| Ticker | Name | Current thesis | Research | Seen in | Total $k |
|---|---|---|---|---|---|
| CEPL | Capstone Energy+ (fmr. Capstone Green Energy) | The flagship bring-your-own-power holding: microturbines of 50 kW–3 MW delivered in 1–3 months against 18–24 months for utility-scale turbines, running at ~10% of a ~1 GW/yr capacity, at ~3× revenue versus Bloom's 15–20×. Three ways to win — hyperscaler data halls, thousands of 1–4 MW edge sites, and grid-constrained manufacturers. | QT · SA · STK · FA | 2026-AUG-07 | — |
| PSFE | Paysafe | Positive with explicit hair: a processor specialising in gaming and prediction markets that Stripe and Adyen avoid, where a portable digital-wallet subsidiary worth 60–70% of enterprise value at 9–10× earnings means "not a lot has to go right for there to be a rerating." Against that, ~$2.5bn of debt on ~$600m book value and high-single-digit organic growth under a new-CEO turnaround. | QT · SA · STK · FA | 2026-AUG-07 | — |
| RELY | Remitly | The digital Western Union killer — migrant remittances where the moat is mindshare, not price: migrants buy on trust and brand, so the take rate erodes only a basis point at a time while send volume grows 30–40% and revenue 25%. Long runway as cash remittances electronify. Rides Visa/Mastercard rails on 90%+ of volume rather than owning the pipes. | QT · SA · STK · FA | 2026-AUG-07 | — |
| SLNG | Stabilis Solutions | Speculative-sleeve (~$75m cap) play on the "virtual pipeline" — liquefying, trucking, storing and vaporising natural gas for data centres with no pipeline access. A consolidated niche with few providers, a new data-centre contract Dean sizes at a couple hundred million over two years, and SpaceX as a large LNG customer scaling launches ~20× over five years. | QT · SA · STK · FA | 2026-AUG-07 | — |
| SMOP.OL | Smartoptics Group ASA (Oslo Børs) | The scale-across bet: as 1–3 GW monolithic data centres top out on power and local moratoriums, hyperscalers must bridge geographically separate sites and bandwidth needs rise ~15×. Smartoptics makes the hardware that turns data into light for inter-site fibre plus best-in-class software, interoperable with Broadcom/Cisco/Arista gear. Oslo Børs listing. | QT · SA · STK | 2026-AUG-07 | — |
| SPT | Sprout Social | A holding and the worked example of the SBC screen: decelerating growth plus growth-era stock comp makes it screen terribly on GAAP, which is why it was priced "for complete decapitation." SBC should be read as a variable expense that normalises, inflecting GAAP earnings with no change in the business. Up 50–60% since he flagged it in April 2026. | QT · SA · STK · FA | 2026-AUG-07 | — |
| TENB | Tenable Holdings | The cyber pick: frontier models collapsed the cost of finding code vulnerabilities, shifting attacks back from identity to exploitation, and Tenable leads the resulting category — exposure management (asset discovery, scanning, ranking what actually matters, remediation) with an Anthropic partnership. 4× EV/sales versus CrowdStrike and Palo Alto north of 20×, with a large scanner base to upsell and growth expected to accelerate. | QT · SA · STK · FA | 2026-AUG-07 | — |
| UPWK | Upwork | Their most contrarian call — a marketplace with network effects that is "impossible to replicate," where AI damage is confined to the sub-$500 job band Upwork itself discloses (Fiverr's core, not Upwork's). Higher-value consultative work holds because clients don't know what to ask AI. A little over 1× EV/sales and ~6× EV/FCF after SBC. | QT · SA · STK · FA | 2026-AUG-07 | — |
| VPG | Vishay Precision Group | The robotics leg: humanoid hardware exists but the brain doesn't, and high-fidelity physics simulation is how AI supplies it. VPG makes the strain gauges and sensors that go into robot hands, with three large customers already in prototype (Tesla, Figure named). Explicitly NOT Vishay Intertechnology (VSH). | QT · SA · STK · FA | 2026-AUG-07 | — |
| WSE | Wise Group plc (Nasdaq primary; LSE: WISE) | Owns the plumbing: direct connections into national payment systems nobody has replicated, moving money cross-border near 2% against correspondent banking's ~5%. The distinctive feature is offence — it cuts its own take rate aggressively as a loss leader rather than defending it ("they're the one causing the storm") — and it targets businesses, banks and ultimately the card networks. Primary listing moved to Nasdaq (WSE) in May 2026; LSE secondary. | QT · SA · STK · FA | 2026-AUG-07 | — |
