbers — while the ~10x forward EV/EBITDA and ~7% FCF yield leave the contracted AI-demand upside underpriced relative to CEG, even allowing for the gas-weighted fleet and integration/leverage risks we flag. At a pinned 0.83% target weight, the position is appropriately sized for a Core name and well inside the 4% single-name cap, requiring no rebalancing action.
    Vistra Corp (NYSE: VST) is a Fortune 500 integrated retail-electricity and competitive power-generation company with a ~41 GW fleet spanning natural gas, nuclear, coal and solar/storage, concentrated in ERCOT (Texas) and PJM (Mid-Atlantic). It sits in the Diamond AI Fund's Power generation / nuclear sub-category as a Core name because its carbon-free nuclear baseload is being contracted to…
- VRT (researched 2026-06-02) — verdict: **Hold.** Fundamentals remain best-in-class — 30% revenue growth, 430bps of margin expansion, a >$15B backlog and raised FY26 guidance keep the thesis intact — but with the stock at ~51x forward earnings, a ~$121B market cap that places it well into mega-cap territory (a Diamond-outgrowth flag under §2.3), and a missing Q1 backlog disclosure clouding near-term visibility, we maintain the position and monitor rather than add, pending re-confirmation of order momentum at the May investor conference and a review of whether the name should be re-sleeved from Diamond to Core.
    Vertiv Holdings (NYSE: VRT) is the cleanest public pure-play on AI data-center critical infrastructure — power (UPS, switchgear, busway), thermal management (air and liquid cooling), and increasingly converged, prefabricated/modular deployments.
- TSM (researched 2026-06-02) — verdict: **Buy.** TSMC is a high-conviction Core hold: an unrivaled monopoly on the advanced silicon underpinning the fund's entire accelerator complex, compounding 30%-plus revenue growth at 66% gross margins with raised capex and revenue guidance, where the principal risks (valuation re-rating, Taiwan chokepoint concentration) are structural and known rather than signs of a deteriorating thesis.
    Taiwan Semiconductor Manufacturing (TSMC; NYSE: TSM) is the world's dominant pure-play foundry, manufacturing the advanced-node silicon behind essentially every AI accelerator the fund holds — Nvidia GPUs, AMD MI-series, and the Broadcom/Marvell custom ASICs.
- TLN (researched 2026-06-02) — verdict: **Buy.** With Q1 EBITDA more than doubling YoY, reaffirmed 2026 guidance, a 17-year ~$18B AWS PPA de-risking the bulk of Susquehanna's output through 2042, and accretive gas-fleet M&A diversifying cash flows, Talen remains a high-conviction expression of the fund's contracted-nuclear-into-AI-demand thesis — though the 1.33% Diamond weight is appropriate given single-customer concentration and elevated leverage that bear monitoring.
    Talen Energy (NASDAQ: TLN) is an independent power producer (IPP) anchored by the 2.5 GW Susquehanna nuclear plant in Pennsylvania, which it owns 90% and operates, supplemented by a growing PJM gas fleet. It sits in the Diamond AI Fund's Power generation / nuclear sleeve as a direct, contracted play on data-center power demand: Talen restructured and significantly expanded its nuclear agreement…
- TER (researched 2026-06-02) — verdict: **Buy** — Teradyne delivered a high-quality, AI-driven record with a credible merchant-GPU optionality leg and reaffirmed full-year targets, and at a 1.07% Diamond weight we view the customer-concentration and lumpiness risks as adequately sized rather than thesis-breaking, making the post-earnings pullback an attractive entry into a pure-play test lever the sleeve otherwise lacks.
    Teradyne (NASDAQ: TER) is one of the two dominant suppliers of automated test equipment (ATE), where it and Advantest together control roughly 80% of the global semiconductor tester market — a pure-play, picks-and-shovels lever on the AI compute and memory build-out that anchors the fund's Semi-cap equipment & EDA sleeve.
- TEM (researched 2026-06-02) — verdict: **Hold.** The growth thesis is intact and on track — 36% revenue growth, raised guidance, expanding TCV and Insights momentum all validate the fund's vertical-AI rationale at a now far-less-demanding multiple — but persistent GAAP losses, a 34% EPS miss, and an unresolved short-seller/accounting overhang argue against adding to the position until profitability visibility improves and the governance questions clear, so we maintain the 1.00% pinned Diamond weight and monitor closely.
    Tempus AI (NASDAQ: TEM) is a precision-medicine diagnostics and healthcare-data company that pairs genomic/molecular testing (oncology, hereditary, MRD) with a large, de-identified multimodal clinical dataset it licenses to pharma — the "vertical AI" angle that earns it a slot in the fund's Servers/storage/robotics/vertical-AI sub-category.
- SYM (researched 2026-06-02) — verdict: **Hold.** The thesis is intact and advancing — second straight GAAP profit, 23% growth, ~$22.7B backlog, expanding adjusted EBITDA, and a credible diversification path via Exol, Fox Robotics, and Medline — but the >80% Walmart concentration, thin GAAP earnings quality, and recent negative sentiment (Goldman PT cut to $45, insider selling) argue against adding at the pinned 1.00% Diamond weight; we maintain and monitor the next two prints for sustained margin expansion and customer-mix progress.
    Symbotic (Nasdaq: SYM) builds AI-enabled, case- and pallet-handling warehouse-automation systems — autonomous "SymBot" fleets plus the software that orchestrates them — for large retailers and distributors, with Walmart as anchor customer and equity holder.
- STX (researched 2026-06-02) — verdict: **Hold.** The thesis is intact and arguably strengthening — nearline is sold out through CY27, HAMR/Mozaic is driving record margins, and data center demand is broadening — but after a ~600% run the stock discounts much of that good news, so we would maintain the 0.67% Core target and monitor hyperscaler capex commentary rather than add at these multiples.
    Seagate Technology (STX) is one of two remaining scale manufacturers of mass-capacity hard disk drives (HDDs), and the fund holds it as the Core "Storage" expression of the AI-buildout thesis — the bet that hyperscaler exabyte demand outruns nearline supply.
- STRL (researched 2026-06-02) — verdict: **Hold.** The operating thesis is intact and arguably strengthening — backlog, mix, and guidance all moving the right way — but with the stock up ~135% YTD and trading at ~45x NTM adjusted EPS, the risk-reward at the 2.21% pinned weight is balanced rather than compelling; we would maintain the position and monitor for a hyperscaler-capex inflection or a valuation reset before adding.
    Sterling Infrastructure (NasdaqGS: STRL) is the "dirt-work and electrical" front end of the data-center build-out — large-scale site development, grading, and (post-CEC) electrical/mechanical services for the mission-critical campuses that must be readied before any UPS, switchgear, or server rack is installed.
