YB new stock pitches (Thu, Jul 23)

Long Galaxy Digital ($GLXY) with a fair value of $45-75 per share (50-150% upside), with the thesis being that pervasive 'crypto hate' obscures the value of the Helios Data Center project hidden in plain sight; the stock has ~407MM shares outstanding (reflecting conversion of the in-the-money 2029 convertible bond) and trades ~$166MM/day. Galaxy has two segments: Digital Assets, a one-stop-shop for the crypto ecosystem with $8B AUM/AUS, recurring fee revenues (trading, lending, staking, brokerage, advisory), $700MM balance sheet crypto, $650MM investments, and $629MM in 3Q'25 EBITDA (viewed as optionality); and Data Centers, anchored by Helios in West Texas—a 1-of-1 asset with 1,500 contiguous acres (with expansion options), underground aquifer water, grid interconnect, built-out infrastructure (owned substation, transformers, generators, chillers), 1,300 workers on site, 1.63GW of ERCOT-approved front-of-meter power, an 830MW recently approved (awaiting customer announcement), 1.8GW of additional plans (3.4GW total planned), and an 800MW 15-year contract with CRWV (end customer a 'multi-trillion' hyperscaler, suspected Microsoft or Meta) worth ~$1B/yr with 90%+ flow-through, ramping this quarter. Helios's four differentiators are land/water, labor/execution (first data hall delivered on time and on budget while competitors face delays), power (ERCOT approvals are scarce given a lengthy queue and new 'batch' system), and location in Load Zone West—an area with 70+ GW of renewables where operators are sometimes paid to take excess power. Valuation approach: spot the balance sheet, value Digital Assets conservatively, and DCF Helios in phases—the contracted 800MW CRWV deal covers today's price, the 830MW approved power gets ~50% upside, and additional sites/approvals drive blue-sky cases; on TEV/approved MW, GLXY is far too cheap versus peers (CORZ, APLD, CIFR, HUT, WULF) with 75-100% upside, a gap that opened after the late-2025 BTC roll-over. A modeling note highlights that once projects stabilize, refinancing on loan-to-appraised-value (e.g., $1B initial to $1.5B refi) frees capital for recycling, avoiding dilutive equity. Catalysts include the Russell 2K index add announcement around May 22 (~7-8MM shares of buying demand), an 830MW customer announcement by year-end (providing customer diversification since it won't be CRWV), Phase 1 delivery to CRWV by June 30, additional data center sites, more ERCOT approvals, and potential crypto upside from Clarity Act progress. Founded in 2018, listed in Canada, moved to Nasdaq in May 2025 post-Genius Act, led by Mike Novogratz (viewed as a net positive, excellent team). Risks: execution (delays, costs, safety), CRWV counterparty risk (mitigated by its large market cap, a motivated hyperscaler end customer, and the diversifying second customer), and crypto downside (though data center reality should eventually outweigh it).

Read the full article here. Read time: 8 min

Convequity is initiating a modest paired position of approximately 1% across Enphase (ENPH) and SolarEdge (SEDG) as complementary exposure within the power-electronics segment of the AI data center value chain, treating this as a capability bet rather than a near-term revenue story. The thesis centers on the structural shift toward 800V DC architectures and megawatt-scale racks, which makes traditional copper-heavy power distribution impractical and pushes the industry toward higher voltages and semiconductor-based solid-state transformers (SST); Enphase's established expertise in bidirectional power conversion—originating in its storage systems—positions it well, and in April 2026 it expanded explicitly into AI data centers with the launch of its IQ solid-state transformer, following SolarEdge into the space, though this is an earlier commercial-scale product at lower voltages rather than a full utility medium-voltage (13 kV) SST. A true next-generation SST would convert utility power (~13 kV) directly down to the 800V DC favored by high-density AI racks in a single compact bidirectional unit, supporting reverse flow for grid export, storage arbitrage, peak shaving, and ancillary services, while handling the extreme 'rocket shifts' of GPU clusters cycling on and off and maintaining fail-safe circuit protection where a single breaker failure could destroy a rack worth millions or billions. Enphase does not currently offer a commercial medium-voltage SST, and the meaningful technical and commercial inflection for direct 13 kV-to-800V DC capability is viewed as roughly two years away, creating a window for firms with proven bidirectional, grid-tied power electronics to build positioning ahead of broader recognition; a supportive secondary dynamic is the parallel evolution of smart grids and residential microgrids that could free grid capacity and provide peak-shaving back to AI loads, an ecosystem where both companies already sit. Following the recent correction, Enphase trades at attractive multiples of approximately 14x gross profit and 38x free cash flow, with SolarEdge similarly attractively valued and viewed as marginally ahead in timing while Enphase catches up; the team believes the AI data center power angle is only partly appreciated by the market, with chart behavior reminiscent of prior holdings 'just about getting noticed.' Conviction is constructive but tempered by the risk that the 800V DC and SST narratives carry reversal risk similar to other high-profile AI infrastructure themes, with mainstream adoption likely arriving later.

Read the full article here. Read time: 5 min

IB Kimyoung (KOSDAQ: 339950) is a Korean cram-school leader with roughly 80% share, preparing students for the university-transfer exam, that trades under 2.5x EV/EBIT ex-cash and screens as a dull, cheap education stock but is materially undervalued once read closely. The core business earns about ₩20bn operating profit on the ₩16bn of capital it actually uses (~152% ROIC), sustains 70%+ gross margins, runs on prepaid student float (₩27bn by Q1) rather than its own capital, and generates cash exceeding profit (₩32.8bn operating cash flow vs ₩20.1bn net income last year, with FCF ~120% of earnings). Of ₩130bn in assets, roughly a third of the ₩115bn market cap sits in cash (₩44bn cash/deposits, plus ₩10bn office, ₩9.4bn private real-estate funds, ₩5bn loans, ₩3.5bn listed shares); the July-initiated ₩100 quarterly dividend annualizes to ₩400 (~15% yield at ₩2,650), costs ~₩17bn against ₩20bn FCF, and is currently a tax-free return of capital lasting about a year before becoming taxable. The apparent shrinkage and lumpy history are largely accounting artifacts (2020 was a ₩3.97bn non-cash shell-listing charge plus COVID; 2022's ₩3.08bn trough came from consolidating loss-making acquired academies; the 3% 2025 dip reflected a small division sold to the parent), while the core grew revenue from ₩56bn (2021) to ₩84bn (2024) and hit its first-ever Q1 profit in early 2026 on 6.3% revenue growth, supported by a transfer quota up 27% since 2020, a 2027 medical-school expansion, and 2028 admissions reform. The bear case is real: Megastudy owns 55% and allocates capital for itself, buying an office from the parent, parking cash in private funds, lending ₩0.5bn to its own managers, injecting ₩14.5bn into a chronically loss-making IT subsidiary (a rescue matching its ₩14bn equity hole), leaving 1.58 million treasury shares uncancelled while issuing 1.23 million dilutive options below market; it is also a one-shot business dependent on a government-controlled exam gate. SOTP values the core at ₩130-160bn (8-10x after-tax), plus ₩42bn net cash and a haircut ₩25bn for non-core, exceeding ₩200bn versus the ₩115bn cap—so investors are paid three ways (cheap dominant core under 2.5x EV/EBIT, ~15% covered yield, and free optionality on cash surfacing), with only the third requiring the controller to reform. Key monitoring triggers: whether the quarterly dividend holds, whether non-core buying stops, and whether treasury shares are cancelled.

Read the full article here. Read time: 5 min