| Ticker | Name | Current thesis | Research | Seen in | Total $k |
|---|---|---|---|---|---|
| 000660.KS | SK Hynix (KRX) | Named alongside SanDisk as the memory complex that fell 40–50% in a month on ROI doubts — the setup for the conversation. Pernas owns no semiconductors at all, and separately flags Chinese DRAM capacity (CXMT) as the structural reason. | QT · SA · STK | 2026-AUG-07 | — |
| AMZN | Amazon / AWS | Cited on both sides of the open-weights argument: evidence that frontier-model pricing had become unaffordable (Amazon and Uber pulling back on AI spend, budgets blown "in a span of a month"), and, as AWS, one of the hyperscalers that hosts the cheaper replacement and captures the GPU rental economics instead. | QT · SA · STK · FA | 2026-AUG-07 | — |
| — | Anthropic (private) | Appears in three roles: a frontier lab whose ~$10-per-million-input-token pricing was blowing up enterprise budgets; the accuser that says open-weight models were built by distilling frontier models (~4 million exchanged messages); and Tenable's partner, supplying the model that reads a customer's code and assets for exposure management. The frontier layer is the one layer open weights are not bullish for. | — | 2026-AUG-07 | — |
| ASML | ASML Holding | The risk framing behind owning no semis: China is "no longer just a producer of $1 widgets," already dominates EVs and drones, and is "going after ASML" as well as the memory makers. The ability to produce DRAM, CPUs and lithography machines far cheaper "gives us pause regardless of how bullish the overall industry looks." Not a call on ASML's stock. | QT · SA · STK · FA | 2026-AUG-07 | 4.9 |
| BE | Bloom Energy | The valuation foil rather than a short: fuel cells at roughly Capstone's scale and "a bit more efficient," selling into the same on-site power shortage — but at 15–20× revenue against Capstone's ~3×. Named as the mainstream way to own the theme they deliberately did not buy. | QT · SA · STK · FA | 2026-AUG-07 | — |
| CEG | Constellation Energy | Named by the host as one of the behind-the-meter names retail hears constantly — the setup for why Pernas owns micro-caps nobody else has instead. No view expressed on Constellation itself. | QT · SA · STK · FA | 2026-AUG-07 | — |
| — | CXMT / ChangXin Memory (Shanghai STAR — 2026 IPO) | CXMT (auto-caption "CXML"), China's largest DRAM maker, IPO'd on Shanghai's STAR Market in July 2026 at roughly a half-trillion-dollar market cap — the concrete evidence for the China risk that keeps Pernas out of semiconductors entirely. No US line; deliberately not tickered. | — | 2026-AUG-07 | — |
| CRCL | Circle Internet Group | Cited in passing as the market's own verdict on the stablecoin story — "Circle's come down significantly as a result of this narrative fading." His view: stablecoins are overwhelmingly a crypto-trading instrument, with "absolutely zero cases" of everyday fiat-like use outside a few broken banking systems, so they are not yet a threat to Wise or Remitly. | QT · SA · STK · FA | 2026-AUG-07 | — |
| CRWD | CrowdStrike | The incumbent benchmark, used twice: the reflex answer to identity-era threats ("let's hire Crowd Strike, let's get their Falcon agent"), and the valuation comparison at north of 20× EV/sales against Tenable's 4× on a platform shift Tenable may be better positioned for. It does earn far more GAAP profit today. | QT · SA · STK · FA | 2026-AUG-07 | — |
| GEV | GE Vernova | A reference point twice over: one of the mainstream bring-your-own-power names, and the maker of the 200–300 MW turbines a microturbine is "a large turbine shrunken down" from. Its 18–24 month lead times are the gap Capstone's 1–3 month delivery fills. No view on the stock. | QT · SA · STK · FA | 2026-AUG-07 | — |
| GOOGL | Alphabet / Google | Grouped with Microsoft as a rational allocator that monetises AI spend through the cloud, and one of "the usual suspects" that will host enterprises' open-weight inference. Same ~10× EV/sales cohort caveat; no separate view on the stock. | QT · SA · STK · FA | 2026-AUG-07 | — |