- SNPS (researched 2026-06-02) — verdict: **Hold.** Synopsys remains a high-quality, thesis-consistent Core EDA holding with a strong EDA/verification engine, a cleared China overhang, an activist-validated margin path, and a relatively undemanding multiple versus Cadence — but the unresolved Design IP erosion and Ansys integration overhang argue for maintaining the 0.67% target weight and monitoring the promised 2H IP recovery rather than adding here.
    Synopsys (SNPS) is the larger half of the EDA duopoly — the chip-design software, verification hardware, and IP that every advanced-node and AI-accelerator design relies on — now augmented by the July 2025 close of the ~$35B Ansys acquisition, which extends the platform into multiphysics simulation and system-level engineering.
- SNOW (researched 2026-06-02) — verdict: **Hold.** The thesis is firmly intact and arguably strengthening — 34% product-revenue reacceleration, 38% RPO growth, the $6B AWS commitment, and fast-ramping Cortex/CoCo monetization all validate Snowflake's place in the AI-software sleeve — but with the stock having re-rated ~60% off its April lows to ~16x forward sales, the risk-reward has rebalanced from compelling to merely fair, warranting a maintain-and-monitor stance rather than an add at the 0.85% pinned weight.
    Snowflake (NYSE: SNOW) operates a consumption-based cloud data platform — the "AI Data Cloud" — that lets enterprises store, govern, and query data and, increasingly, run AI/ML workloads (Cortex, Snowflake Intelligence, Cortex Code) directly against that governed data.
- SMR (researched 2026-06-02) — verdict: **Hold** — NuScale retains a genuine, hard-to-replicate moat as the only NRC-approved SMR vendor with the ENTRA1/TVA program as a multi-GW AI-power option, but with no binding orders, first revenue not expected until 2030, a standing ATM dilution overhang, and a Q1 print that confirmed deep, lumpy losses, we see no case to add and would keep the position at its modest 0.97% Diamond weight while monitoring for a binding offtake catalyst that would re-rate the name toward Buy — or, failing that, treat it as the substitution candidate flagged at inception if the sub-category needs a slot.
    NuScale Power (NYSE: SMR) is the only US-listed small modular reactor (SMR) developer with a US Nuclear Regulatory Commission (NRC)-approved design — its light-water NuScale Power Module (NPM), uprated to 77 MWe per module and approved by the NRC on May 29, 2025 for a 462-MW, six-module plant.
- SMCI (researched 2026-06-02) — verdict: **Hold.** The operating thesis remains intact — triple-digit growth, >80% AI-GPU revenue mix, record backlog, a raised FY26 guide, and an undemanding ~18x forward P/E — but the re-escalated DOJ/export-control and securities-litigation overhang, paired with Q4 margin guidance stepping back to ~8.3%, warrants maintaining the small pinned 0.56% Core weight and monitoring closely rather than adding.
    Super Micro Computer (SMCI) is a Total IT Solution Provider building rack-scale AI servers, storage, and direct-liquid-cooling (DLC) infrastructure around Nvidia and AMD accelerators; it sits in the fund's Core sleeve (Servers, storage, robotics, vertical AI) as a high-beta, supply-side read on the hyperscaler capex wave.
- SITM (researched 2026-06-02) — verdict: **Hold.** The operational thesis is intact and arguably strengthening — 88% YoY growth, 158% CED growth, raised guidance and a strategically sound, antitrust-cleared Renesas deal — but with the stock trading well above even bullish Street targets, a newly levered balance sheet, and integration risk ahead, we see the risk/reward as balanced rather than additive at current levels, and would monitor for a valuation reset or deal-close confirmation before adding. (Note for the platform: at a 1.40% Diamond target weight, SITM remains comfortably within the 4% single-name cap;
    SiTime Corporation (SITM) is the pure-play, MEMS-based precision-timing leader — oscillators, clocks, and resonators that function as the heartbeat of electronic systems — and it sits in the Diamond sleeve as a "hidden pick-axe" beneficiary of rising AI-server timing complexity (NVLink scale-out, HBM, optical/CPO clocking).
- RMBS (researched 2026-06-02) — verdict: **Hold** — the franchise quality, share gains, AI-content tailwind, and pristine balance sheet keep us constructive on the thesis, but at ~$16B and ~50x+ NTM non-GAAP the stock has re-rated well ahead of its mid-cap "diamond" entry profile, so we maintain the 1.07% position and monitor for a Diamond-outgrowth/valuation review rather than adding here.
    Rambus is a pure-play memory-interface franchise straddling two businesses: a high-margin patent-licensing/royalty stream and a fast-growing memory-interface chip (product) business — DDR5 registering clock drivers (RCDs), companion chips (PMICs, SPD hubs, temp sensors), MRDIMM/MCRDIMM chipsets, plus a silicon-IP portfolio spanning HBM controllers, PCIe, and security.
- PWR (researched 2026-06-02) — verdict: **Hold.** The thesis is executing cleanly — record $48.5B backlog, a 26% revenue beat, raised guidance, and direct AI-power tailwind — but a ~49x forward multiple after a 118% run, an 8.5% drawdown below our $751.50 cost basis, and the stock's drift into mega-cap territory argue for maintaining the 1.07% Diamond weight and monitoring rather than adding here.
    Quanta Services (NYSE: PWR) is the largest US specialty-infrastructure contractor for electric power transmission, distribution, grid hardening, gas generation, and increasingly the high-voltage interconnection and substation work that AI data centers require.
- POWL (researched 2026-06-02) — verdict: **Hold.** We maintain the position — the backlog at $1.8 billion (+33%), a 1.7x book-to-bill, fiscal-2028 visibility, and the >$400 million behind-the-meter data center order entering Q3 keep the AI-electrification thesis firmly intact — but the ~184% trailing-year re-rating, a forward P/E near 30x, and consensus targets below spot leave the risk/reward balanced rather than compelling at the 1.07% target weight, warranting monitoring rather than adding here.
    Powell Industries (NASDAQ: POWL) designs and manufactures custom-engineered medium-voltage switchgear, electrical houses, breakers, and control systems for the management and distribution of electrical energy. It sits in the Diamond AI Fund's Power, cooling & electrification sleeve as a small-cap pure-play on electrical distribution feeding the data center buildout — a high-conviction…
- POWI (researched 2026-06-02) — verdict: **Hold.** The Diamond thesis is on track and structurally strengthening — POWI's sole-supplier position in high-voltage GaN, the NVIDIA 800V engagement, and a credible $1B 2030 data-center SAM justify retaining the 1.00% pinned position — but a ~65x FY26E non-GAAP multiple on a name where AI revenue is still negligible, set against intensifying competition from Infineon and Navitas, leaves insufficient upside to add here; maintain at target weight and monitor design-win-to-revenue conversion.
    Power Integrations (POWI) is a fabless supplier of high-voltage power-conversion ICs and gate drivers, and — crucially for this fund — the only merchant supplier with 1,250V/1,700V gallium-nitride (GaN) switches in mass production, positioning it directly in the data-center 400V/800V transition that anchors its Diamond thesis.
