New ticker reveal: The stock I have been looking into is $SILC. I'm not the first to write about it @PepInvestStocks and @WealthyReadings have written about it before, but I believe we're still early. Let's get into it. To start, they have a big catalyst coming: H2 2026 proof of concept at a Tier 1 hyperscaler. PoC conversion brings tens of thousands of FPGA inference cards at multi-thousand-dollar prices per unit, mix-shifting consolidated revenue toward higher-margin product and triggering gross margin recovery from depressed 30% toward 2022 peak of 34%. It hit both revenue ramp + margin inflection, and forward EV/EBITDA prints 6.6x, below every direct networking peer. Cash + working capital covers ~50% of market cap, so the bear case is bounded while you wait. Now that I have your attention, let me give you a brief rundown using my framework: AI enabler or beneficiary? Enabler. Three AI product lines built on different chips. First, FPGA inference cards using chips from Intel and AMD (FPGAs are programmable chips, meaning they can be reconfigured as AI models change every few months) targeting cloud datacenter inference workloads. Second, NVIDIA GPU edge appliances (Marbella line, using NVIDIA's L4 chip designed specifically for video analytics and AI at the edge of the network rather than the cloud) for telecom and enterprise customers. Third, Hailo-8 AI appliances (Madrid line, using Intel's low-power CPU plus Hailo's specialty AI chip, 26 trillion AI operations per second using only 2.5 watts of power) for factory robots, security cameras, smart retail systems, and automated sorting. Three customer engagements stacked: a May 5 proof of concept targeting hyperscalers at multi-thousand-dollar prices per unit, two AI compute customers already shipping product, plus a third inference product co-developed. Mgmt has explicitly guided "significant" AI inference revenue in 2027. ✅ Leadership. CEO Liron Eizenman (since July 2022) built the Edge Networking division that became Silicom's primary growth driver pre-2022. The business is back to executing: Q4 25 +17% YoY, Q1 26 +33% YoY (materially ahead of the 18% guide). Two consecutive beat-and-raise quarters. 4 design wins booked in 4 months versus 7 to 9 annual target, running well ahead of pace. ✅ Revenue Quality. $61.9M FY25 real revenue. 200+ global customers, 400+ active design wins, 75% North America. Design-win driven business, i.e. each customer takes 6 to 18 months to onboard and re-qualify, meaning once a customer chooses SILC it's expensive and slow for them to switch away. Edge AI line opens diversification away from traditional telco/cybersecurity/cloud customer base into manufacturing, retail, security, smart cities, and robotics end markets. ✅ Revenue Growth. FY26 guide $82 to 83M (+33%). Layered growth engines: design-win compounding, AI inference across all three product lines, post-quantum cryptography (3 wins to date, $3 to 4B market by 2030, new encryption standards needed to defend against quantum computers cracking today's encryption), white-label switching ($6 to 7B market by 2030), edge AI appliances for robotics/security/retail. CAGR framework points FY28 to $130 to 140M baseline; mgmt's long-term target is $150 to 160M + $3+ EPS. Margin leg: current gross margin 30% vs 2022 peak of 34.5%. Mgmt's $3 EPS target implicitly requires recovery to ~34% -> each 1% of gross margin recovery on a $135M base = +$1.35M of operating income.✅ Moat. 20+ years of proprietary hardware and software engineering. Proprietary cybersecurity packet-capture technology running at 400 Gigabits per second on FPGA chips. Nanosecond timing technology used in 5G networks and high-frequency trading (where microseconds equal money). Physical fail-safe circuits that keep networks running when security appliances crash: cybersecurity vendors specifically design products around these. Plus multi-silicon platform engineering: integrating chips from Intel + NVIDIA + Hailo + AMD into single appliances with industrial-grade certifications. The moat lives in niches that silicon vendors don't find economic to integrate at SILC's scale.✅ Validator stack: NVIDIA (L4 GPU partnership for the Marbella line), Hailo (AI partnership since July 2023 across multiple product lines), Tier 1 cyber security ($5M switch family + $2M Edge system), European secure communications ($3M post-quantum cryptography deal), major streaming provider ($25 to 30M five-year contract), global networking + SaaS leader (raised expected spend from $3 to 4M to $8 to 10M annual), two AI compute customers, hyperscaler-focused inference proof of concept. Asymmetry. EV ≈ $154M against $109M working capital + securities. Comp ladder: Bear $32 (Adtran/Ceragon at 1.5x EV/Sales), Base $71 (Radware at 2.5x — sister RAD Group), Bull $120 (RDWR premium at 3.5x), Moonshot $245 (specialty + AI capped at 5x — NOT silicon-designer multiples). Probability-weighted EV $85 vs spot $38 = 2.23x.✅✅ Forward EV/EBITDA: > Base ($135M FY28, 32% gross margin): 10.4x > Aspirational ($155M FY28, 34% gross margin = mgmt target): 6.6x > Moonshot ($250M, 35% gross margin, PoC converts): 2.9x Conviction Gap. Two discrete near-term catalysts: H2 2026 AI inference proof of concept outcome AND gross margin recovery trajectory. Q2 26 print (late July / early Aug) is the first checkpoint on both. Watch for the first quarter printing ≥31% gross margin imo, that's the structural inflection signal, more important than headline revenue beat. Single active analyst with stale $28 PT (April 2025) vs Morningstar quant fair value $96.38 = 3.4x dispersion. Thin coverage.✅ TL;DR. Forgotten Israeli RAD Group networking name printed a clean 33% Q1 inflection. Cash + working capital covers ~50% of market cap, the bear case is bounded. Three AI product lines across different chip vendors, FPGA cards for cloud datacenters, NVIDIA GPUs for telecom edge, Hailo specialty AI chips for industrial robotics/security/retail, which is genuine diversification away from telco-heavy historical mix. Real moat in 20+ years of proprietary networking and edge appliance engineering. Bull case has two legs: revenue ramp to mgmt's $155M target AND gross margin recovery to 2022 peak of 34%. Hit both and the multiple rerates from 10x to 6.6x EV/EBITDA, below every direct peer. Margin recovery is the leg that doesn't require a binary catalyst, mgmt's own EPS target is the receipt. Technicals: Chart looks like it is cooling down after a run from $18 to $50 recently. Coming down to test the 10EMA and 0.382 Fib level around $37 on lower volume. 50/200EMA golden cross occurring soon on weekly timeframe, indicating a bullish trend for the mid-long term. Still overall very bullish. ✅ I truly believe that despite the recent run up, we are still early. I won't be surprised if we are at $100 by EOY. Just my 2c, as always. - Leki the investing monkey 🐒