| MA | Mastercard | Paired with Visa throughout as "the king and queen of the space" — the only payments names spared huge multiple compression, and the rails Remitly settles over. Same qualification: Wise is the competitor genuinely building around them. | QT · SA · STK · FA | 2026-AUG-07 | 7.0 |
| — | Moonshot AI — Kimi (private, China) | The catalyst for the whole conversation: Moonshot's open-weight Kimi model, described as now number three in the world, triggered the fear that frontier-lab pricing power collapses. Dean thinks the gains are "not purely due to innovation, architectural improvements" but substantially distillation — while still being bullish for the chip, cloud and application layers. | — | 2026-AUG-07 | — |
| MSFT | Microsoft | A beneficiary of open weights: enterprises running cheap open models still rent someone's cloud, and "the economics translates to whoever controls the GPU." Capital allocation is "a lot more measured and a lot more rational" than Meta's because Microsoft actually monetises through the cloud — but it sits in the ~10× EV/sales hyperscaler cohort where a lot has to go right. | QT · SA · STK · FA | 2026-AUG-07 | — |
| — | OpenAI / ChatGPT (private) | The other frontier lab in the four-layer framework — the layer whose unit economics compress when open weights arrive. Enterprises no longer pay ~$10 per million input tokens, creating "an air pocket in demand for the frontier models," with that demand re-hosted on the hyperscalers instead. | — | 2026-AUG-07 | — |
| ORCL | Oracle | Named at the head of the hyperscaler list; the only view attaching to it is the cohort view — roughly 10× EV/sales on average while spending all their free cash flow and more on a step-function capex, so "a lot is going to have to go right for that multiple to be roughly correct." | QT · SA · STK · FA | 2026-AUG-07 | — |
| PANW | Palo Alto Networks | The other larger cyber rival north of 20× EV/sales — cited to size the discount on Tenable rather than as a view on Palo Alto itself. | QT · SA · STK · FA | 2026-AUG-07 | — |
| PCOR | Procore Technologies | The worked example of the post-AI software-defensibility test, not a position: asked where a moat comes from now that production-ready code costs 90–95% less to write, Deiya lists unique proprietary data, security/compliance, enterprise-grade features, integration depth and "a real world component" — helping with giant construction projects — versus a purely digital tool. | QT · SA · STK · FA | 2026-AUG-07 | — |
| SNDK | SanDisk | Market backdrop rather than a pick: SanDisk, SK Hynix and "that entire memory complex and AI complex" down 40–50% in a month with the market questioning AI ROI. Pernas' whole argument is that the drawdown misreads what open weights do to the stack. | QT · SA · STK · FA | 2026-AUG-07 | — |
| TOST | Toast | Positive-leaning but not a position: pure payment processing is commoditising and "the value seems to be accruing to the application layer, to the Toasts of the world" — verticalised software that runs inventory and the nervous system of a business with payments embedded. The rule drawn from it: a payments investment has to do more than process. | QT · SA · STK · FA | 2026-AUG-07 | 0.3 |
| V | Visa | The archetypal payments compounder and the rails almost every fintech ends up riding — 90%+ of Remitly's transactions settle over them, which is why the recurring fintech-disruption bear case keeps failing. The one genuine threat named is Wise, which built its own infrastructure: "Wise is bad for Visa and Mastercard." | QT · SA · STK · FA | 2026-AUG-07 | 0.8 |
| Ticker | Name | Current thesis | Research | Seen in | Total $k |
|---|---|---|---|---|---|
| FVRR | Fiverr | The short side of the Upwork trade: Fiverr "productized a lot of these lower value jobs" — the sub-$500 tasks Upwork's own data shows AI is competing away, which is most of Fiverr's business and a minority of Upwork's. "I think they're going to have a much tougher time." | QT · SA · STK · FA | 2026-AUG-07 | — |