- POET (researched 2026-06-02) — verdict: **Hold.** We maintain the 0.50% pinned Diamond position — the optical-interposer IP and the Lumilens framework keep POET aligned to the fund's 1.6T/CPO thesis — but the Celestial/Marvell cancellation, the active short-seller and PFIC/litigation overhang, persistent dilution, and a share price trading at roughly 1.7x the Street's own targets argue against adding until 2H26 shipment data converts pipeline into recognized revenue.
    POET Technologies designs optical-interposer-based optical engines and light sources for AI data-center networking, targeting the 800G/1.6T transceiver tier hyperscalers need for next-gen GPU fabrics — exactly the co-packaged-optics (CPO) narrative that anchors its place in the Diamond sleeve's Networking & optical sub-category.
- PLTR (researched 2026-06-02) — verdict: **Hold.** The fundamental thesis is firing on all cylinders — 85% growth, accelerating US commercial, raised guidance, 150% NDR, and 88% gross margins more than validate PLTR's place in the Core AI-software sleeve — but at ~49x forward sales the valuation prices in years of perfection, so we maintain the existing 0.67% target weight and monitor for either a multiple reset (an opportunity to add) or a deceleration in US commercial RDV (a trim signal) rather than adding at current levels.
    Palantir Technologies (PLTR) builds operational AI software — its Gotham, Foundry, and Artificial Intelligence Platform (AIP) products embed data, ontology, and AI agents directly into mission-critical government and enterprise workflows. It sits in the fund's **Core / AI software & cyber** sleeve at a 0.67% target weight, capturing the application layer of the AI value chain — where hyperscaler…
- PENG (researched 2026-06-02) — verdict: **Hold.** The fundamental thesis is intact and even strengthening on the memory/inference axis — 12% FY26 sales-growth guidance, 60%+ AI revenue, and seven first-half AI/HPC logos all support the Diamond-sleeve rationale — but with the stock up ~200% in twelve months and trading roughly double the consensus target while the Advanced Computing leg is down 42% and customer concentration has risen to 66%, we see the risk/reward as balanced rather than additive at current levels; maintain the 0.93% position, bank the gains via scheduled rebalancing back to target, and monitor whether strong bookings convert to Advanced Computing revenue in FY27 before adding.
    Penguin Solutions, Inc. (Nasdaq: PENG) is an end-to-end AI-factory systems company spanning three segments — Advanced Computing (HPC/AI systems, OriginAI, Stratus fault-tolerant platforms), Integrated Memory (SMART Modular specialty DRAM/flash), and Optimized LED — selling into enterprises, governments, and neocloud providers rather than the top-tier hyperscalers, which is precisely the…
- PANW (researched 2026-06-02) — verdict: **Hold** — PANW's platformization, Cortex/Prisma AI traction, and the CyberArk identity expansion keep the fund's AI software & cyber thesis firmly intact, but a 91% YTD re-rating into the June 2 print, a full ~45x forward multiple, and fresh sector-sentiment risk from Zscaler's guide-down argue for maintaining the 0.67% Core weight and monitoring the Q3 result rather than adding here.
    Palo Alto Networks (PANW) is the largest pure-play cybersecurity vendor, spanning network security (Strata), cloud/SASE (Prisma), and AI-driven security operations (Cortex XSIAM), now augmented by identity security following the February 2026 close of its CyberArk acquisition.
- ORCL (researched 2026-06-02) — verdict: **Hold.** The demand thesis and $553B backlog remain firmly intact and the position is a small, pinned Core weight (1.10% target), but with FCF deeply negative through FY28, debt above $100B, a single-customer dependence now flashing (OpenAI's April miss), and the decisive Q4 print due 2026-06-10, we see no case to add ahead of that catalyst and would maintain-and-monitor rather than chase the late-May rally.
    Oracle (ORCL) is a legacy enterprise database and applications vendor that has reinvented itself as a fourth hyperscaler, with Oracle Cloud Infrastructure (OCI) now the central growth engine and the reason it sits in the fund's Hyperscalers & cloud sleeve.
- ONTO (researched 2026-06-02) — verdict: **Hold.** The thesis is intact and arguably strengthening — record revenue, a Q1 beat, accelerating Dragonfly G5 ramp, confirmed share gains versus KLA, and a sensible Rigaku expansion — but at ~36x forward earnings with a recent de-rate, concentrated memory/foundry customers, and Rigaku integration/leverage to digest, the risk-reward is balanced rather than compelling at the current 1.01% Diamond weight; we would maintain and monitor the G5 ramp and Q2 print before adding.
    Onto Innovation (ONTO) is a Wilmington, MA-based supplier of metrology, inspection, lithography, and process-control software for semiconductor front-end and advanced-packaging applications — the inspection chokepoint where HBM stacks, 2.5D logic, and gate-all-around nodes get qualified.
- ON (researched 2026-06-02) — verdict: **Hold.** The thesis is demonstrably on track — AI data-center revenue is doubling, the cyclical trough is past, and margins are expanding — but with the stock up ~88% YTD and trading at ~40x forward earnings versus mid-teens peers, the risk-reward is now balanced rather than compelling, warranting maintenance of the existing 1.09% Core weight while monitoring AI revenue cadence and auto demand rather than adding at these levels.
    onsemi (Nasdaq: ON) is a US-domiciled power and sensing semiconductor maker — silicon carbide (SiC), IGBTs, analog/mixed-signal, and CMOS image sensors — historically levered to automotive and industrial, but increasingly to AI data-center power delivery.
- OKLO (researched 2026-06-02) — verdict: **Hold.** The regulatory progress (PDC approval), the prepaid Meta/Equinix structures, and a fortress $2.54bn balance sheet keep the inception thesis intact and justify the existing 0.83% Core position, but with the company still pre-revenue, first power two-plus years out, the bulk of the 14 GW pipeline non-binding, and the equity richly valued, we see insufficient near-term margin of safety to add — maintain at target weight and monitor the COLA cadence, Meta-deal conversion, and cash-burn trajectory.
    Oklo Inc (NYSE: OKLO) is a pre-revenue advanced-fission developer building its ~75-MWe Aurora "powerhouse" fast-reactor under a build-own-operate model, selling power-as-a-service through long-term power purchase agreements (PPAs) rather than selling reactors outright, with a secondary ambition in nuclear-fuel recycling.
- NVTS (researched 2026-06-02) — verdict: **Hold.** We maintain the fund's small 0.50% pinned Diamond position for its differentiated, unhedgeable read-through to the 800V HVDC rack-power transition and genuine Nvidia design engagement, but the equity trades at a triple-digit sales multiple roughly 90% above consensus PT on a sub-$10mn quarterly revenue base, so we would not add until 2027 monetization shows tangible traction — and we flag the position for trim on any further multiple expansion or capital raise that dilutes the thesis.
