YB new stock pitches (Wed, Jun 17)

Elastic N.V. ($ESTC, $6.3B market cap), creator of the widely used open-source Elasticsearch engine (est. 2010), is mispriced as legacy software at 18.6x forward P/E versus a 47.4x median across 17 AI-exposed software peers, despite sitting at the intersection of enterprise AI infrastructure. The thesis rests on three 'axioms' that won't change in ten years: AI always needs the right context (a retrieval/search problem that bigger context windows don't solve, since context is costly and 'lost in the middle' degrades performance), AI can never be fully trusted (requiring an external monitoring layer of logs, traces, permissions and anomaly detection, served by Elastic's Observability and Security products), and AI prefers what it already knows (agents default to the most-documented tools, like Postgres in databases—80%+ of Neon databases are agent-created—giving Elasticsearch, with 5.5B downloads, 120K+ GitHub stars, and use by 17% of professional/19% of AI developers, an incumbency moat). Elastic runs one engine across three products (Search, Observability, Security) over one copy of data, runs self-managed for data-sovereignty-sensitive customers, offers hybrid (keyword + vector) and multimodal search (via its 2025 Jina AI acquisition of open-weight, non-commercially-licensed embedding models), and is founder-led by CTO Shay Banon, a major shareholder. Lock-in is proven by land-and-expand cohorts: ARR grows from ~$32K (one solution) to ~$103K (two) to ~$390K (three); pre-FY21 customers drove ~61% of growth since FY20; and the AI tailwind is visible as GenAI penetration among >$100K customers rose from 4% (FY23) to 11% (FY24) to 21% (FY25), with the FY24 cohort expanding ARR 42% year-1-to-year-2 (double the ~20% five-year average). Elastic has 21,550+ customers (1,550+ over $100K). Management targets 20% growth (15% base + 5% GenAI) and Rule of 40 (~20% FCF margin), plus net dilution below 2.5% and a first $500M buyback (2025). Using FY26's $1.7B revenue baseline, 20% growth for five years yields ~$4.2B revenue and, at a haircut 15% FCF margin, ~$635M adjusted FCF by FY31—doubling cash earnings in four years even with no multiple re-rating, with re-rating as free upside. Key risks: OpenSearch (the free AWS fork) becoming the agent default and flipping incumbency from moat to trap (the most-watched risk), search collapsing into 'good enough' capabilities inside Postgres or hyperscalers, and Datadog continuing to take observability share. The author frames it as a rare asymmetric AI bet: heads you win a lot, tails you own a sticky, profitable, founder-led search business bought at a reasonable price.

Read the full article here. Read time: 13 min

7FIT S.A. (7FT.WA, added to position) is a Polish sports nutrition micro-cap (~42 million PLN market cap at ~23 PLN/share) trading on NewConnect with zero analyst coverage, founder-led by Michał Wasilewski (51% ownership) with CEO Arkadiusz Bała (19.5%), and notably a new ~7.75% holder, Michał Kopiczyński, who quietly accumulated a stake. The company sells over 170 products across 13 categories under the 7Nutrition brand, primarily wholesale B2B (95%) across Europe, the Middle East, and now Asia, with international sales at 38.5% of 2025 revenue. Since Bała took over in 2016, revenue grew from 9M to 35.3M PLN (FY2025, 8.1% net margin, 2.85M net income), with Q1 2026 revenue up 60.7% YoY to 13.4M PLN, net income up 137% to 1.57M PLN, and a 20-year-high 11.7% net margin. Gross margins sit stable at ~32% (whey-price dependent), with strong operating leverage (only ~25 employees, ~2.1M PLN revenue/employee, 7.5M PLN opex). At a 6.7x run-rate PE—down from 3.9x at the report—it is deeply undervalued versus peers SFD, Applied Nutrition, and Glanbia (median ~14.7-16.8x). Catalysts include the 1.6-hectare Stanowice land bought cheaply (490K PLN) for a future facility, Asia expansion (Singapore, Osaka, Hong Kong, Shanghai) into the broader health/wellness segment, new product lines, aggressive hiring, and compounding earnings forcing a rerating; the author conservatively models FY2026 revenue of ~48M PLN (35.8% growth), and a DCF yields ~40.5 PLN/share (+76%). Bear case (raw material spikes, plant delay, slowing growth) sees 18-20 PLN (-15-20%); base case targets 60-70 PLN (a triple) on 55-65M PLN revenue by 2028; bull case (Asia success, new facility, 80M PLN revenue at 15% EBIT margin) sees 80-100+ PLN (4-5x). Risks include extreme illiquidity (11.9% free float, ~213,210 shares, ~1.2M euros tradeable), whey-price cyclicality (200-300bps margin compression in bad years like 2024), single-plant concentration in Stanowice, governance concerns (family-dominated supervisory board, unexplained growing 609K PLN loan), undisclosed customer concentration, competition (Myprotein, Scitec, Optimum Nutrition), uncertain/indefinite China regulatory timelines, and key-man risk in Bała. The author began buying at ~17 PLN (~8M euro market cap) and has since added, viewing the base case as most likely.

Read the full article here. Read time: 15 min

VersaBank (VBNK), a founder-led, branchless B2B lending machine trading on both the TSX and NASDAQ at ~US$650M market cap (~US$20/share, roughly 1.5x book value and ~10x 2027 EPS), is a 'software company hiding inside a bank charter' that just grew revenue 27% and earnings 45% year-over-year. The core business buys bundled loan stacks via its Structured Receivable Program—where each lending partner leaves a cash cushion that absorbs losses first—producing a 0.04% loss rate (among Canada's lowest) and a fat ~2.7% net interest margin; deposits flow in cheaply via brokers and bankruptcy/insolvency cash (a hedge that strengthens funding in downturns), and its dual federal bank charters (the first new US charter in 18 years) plus a 50-year loss record form the moat. The thesis is two ideas stacked: (1) a proven Canadian model (C$5.7B assets, 30%+ annual growth) being copy-pasted into the US market—roughly 10x larger—where year-one fundings hit US$310M (past the US$290M target), the US book already crossed US$600M, management targets +US$1B US funding this year and C$10B+ total assets, with the US already 20%+ of revenue on a fifth of assets and heading toward 50/50; three forces (operating leverage from a fixed tiny cost base, a widening margin, and an AI-powered Real-Time SRP accelerant that buys loans instantly and could allow fee-generating syndication of overflow) supercharge growth, with US efficiency already improving from 41% to 37% in one quarter, supporting estimated 2027 core EPS of ~$2 (stock could double in a year) and 2030 core EPS of $5-6; and (2) a free, zero-valued option in tokenized deposits—VersaBank became the first bank to complete a tokenized-deposit pilot (live on Algorand, Ethereum, Stellar), interest-bearing and expected to be federally insured, sitting on its balance sheet as cheap funding, where just US$2B lent at the spread throws off ~US$40M after tax (more than the entire bank's current ~US$36M), US$10B would be ~US$200M (~5x the company), against US$1.3 trillion in annual US-Canada cross-border trade. The key risk is capital: CET1 fell to 12.3% from 14.3% and the leverage ratio dropped to 7.9% from 9.6%, so rapid US growth may require an equity raise (a planned cybersecurity arm sale could free capital), though raising above 1.5x book and deploying at the fat margin is accretive, not dilutive; smaller risks include the unproven US loss record and a stock that has run hard. The author owns VBNK.

Read the full article here. Read time: 9 min