| META | Meta Platforms | Once a large position — bought after the 2022 collapse near $90 from $300-something and sold above $700 last year — now explicitly negative. The open-source content strategy was right but Llama underdelivered (cultural mishaps, Yann LeCun's opposition to LLMs) and hundreds of millions spent poaching researchers "didn't really yield anything." Meta isn't in the cloud business, so its capital allocation is far less rational than Microsoft's or Google's. | QT · SA · STK · FA | 2026-AUG-07 | 3.0 |
| PYPL | PayPal | A published pass. They liked the Braintree processing segment — volumes "growing like crazy" with scale economics ahead — but not branded checkout, the high-margin engine, where "there's just a lot of vulnerability… it's hard to see the future growth." Liking the small unprofitable half and distrusting the large profitable half meant the whole didn't work. | QT · SA · STK · FA | 2026-AUG-07 | — |
| WU | Western Union | The disrupted incumbent in the Remitly thesis, named only as the loser: Remitly is "a Western Union killer… much more enhanced experience, cheaper fees, and has really eroded Western Union's business model." The cash still walked into branches is the volume being electronified away from it. | QT · SA · STK · FA | 2026-AUG-07 | — |
The offerings below are the brothers' own description of the service on 2026-AUG-07 (~41:31–48:50), plus the positions they name in that same conversation.
| Offering | What it is | How they run it | Seen in the index |
|---|---|---|---|
| The model portfolio (3 sleeves) | Access to every position with its rating and weight. Core 5–15% — high conviction, long horizon, ~70% historical batting average. Starter ~3% — taken before the full three months of work is possible, able to graduate. Speculative 1–3% — little downside protection, played for slugging percentage. | Sizing follows payoff shape, not enthusiasm. Cash weighting rises when origination isn't producing qualifying names. | CEPL, VPG, TENB, SPT, UPWK, WSE, RELY, PSFE — and SLNG, named explicitly as speculative at a ~$75m cap |
| Initiation reports | "Every single name that we put in the portfolio there's an initiation report on." | Deliberately short — "two to three paragraphs," "no fat at all," which is what makes covering a hundred names a year sustainable. | Every position above |
| Action updates | A written communication attached to every trim, add or exit — "there's always research communications associated with it." | Forces the fundamental case to be restated before capital moves. The add rule: fundamentals confirming while price falls. Averaging down that way is "about 20 to 30% of our alpha." | META (bought ~$90 in 2022, sold above $700 last year) |
| Monthly report | "Every month we come out with a report on something" — a longer piece on a theme or a name. | Regular cadence rather than event-driven. | The AI-stack and bring-your-own-power themes |
| Weekly "Stock Sonar" | Three names a week they looked at — some interesting, some needing more work, some passed on. | "We try to give our members an inside view into the research process and where we're looking." | — |
| Pass notes / research vault | Published write-ups on open positions, closed positions and names they are neutral on — including explicit passes with the reason. | Origination runs at "roughly one in a 100," so the rejections are the bulk of the work; publishing them "gives people a deep understanding into our filter process." | PYPL (liked Braintree, refused branded checkout); Backblaze and a "TIC solutions" name cited as pass notes |
| Performance / audited record | The portfolio's tracked return since 2017, published rather than asserted. | "Track records are important. How has your portfolio done? Are you any good at what you do? If you're not you should just go home." | — |
How it serves retail investors:
One dated page per appearance — each has its stock table, talking points, and the saved transcript. Newest first.
| Date | Title / analysis page | Show | Link | Transcript | Actionable insights |
|---|---|---|---|---|---|
| 2026-AUG-07 | These Power Stocks (Not Semis) Are The Real AI Beneficiaries, Say Investing Brother Duo Up 1,300% | Monetary Matters (Jack Farley) | ▶ YouTube | transcript | actionable insights |
Appearances not yet processed — newest first. None queued yet.
For personal study — not investment advice. Source material © The Monetary Matters Network.