    Navitas Semiconductor (NVTS) is a micro-cap gallium-nitride (GaN) and silicon-carbide (SiC) power-IC vendor mid-pivot into AI data-center power delivery — the "Navitas 2.0" strategy — anchored by a high-profile design collaboration with Nvidia on 800V high-voltage DC (HVDC) rack power architecture.
- NVT (researched 2026-06-02) — verdict: **Hold.** Execution and AI data-center backlog visibility are excellent and the original sleeve thesis is intact, but with the stock trading at ~38x FY26E adjusted EPS near 52-week highs, copper/tariff margin recovery still to prove out, and the comp base hardening into H2, we see balanced risk-reward and would maintain — not add to — the 0.67% Core position while monitoring Q2 guidance for the next leg.
    nVent Electric (NYSE: NVT) is a mid-cap maker of electrical connection, protection, and enclosure systems whose Systems Protection segment has become a fast-growing supplier of liquid-cooling and "gray-space/white-space" infrastructure for AI data centers.
- NVDA (researched 2026-06-02) — verdict: **Buy.** With Q1 FY27 delivering an all-round beat ($81.6B revenue, +85% YoY; 75% gross margin), a $91.0B Q2 guide that excludes China, and ~$1T of Blackwell/Vera Rubin orders underpinning multi-year visibility — all at an undemanding ~21x forward P/E (PEG ~0.48) — NVIDIA remains a high-conviction core holding, with its modest 0.67% fund weight already prudently capping the single-name and Nvidia-circularity risk that warrant ongoing monitoring.
    NVIDIA is the foundational compute platform of the AI build-out — the merchant GPU, networking, and CUDA software stack that hyperscalers and neoclouds rent capacity on — and sits in the Diamond AI Fund's Core sleeve as the "hyperscaler-capex tax collector" anchoring the Semiconductors & AI chips sub-category at a deliberately modest 0.67% target weight.
- NRG (researched 2026-06-02) — verdict: **Hold.** NRG offers a fairly-valued, genuinely differentiated gas-fired AI-power optionality that fits its Core sub-category role, but a soft Q1, reaffirmed-not-raised guidance below consensus, a fresh CEO, post-LS-Power leverage, and a 2029-weighted build-out leave near-term catalysts thin — we maintain the pinned 0.83% target and monitor for the expected 2026 data-center contract announcement as the next conviction trigger.
    NRG Energy (NYSE: NRG) is a Houston-based integrated independent power producer (IPP) and competitive retailer that, following its ~$12B LS Power acquisition, now operates roughly 25 GW of generation (predominantly natural-gas-fired) and serves ~6mn-plus retail customers, with a Vivint Smart Home consumer-services arm bolted on.
- NOW (researched 2026-06-02) — verdict: **Buy.** The thesis that justified inclusion — dominant, monetized enterprise agentic-AI exposure diversifying the software sleeve — is intact and arguably strengthening (Now Assist ACV inflecting toward a raised $1.5B target, 22% subscription growth, expanding margins), while the ~50% drawdown has compressed the forward multiple to ~24.6x, leaving a favorable risk/reward at a 0.67% Core weight provided federal-spend weakness and SaaS-displacement fears prove cyclical rather than structural — both of which bear close monitoring into Q2 FY26 results.
    ServiceNow (NOW) operates an enterprise workflow-automation platform — the Now Platform — spanning IT, HR, customer service, and security workflows, now repositioned by management as the "AI control tower" for agentic enterprise AI. It sits in the fund's **Core / AI software & cyber** sleeve at a 0.67% target weight because, alongside Palantir, it is the clearest large-cap vehicle for *monetized*…
- NET (researched 2026-06-02) — verdict: **Hold.** We maintain the position given Cloudflare's accelerating 34% growth, a credible agentic-AI platform thesis validated by the Anthropic win and record ACV, and clean alignment with the fund's AI-software mandate — but at low-30s EV/sales (well above both peers and analyst targets), with gross margins compressing and a disruptive 20% restructuring still to be digested, the risk-reward at $270 no longer warrants adding, and we would monitor the Q2 print and Analyst Day before revisiting the weight.
    Cloudflare (NYSE: NET) operates a global edge network spanning connectivity, security (WAF, DDoS, Zero Trust/SSE), and a serverless developer platform (Workers, Workers AI, AI Gateway, Durable Objects) increasingly positioned as the execution and security layer for AI agents.
- NBIS (researched 2026-06-01) — verdict: **Buy.** We maintain the Diamond-sleeve overweight: NBIS pairs the fund's strongest attribute — contractually visible, multi-year hyperscaler-backed revenue (~$50B backlog, $7–9B exit ARR) — with a genuine purpose-built, full-stack differentiation that keeps it clear of the converted-miner sub-limit, and although valuation is stretched and execution/concentration risks are real, the 1.84% pinned target appropriately sizes a high-conviction, high-variance position rather than overcommitting to it.
    Nebius Group (NBIS) is an Amsterdam-headquartered, vertically integrated "AI-native hyperscaler" — a neocloud renting GPU compute and a full-stack AI platform (Aether cloud, Token Factory inference) to AI builders and hyperscalers. It sits in the Diamond sleeve under Neoclouds as a high-conviction, smaller-cap expression of the hyperscaler-capex thesis, and — crucially for the fund's…
- MU (researched 2026-06-01) — verdict: **Hold.** The thesis is firmly intact and the fundamentals are exceptional — contract-backed, sold-out HBM through calendar 2026, record AI-driven results, and a >$25B capex commitment that confirms hyperscaler read-through — but with the stock having re-rated to ~$1,035 (a ~28% gain over the $807 cost basis), near-universal Street bullishness, and peak-cycle margins embedded in the forward multiple, we would maintain the pinned 0.85% Core weight and monitor for capacity-unlock and pricing-rollover signals rather than add at these levels.
    Micron Technology (MU) is one of only three scaled producers of DRAM and NAND memory globally — and, critically for this fund, one of three makers of high-bandwidth memory (HBM), the binding supply constraint on AI training and inference. It sits in the Diamond AI Fund's Core sleeve (Semiconductors & AI chips, 0.85% target) as a direct, supply-constrained play on the hyperscaler capex wave rather…
- MSFT (researched 2026-06-01) — verdict: **Hold.** The thesis is firmly intact — 40% Azure growth, a $37B AI run-rate (+123%), and a $627B RPO confirm the contracted, supply-led demand the fund is built around — but as a pinned 1.10% Core name deliberately capped to express the hyperscaler-capex read-through via suppliers rather than direct mega-cap ownership, the correct action is to maintain the position at target while monitoring margin compression and the $190B capex digestion, not to add.
    Microsoft is the world's second-largest hyperscaler, anchoring the Diamond AI Fund's **Hyperscalers & cloud** sleeve as a Core, pinned name at a deliberately modest 1.10% target — the fund gets its leveraged read-through to the AI capex cycle through the *supplier* base, not through direct mega-cap ownership.
