YB new stock pitches (Mon, Aug 3)

Galaxy Gaming ($GLXZ) trades at $1.63/share, a 5-year low, after its acquisition by Evolution ($EVO.ST) fell through, leaving the stock deeply beaten down despite high-quality financials that don't justify such a low multiple. With a fully diluted share count of 26.19M, the FD market cap is $42.69M on $11.93M TTM EBITDA (sub-4x, though this is somewhat deceptive given sizeable net debt and a working capital swing that boosted LTM FCF), with TTM EV/EBITDA sub-7x. Unlike in 2022, the company can now service its debt—incurred from a levered share repurchase to retire the founders' shares and carried since—via operational cash flows, and the business is growing. Post-breakup, Galaxy announced a $4M buyback funded entirely by the $5.23M Evolution termination fee, and the author believes it can afford a >15% buyback yield (~$6M at current prices) while paying down net debt in the next year. Former CEO Todd was fired in late 2023 (the year perpetual license sales, deemed low-quality non-recurring revenue and described by some as borderline fraudulent, peaked); those sales are now down 80% YoY, a positive. The author expects $GLXZ to deliver an average return above 30% annually over the next 3 years—much of it within 12 months—driven by sentiment recovery and a likely rerate to a higher multiple. The author is long shares.

Read the full article here. Read time: 3 min

Celestica (CLS), an AI-infrastructure design and manufacturing partner, delivered a Q2 2026 that beat on both lines above the high end of guidance—revenue of $4.70B (+62% y/y, vs. $4.3B est.), adjusted EPS of $2.54 (+83% y/y), and a record 8.2% adjusted operating margin—marking its fourth consecutive quarter clearing its own EPS guide. Management raised full-year 2026 guidance for the third time in six months to $20.5B revenue (from $19.0B), $11.30 adjusted EPS (from $10.15), 8.4% operating margin, and $600M FCF (up from a long-stuck $500M), while stating 2027 revenue growth will accelerate past the 65% guided for 2026—implying a 2027 revenue base above $33.8B and adjusted EPS above $18.65. Key catalysts include the reveal that the mystery digital-native customer is OpenAI (custom accelerator racks built with Broadcom, mass production 2027) and the AMD Helios scale-up switch program (samples end-2026, ramp 1H27), both described as multi-billion-dollar 2027 opportunities that didn't exist in the original thesis; the CCS segment grew 84% (81% of revenue), Communications +62% with 1.6T mass production beginning (10 active programs), and Enterprise +167%. The bull case rests on structural hyperscaler AI capex (visibility into 2028/2029), a 400G/800G/1.6T networking ramp, a design-plus-manufacturing moat (2,000 engineers by year-end, programs being consolidated back from second sources), 55.5% adjusted ROIC, 0.5x leverage, $204M net debt, and $2.3B liquidity, with customer concentration improving sequentially (top three 32%/17%/14% vs. 35%/15%/15%). The stock fell 28% from a June high of $474 on sentiment-driven AI-sector fears (Chinese domestic DUV lithography, CXMT's Shanghai listing and DRAM oversupply worries, AI financing/circular-deal risks, and Chinese hardware competition)—all of which the author argues don't materially touch Celestica (cheaper memory helps its bill of materials, no vendor financing, non-cancellable orders, export-controlled silicon protecting it from Chinese competitors) except AI capex financing, now the top watch-list risk as a potential 2028 problem via order-book length. Bear cases and risks include: 30x guided 2026 EPS (RankedStocks valuation score of 20, RS 72 overall), thin 2.9% FCF conversion with no buybacks as capex ($1B in 2026, ~$1.5B placeholder 2027) absorbs cash, the 32% top-customer concentration (top 10 = 83% of revenue, up from 78%), tight component supply, high volatility (12-month range $169–$474), and dependence on trade credit (payables days extended 54 to 73). Q3 2026 guidance is $5.25B–$5.55B revenue (+69%) and $2.88–$3.08 adjusted EPS (+89%). The author, who has held CLS since 2023 at 8.3% portfolio weight (up 530%), maintains a BUY with a $575 target (68% upside from $342), deliberately holding it flat despite the raise as a conservative anchor since his DCF's 2027 revenue forecast ($21.9B) sits 35% below management's implied guidance.

Read the full article here. Read time: 21 min

Dong-A Eltek (088130.KQ), a Korean holding company on the KOSDAQ, trades at ₩4,325 versus an estimated NAV of ₩17,780 per share, a 76% holdco discount. Its primary asset is a 47% stake in listed Sunic System (171090.KQ), worth ₩12,810 per Dong-A share; Sunic broke Canon Tokki of Japan's twenty-year monopoly on OLED thin-film vapor deposition, won the deposition order for BOE's 8.6-generation IT OLED line in China (supplying iPads and MacBooks), and now forms a duopoly with Canon in a market where global OLED IT shipments are forecast to grow from 24 million units this year to 53 million by 2029. Sunic generated ₩515 billion revenue and ₩96.5 billion net income in 2025 with a ₩582 billion backlog and trades under 5x P/E. Dong-A also owns DA Value Investment, a licensed VC firm with ₩23.7 billion net assets (₩1,360/share) holding stakes in 60+ funds, and the parent's optical inspection business (customers Samsung Display, LG Display, BOE) generating ~₩6 billion annual profit, with parent equity of ₩63.2 billion (₩3,610/share) including undervalued real estate. The group has net cash of ~₩97 billion (₩180 billion cash plus ₩5 billion investments vs. ₩88 billion borrowings, no bonds) against a ₩75 billion market cap. Tailwinds include Korea's value-up program, the Commercial Act amendment on directors' duties, and activist pressure; the company also cancelled 18% of shares (₩15 billion) in September 2024. Chairman Park Jae-kyu holds 14.83% personally (37.74% with related parties) and is co-CEO of both listed companies. Risks are that the discount could persist for a long time or that China and Korea enter a trade war. The author views this as a timely, cheap entry given the recent indiscriminate sell-off of Korean memory names.

Read the full article here. Read time: 3 min