- MRVL (researched 2026-06-01) — verdict: **Buy.** With Q1 FY27 delivering 28% YoY record revenue, a $1.5bn upward revision to the FY28 target, an expanding custom-silicon socket base, and direct Nvidia ecosystem validation, the inception thesis is intact and inflecting — and at a 0.67% Core weight the position size leaves room to add even after the re-rating; we would, however, monitor the Nvidia-circularity cluster against the fund's ~12% soft cap and watch custom-mix margin dilution as the ramp accelerates.
    Marvell Technology (MRVL) is a fabless semiconductor supplier of custom silicon (ASICs), optical/electrical interconnect, switching, and storage solutions positioned squarely at the infrastructure layer of the AI build-out — exactly the "supply-constrained value chain" exposure the Diamond AI Fund is built to capture.
- MPWR (researched 2026-06-01) — verdict: **Hold.** The fundamentals and backlog firmly validate the AI-power thesis — a 26% revenue beat, an 85% enterprise-data growth floor, and a $900mn Q2 guide ~10% above consensus — but with NVIDIA content normalizing toward ~50%, a securities-fraud claim now cleared to proceed, and a ~52x forward multiple offering only low-single-digit consensus upside, we see the risk-reward as balanced rather than additive at current levels; we would maintain the pinned 0.84% position and monitor share-loss data points and the litigation rather than add.
    Monolithic Power Systems (MPWR) is a fabless analog and mixed-signal chipmaker specializing in highly integrated power-management ICs and, increasingly, full power-delivery modules and systems — the "invisible" power layer feeding AI accelerators, server CPUs, optical modules, and switches.
- MOD (researched 2026-06-01) — verdict: **Hold.** The thesis is firmly intact and arguably strengthening — record data-center growth (+119% FY26), a $4B five-year hyperscaler LTA, FY27 data-center guidance of +60–80%, and a pure-play-creating spin-off — but with the stock at ~57x trailing adjusted EPS, gross margins compressing ~320bps on expansion costs, rising customer concentration (~11% single customer), and a negative post-earnings reaction, the risk-reward at the current ~$289 level is balanced rather than compelling, arguing for maintaining the 1.07% target weight and monitoring margin recovery and the Gentherm close rather than adding.
    Modine Manufacturing (NYSE: MOD) is a Racine, Wisconsin-based thermal-management specialist whose data-center cooling franchise — chillers, coolant distribution units, and air-side systems for hyperscale buildouts — has become the dominant value driver, justifying its place as a Diamond-sleeve, non-Vertiv exposure within Power, cooling & electrification.
- MIR (researched 2026-06-01) — verdict: **Buy.** We maintain the Diamond-sleeve position at its 1.74% target: the order/backlog inflection ($288M orders, $1.1B backlog) and accelerating SMR content validate the picks-and-shovels nuclear thesis, and while the rich ~26.6x EV/EBITDA multiple and Q1 margin dilution warrant monitoring, the small-cap-relative, horizontally differentiated exposure is exactly the "diamond in the rough" the fund is designed to capture — and at a $4.67B market cap it remains comfortably within the Diamond profile.
    Mirion Technologies (NYSE: MIR) is the radiation-detection, measurement, and instrumentation layer of the nuclear stack — a "picks-and-shovels" supplier to both the existing reactor fleet and the SMR build-out, spanning a Nuclear & Safety segment and a Medical (radiation therapy QA, dosimetry) segment.
- META (researched 2026-06-01) — verdict: **Hold.** The thesis is intact and the capex profile (now confirmed at $125–145B for 2026, multi-GW Hyperion commitments) precisely validates the fund's reason for owning the name — but Meta belongs in the Core/Hyperscaler sleeve at its modest pinned 1.10% weight by design, so we maintain and monitor rather than add, watching capex-ROI signals, the EU DSA fine, and any further DAP softening as the key variables into Q2.
    Meta Platforms (META) is the world's dominant social-media advertising business — Facebook, Instagram, WhatsApp, Messenger, and Threads — and increasingly an AI-infrastructure builder spending at hyperscaler scale. It sits in the Diamond AI Fund's **Core / Hyperscalers & cloud** sleeve at a deliberately modest 1.10% target weight, capturing read-through to the capex cycle without overweighting a…
- MDB (researched 2026-06-01) — verdict: **Buy.** The Q1 FY27 beat-and-raise — 25% revenue growth, 29% Atlas growth, 121% net ARR expansion, 88% RPO growth, first GAAP profit, and a Voyage AI/agentic product story now in production — confirms the fund's thesis is firmly intact, and while the ~12x forward EV/sales leaves limited valuation cushion, the accelerating Q2 guide and reaccelerating consumption justify maintaining (and, given the sub-1% weight, potentially adding toward) the Core target.
    MongoDB (NASDAQ: MDB) is the leading document-database vendor, monetizing primarily through Atlas — its fully managed multi-cloud database-as-a-service — and positioning itself as the operational data layer for production AI and agentic applications.
- LSCC (researched 2026-06-01) — verdict: **Hold.** The thesis is executing ahead of plan — 42% Q1 revenue growth, record data-center-AI-driven Compute and Communications, rising server attach, and an AMI deal that credibly delivers the $1B+ run-rate milestone — but a ~1000x TTM / ~70–80x NTM multiple and fresh integration/leverage risk argue for maintaining the existing 1.07% Diamond weight and monitoring rather than adding into strength.
    Lattice Semiconductor (LSCC) is the merchant leader in low-power, small-footprint FPGAs (its Nexus and AVANT families), built on mature process nodes and increasingly positioned as the "companion chip" attaching to CPUs/GPUs in AI servers for power sequencing, security and platform control.
- LRCX (researched 2026-06-01) — verdict: **Hold.** The thesis is intact and arguably strengthening — record results, a raised $140B WFE outlook, the HBM tool-of-record win, and improving China risk profile all validate the etch-and-deposition-chokepoint logic — but with the stock up ~248% YoY, trading at ~36x forward earnings above its historical range and slightly above revised fair value, and with the sub-category already over the fund's ~8-name soft guidance, we see no compelling case to add at current levels and flag LRCX as the natural substitution candidate should a higher-conviction, less-extended Diamond name in semi-cap warrant the slot.
    Lam Research (LRCX) is the etch-and-deposition specialist that gates capacity additions in 3D NAND and high-bandwidth memory (HBM) — precisely the memory tiers most directly bottlenecked by AI training and inference. It sits in the fund's Core sleeve at a 0.78% target weight as the equipment chokepoint complementing the demand-side memory/ASIC names (MU, AVGO).
- LITE (researched 2026-06-01) — verdict: **Hold.** The fundamental thesis — AI optics leadership, NVIDIA-backed capacity, >$400mn OCS backlog and a ramping CPO order — is intact and arguably strengthening, but with the stock having re-rated to ~59x FY27E EPS, ~40% revenue concentration in two customers, a textbook Nvidia-circularity profile, and only breakeven-ish free cash flow, the risk-reward no longer supports adding; we maintain the existing 1.07% Diamond weight and monitor sleeve-level Nvidia-circularity exposure and any hyperscaler capex inflection.
    Lumentum Holdings (LITE) is a San Jose–based supplier of optical and photonic components, modules and subsystems — laser chips, EML/200G transceivers, pump and narrow-linewidth lasers, optical circuit switches (OCS) and co-packaged optics (CPO) — that sit directly in the AI data-center interconnect stack.
- LEU (researched 2026-06-01) — verdict: **Hold.** The thesis remains intact — Centrus is the proven domestic enrichment choke-point with a $3.9bn backlog, a $900M DOE HALEU award, and tightening alignment to advanced-reactor and AI-power demand — but a ~60x multiple, declining near-term EPS, lumpy contract timing, and DOE/appropriations dependence argue for maintaining the pinned 1.33% Diamond position and monitoring milestone execution rather than adding at current levels.
    Centrus Energy is the only US-licensed commercial producer of enriched uranium and the sole domestic supplier delivering High-Assay Low-Enriched Uranium (HALEU) to the government — the fuel that SMRs and advanced reactors require — placing it squarely at the choke-point of the domestic advanced-reactor fuel chain that underpins its Diamond-sleeve role in the fund.
- KLAC (researched 2026-06-01) — verdict: **Buy.** KLA is a high-conviction Core holding whose process-control franchise — number-one share growing seven-times its nearest rival, accelerating advanced-packaging/HBM exposure, and high-teens FY26 growth guidance — captures the fund's AI semi-cap thesis cleanly; the full valuation and China export-control overhang are real but, at a sub-1% target weight, are appropriately sized risks we would hold through rather than trim.
    KLA Corporation (KLAC) is the dominant supplier of process-control, inspection, and metrology equipment to the semiconductor industry — the tools that detect defects and police yield across wafer fabrication and, increasingly, advanced packaging.
- IRM (researched 2026-06-01) — verdict: **Hold.** The thesis is firmly on track — data center +47% and ALM +92% in Q1 2026, a guidance raise, and a 100MW+ leasing runway validate the original case — but with the stock near consensus targets at ~22x 2026E AFFO, carrying 40% net-debt/EV leverage and structural hyperscaler-concentration beta, we see fundamentals and valuation broadly balanced and prefer to maintain the existing 0.67% Core weight rather than add into strength.
    Iron Mountain (IRM) is a global information-management REIT whose investable narrative has shifted decisively from legacy records storage toward higher-growth "growth businesses" — data center colocation, digital solutions, and asset lifecycle management (ALM) — that ride the hyperscaler buildout.
- IREN (researched 2026-06-01) — verdict: **Buy.** The thesis is tracking ahead of inception — the Microsoft contract is now IG-financed at a 3.31% all-in cost, the NVIDIA partnership lifts the forward ARR target to $4.4bn, and IREN's owned-power vertical integration is precisely the scarce-supply exposure the fund is built to capture; we would hold the Diamond overweight while flagging IREN's new NVIDIA equity tie for inclusion in the fund's Nvidia-circularity review and monitoring early-2027 commissioning timelines for slippage.
    IREN Limited (NASDAQ: IREN) is a vertically integrated AI cloud provider that owns grid-connected land, power, and data centers — built on a Bitcoin-mining foundation it is now actively converting into GPU capacity. It sits in the Diamond AI Fund's Neocloud sleeve as a converted-miner pivoting to contracted hyperscaler hosting, the original thesis anchored on a five-year, USD 9.7 billion…
- INOD (researched 2026-06-01) — verdict: **Buy.** We maintain the 1.00% Diamond position: INOD is executing materially ahead of plan — 54% revenue growth, a near-doubling of adjusted EBITDA, raised FY26 guidance to ~40%+, improving (if still elevated) customer concentration, and a now-monetized agentic-evaluation platform that validates the original thesis — and while the ~44x forward multiple leaves little margin for a stumble, the upside-skewed guidance and visible pipeline justify the high-conviction overweight at a sub-1% portfolio weight.
    Innodata Inc. (INOD) is the picks-and-shovels supplier of training data, fine-tuning datasets, and — newly — agent evaluation/observability tooling for the hyperscalers and frontier AI labs building foundation models, which is precisely why it sits in the Diamond AI Fund's AI-software-&-cyber sleeve as a high-conviction, smaller-cap "diamond." On the latest print the thesis is firmly on track…
- HUBB (researched 2026-06-01) — verdict: **Hold.** HUBB is executing well with a genuine, visible data-center and grid-modernization tailwind that validates its place in the Power, cooling & electrification sleeve, but a full ~24–27x forward multiple, flattish near-term HES margins, and the market's demonstrated unwillingness to reward beats argue for maintaining the small Core position and monitoring rather than adding here.
    Hubbell Incorporated (NYSE: HUBB) is a US-centric manufacturer of electrical and utility infrastructure components, split across two segments — Hubbell Utility Solutions (HUS, ~$3.7B annual sales: grid hardware, transmission/substation gear, and grid automation) and Hubbell Electrical Solutions (HES, ~$2.1B: power-distribution and balance-of-system electrical products serving data centers, light…
- HPE (researched 2026-06-01) — verdict: **Buy** — Q2 FY26 was a high-quality, all-round beat with raised guidance, accelerating networking and AI-server revenue, doubling Cloud & AI margins, and a repaired balance sheet, validating the fund's AI-server-plus-Ethernet thesis at an undemanding mid-teens forward multiple, while we monitor networking-margin sustainability and 2H backlog conversion as the key swing factors.
    Hewlett Packard Enterprise (HPE) is a global enterprise IT vendor spanning compute servers, storage, high-performance computing, and — post-Juniper — a materially larger networking franchise, positioning it as a full-stack AI infrastructure supplier.
- GOOGL (researched 2026-06-01) — verdict: **Hold.** The operating thesis is excellent and accelerating — Cloud at +63% with a $462B backlog and a 32.9% margin is exactly the supply-constrained AI exposure the fund seeks — but with the stock near all-time highs, a low-quality headline EPS, an unusual $80B dilutive equity raise landing on the report date, and a live antitrust appeal, we would maintain the pinned 1.10% Core weight and monitor the raise's reception rather than add here.
    Alphabet is the parent of Google Search, YouTube, the Android/Chrome platform stack, and Google Cloud — and, increasingly, a vertically integrated AI supplier whose Gemini models run on its own Tensor Processing Unit (TPU) silicon. It sits in the Diamond AI Fund's Core sleeve as a Hyperscalers & cloud name at a deliberately modest 1.10% target weight, consistent with the fund's anti-concentration…
- GLW (researched 2026-06-01) — verdict: **Hold.** The AI-optical thesis is executing exceptionally well — accelerating Optical growth, multiple contracted hyperscaler deals, and the Nvidia partnership all validate the chokepoint thesis — but with the stock at ~53–56x forward earnings near all-time highs and Street targets implying limited upside, we see the risk-reward as balanced rather than compelling, and would maintain the 0.67% Core weight without adding here while monitoring the Nvidia-circularity cluster and any sign of hyperscaler capex digestion.
    Corning Incorporated (GLW) is the world's leading optical-fiber and cable maker and a materials-science specialist spanning optical communications, display, specialty materials, automotive, and solar — but the investment story has narrowed sharply to one theme: it has become the fiber-and-connectivity chokepoint for hyperscaler AI campus build-outs, which is precisely why the fund holds it in its…
- GFS (researched 2026-06-01) — verdict: **Hold.** GF is executing well — a clean margin-led Q1 beat, a credible photonics roadmap, an inaugural dividend, and deepening design-win and IP-driven differentiation — but with organic growth only ~4%, photonics still a small (if fast-growing) sliver of revenue, and the stock re-rated to a high-50s P/E near its 52-week high, the risk-reward is balanced rather than compelling; we would maintain the 1.56% Diamond position and monitor photonics ramp and margin progression rather than add at current levels.
    GlobalFoundries (GFS) is a specialty/non-leading-edge foundry that, unlike TSMC, competes on differentiated process platforms — RF SOI, FD-SOI (FDX), SiGe, BCD power, eNVM, and increasingly silicon photonics — rather than on bleeding-edge logic nodes.
- GEV (researched 2026-06-01) — verdict: **Hold.** The fundamental thesis is firmly on track — record orders, a $163B backlog, sold-out turbine slots into 2030, and a raised FCF outlook — but with the stock trading near the consensus median target on a high-20s forward P/E and a sell-side downgrade flagging that catalysts are largely priced in, we see balanced risk-reward and would maintain the Core 0.67% weight rather than add at current levels.
    GE Vernova (GEV) is the pure-play power-and-grid spinoff from General Electric, operating three segments — Power (gas turbines, nuclear, hydro), Electrification (grid equipment, transformers, HVDC), and Wind — and sits in the fund's Core sleeve under Power, cooling & electrification at a 0.67% target weight.
- FORM (researched 2026-06-01) — verdict: **Hold.** The HBM4 super-cycle thesis is intact and over-delivering — record revenue, margin model achieved early, and a widening HBM/logic customer base — but with the stock having rerated ~130% YTD to ~48–52x NTM EPS, ~6% below our cost basis, and customer concentration near 40% in three accounts, we see balanced risk-reward at a 1.07% Diamond weight and would maintain rather than add until either a pullback or further HBM4 volume confirmation improves the entry.
    FormFactor (FORM) is the global leader in probe cards — the precision interfaces that test semiconductor wafers — with a particular franchise in DRAM/HBM and Foundry & Logic test, where rising AI test intensity is the structural tailwind. The name sits in the fund's Diamond sleeve under Semi-cap equipment & EDA (HBM probe cards) at a 1.07% target weight, capturing the thesis that HBM4 roughly…
- FN (researched 2026-06-01) — verdict: **Hold.** The thesis is intact and arguably strengthening — record 39% YoY growth, a supply-not-demand datacom constraint, two qualified 800G hyperscaler programs, and falling Nvidia concentration that improves the fund's diversification rationale — but with the stock near 52-week highs at ~47x FY26E non-GAAP EPS and FX/cash-flow headwinds flagged into Q4, we see the risk-reward as balanced at the current 2.00% target weight; maintain the position and monitor the Q4 print and the 800G/CPO ramp into FY27 rather than adding here.
    Fabrinet (NYSE: FN) is the dominant Thailand-based optical original-design-manufacturer (ODM), providing advanced optical packaging and precision electro-mechanical and electronic manufacturing to OEMs of optical communication modules, transceivers, industrial lasers, and increasingly AI/HPC hardware.
- FLEX (researched 2026-06-01) — verdict: **Hold.** The AI-infrastructure thesis is validated and the CPI spin-off is a genuine value-unlock catalyst, but with the stock having already re-rated ~34x NTM EPS to roughly the consensus target and FY27 guidance leaving little execution cushion, the risk-reward is now balanced — we maintain the position at its 1.01% Core target and monitor for either spin-off execution detail or a hyperscaler-capex wobble before adding.
    Flex Ltd. (FLEX) is one of the world's largest advanced-manufacturing / EMS providers, but the relevant story for this fund is its rapidly scaling AI data-center power, cooling, and electrical-infrastructure franchise. The name sits in the Power, cooling & electrification sub-category, and the original thesis — that the NVIDIA-co-developed 800 VDC Power Rack and Omniverse DSX reference designs…
- FIX (researched 2026-06-01) — verdict: **Hold.** The fundamental thesis is firing on every cylinder — record backlog, 56% data-center mix, record margins, strong organic growth and cash generation — but at ~53x trailing / ~43x forward earnings, a ~$64B market cap, and extreme tech-customer concentration heading into tougher 2H comps, the risk-reward no longer supports adding; we maintain the Core 0.67% weight and monitor for hyperscaler-capex and margin-normalization signals.
    Comfort Systems USA (NYSE: FIX) is a leading US mechanical and electrical (MEP) contractor that has transformed into one of the purest public proxies for data-center construction. It sits in the Diamond AI Fund's Core sleeve under Power, cooling & electrification at a 0.67% target weight, filling the data-center MEP build-out gap alongside electrification names like Quanta (PWR).
- ETN (researched 2026-06-01) — verdict: **Hold.** The thesis is intact and arguably stronger than at inception — 240% data center order growth, a 228 GW backlog, and raised FY26 guidance validate the fund's supplier-backlog logic — but a full ~30x forward multiple, near-term margin compression that must reverse on schedule, elevated post-deal leverage, and CEO insider selling collectively argue for maintaining the position at its 0.67% Core weight and monitoring Q2 margin recovery rather than adding here.
    Eaton Corporation (NYSE: ETN) is an intelligent power management company supplying "grid-to-chip" electrical infrastructure — switchgear, busway, UPS, power distribution, and now liquid cooling — across data center, utility, industrial, aerospace, and commercial end-markets.
- EQIX (researched 2026-06-01) — verdict: **Hold.** The thesis remains intact — record AI bookings, raised guidance, a 51% EBITDA margin and a doubling-capacity capex plan validate Equinix's role as the fund's premier interconnection REIT — but consecutive consensus misses, a full ~25x forward-AFFO multiple, and the deliberately Core 1.25% weight argue for maintaining the position and monitoring booking conversion and capex-driven FCF, not adding.
    Equinix (EQIX) is the world's largest carrier-neutral interconnection and colocation operator, with 10,000+ customers across 260+ IBX data centers in 70+ metros and more than 500,000 interconnections — a dense, network-effect ecosystem that anchors the Diamond AI Fund's Data Center REITs sub-category.
- ENTG (researched 2026-06-01) — verdict: **Hold.** The content-growth and margin-inflection thesis is intact and the moly/HBM read-through validates the original case, but with the stock at ~40x FY26E after a near-79% twelve-month run, limited near-term revenue acceleration outside the ~30% AI core, and a 1.07% target weight that already reflects appropriate Diamond-sleeve sizing, we see balanced risk-reward and would maintain rather than add at current levels — revisiting on either a multiple reset toward the low-$120s or evidence that the 2027 capacity ramp is converting to faster top-line growth.
    Entegris (NASDAQ: ENTG) is a leading supplier of critical advanced materials and process solutions — CMP slurries and pads, deposition and selective-etch chemistries, specialty gases, and liquid filtration — that sit as recurring consumables across the semiconductor manufacturing flow.
- EME (researched 2026-06-01) — verdict: **Hold.** The thesis is intact and arguably strengthening — record backlog, a beat-and-raise, and ~30% data-center mix confirm EMCOR as a clean AI pick-and-shovel — but with the stock having re-rated ~88% in twelve months to ~28x earnings and the position sitting comfortably within its 1.21% Core target, we would maintain rather than add, monitoring hyperscaler capex commentary and mechanical-margin mix as the key signals for any change in stance.
    EMCOR Group (EME) is a US specialty electrical and mechanical construction and facilities-services contractor — the "pick-and-shovel" trade base that builds the power, cooling, and electrical guts of hyperscale and colocation data centers. It sits in the Diamond AI Fund's Core sleeve under Power, cooling & electrification at a 1.21% target weight, capturing the supply-side AI capex thesis without…
- DELL (researched 2026-06-01) — verdict: **Buy.** With the AI backlog rebuilt to a record $51.3B, the FY27 AI-server guide raised to ~$60B, and a 64% EPS beat all confirming that supply — not demand — remains the constraint, Dell is executing squarely against the fund's supplier-side thesis; we are comfortable maintaining the pinned 0.92% Core weight despite a full trailing multiple, while monitoring AI-mix margin trajectory and the post-run insider-selling signal.
    Dell Technologies (DELL) is the world's largest server and storage OEM, now repositioned as a leading end-to-end AI-infrastructure integrator selling NVIDIA-, AMD-, and custom-silicon-based "AI-optimized" servers alongside its legacy PC (CSG) and traditional infrastructure (ISG) franchises.
- DDOG (researched 2026-06-01) — verdict: **Hold.** The fundamental thesis is intact and arguably strengthening — broad-based 32% growth, a clean guidance raise, and genuine new AI-training wins — but with the stock up ~66% YTD trading at ~20x ARR against a consensus target below the current price, top-customer concentration explicitly flagged by management, and the sub-category already over the ~8-name guidance, we see this as a name to maintain and monitor as a substitution candidate rather than add to at current levels.
    Datadog (DDOG) is the leading AI-powered observability and security SaaS platform, monitoring infrastructure, applications, logs, and increasingly AI/LLM workloads for ~33,200 customers. It sits in the fund's Core sleeve (AI software & cyber) as the cleanest B2B-software read on AI-workload monetization: native LLM-observability integrations meter incremental spend as customers push inference…
- CSCO (researched 2026-06-01) — verdict: **Buy.** The inception thesis is not just intact but strengthening — FY26 AI orders raised to $9B and AI revenue to $4B, five fresh hyperscaler design wins, 35% product-order growth, and secured silicon supply — and at a ~6x forward P/S discount to Arista the name remains an attractive, lower-beta Core anchor for the Networking & optical sleeve, with customer concentration and margin mix the key items to monitor at its modest 0.67% target weight.
    Cisco Systems (CSCO) is the incumbent leader in enterprise networking — switching, routing, security, and increasingly AI data-center infrastructure via its Silicon One merchant-silicon and Acacia optics franchises — held in the fund's Networking & optical sleeve as the Core, large-cap read-through to the hyperscaler buildout. The inception thesis (FY26 AI orders raised from $5B to $9B;
- CRWV (researched 2026-06-01) — verdict: **Hold.** The demand thesis is intact and the backlog is extraordinary, but light Q2 guidance, compressing margins (56% vs. 62% adj. EBITDA), ~$31–35bn of 2026 capex against a stretched balance sheet, ~45% single-customer concentration, and a freshly cut sell-side target collectively argue for maintaining the pinned 1.52% Core position and monitoring backlog conversion rather than adding here.
    CoreWeave (CRWV) is a GPU-as-a-service "neocloud" that rents Nvidia-accelerated compute to AI labs and enterprises, and it sits in the Diamond AI Fund's Core sleeve as a Neoclouds constituent (target weight 1.52%, pinned) precisely because it offers concentrated, contractually visible read-through to the hyperscaler capex cycle.
- CRWD (researched 2026-06-01) — verdict: **Hold.** The inception thesis has been decisively achieved and fundamentals — 24% ARR growth, record USD 1.24B FY26 FCF, an inflecting margin profile, and a credible agentic-AI catalyst in Charlotte AI — remain best-in-class, but at ~32x EV/FY27E sales and ~150x forward earnings, with net retention moderating and a headline revenue miss on the latest print, the risk-reward no longer supports adding; we maintain the Core base weight and monitor net-new ARR and competitive dynamics for either a de-rating entry point or a thesis crack.
    CrowdStrike Holdings (CRWD) is the leading pure-play, cloud-native cybersecurity platform, selling its Falcon suite of endpoint, identity, cloud, SIEM, and increasingly agentic-AI security modules on a subscription basis. It sits in the fund's Core sleeve under AI software & cyber as the consolidation winner in security software and the primary vehicle for "securing AI" — protecting GPUs, models…
- CRDO (researched 2026-06-01) — verdict: **Hold.** Credo is executing exceptionally — triple-digit revenue growth, ~69% gross margins, and a credible optical/vertical-integration roadmap firmly validate the Diamond thesis — but with the stock at 52-week highs above consensus targets, a forward P/E of ~45x, extreme two-customer concentration (39%/32%), and Q4 margin guidance signaling compression, the risk-reward is balanced rather than asymmetric; we maintain the 1.65% target weight and would revisit adding only on a pullback or fresh evidence of customer diversification, while noting the name sits comfortably inside the 4% single-name cap.
    Credo Technology (CRDO) is a pure-play, high-speed connectivity company whose flagship product — HiWire active electrical cables (AECs) — has become the standard copper interconnect for GPU-to-GPU and rack-scale links inside AI data centers, supplemented by optical PAM4 DSPs, SerDes IP/chiplets, and PCIe retimers.