Chip Stock Investor (Nick & Kasey Rossolillo)
Nicholas Rossolillo (CFP, founder of RIA Concinnus Financial, investing since 2005, managing client money since 2012) with co-host Kasey Rossolillo; runs the Chip Stock Investor YouTube channel + research service.
Semiconductor-specialist creator duo with a large YouTube/podcast following who run a transparent, members-facing model portfolio focused entirely on chips, compute, networking, and AI-datacenter infrastructure. The principals are a real registered investment advisor, so the public commentary maps to an actual managed book rather than a paper portfolio — distinguishing them from anonymous FinTwit callers.
- Not a fund — A research, media or advisory business with no managed book to score.
- No public record — No published return series — performance cannot be verified in either direction.
- Self-reported — Headline returns are manager- or firm-reported and unaudited.
Top holdings
core long (unchanged) — Held as the core AI-datacenter / custom-silicon (XPU) and networking name; position explicitly described as unchanged in
long — Long-term hold; in 2026 they signaled "time to take a little nibble again" for decade-plus holders.
long — Held as the "value" optical/AI-networking play; framed alongside Lumentum (the "momentum" play).
long — Held alongside Coherent; optical transceiver / AI-datacenter interconnect exposure.
Recent moves
2026: maintained Broadcom as a core AI-infrastructure position (unchanged); commented favorably on adding to Nvidia for long-term holders; reaffirmed holding both Coherent (value) and Lumentum (momentum) in optical networking (chipstockinvestor.com / YouTube, 2026).
Our take
A rare viral creator pair whose calls are backed by an actual RIA, which raises the credibility floor versus anonymous accounts — the disclosed longs (AVGO, NVDA, COHR, LITE) are a clean read on where semi specialists see AI value migrating: custom silicon and optical interconnect over pure GPU beta. The limitation is the inverse of Beth Kindig's: they disclose positions but publish no performance numbers, so there is no verifiable track to size their edge. Treat as a high-quality qualitative signal on chip/optical positioning, not a return-validated book.
Two things worth separating. First, single-sector AI-infrastructure research is, to the extent it centres on the obvious beneficiaries, one macro factor expressed several ways — custom silicon, GPUs and optical components all lever to the same hyperscaler capex line.
Within-sector dispersion reads as breadth from the inside, but correlation converges toward one in exactly the drawdown where separation would matter. We could not verify a published position list, so this is a property of the category, not a documented property of this book. Second, the adviser credential does less work than the framing implies. Chip Stock Investor LLC and Concinnus Financial are separate entities, and the adviser is a two-person practice with 158 households at a $193K average — a planning book under suitability and diversification obligations, which cannot mirror the conviction the content expresses. Read the credential as "the principals are regulated and identifiable": an integrity floor, not evidence a concentrated chip strategy is run at scale or measured against anything.
- 2026-07July 2026 supplied the drawdown and dispersion widened rather than collapsed: NVIDIA +0.3% and Broadcom +3.1% for the month while Coherent fell from $331.57 to $262.89, and SOX -20.6% against S&P 500 -0.1%.
Thesis
A semiconductor-research media brand — Chip Stock Investor LLC, $10/month, "1,000+ members" (self-reported marketing copy) — run by the founder of a separate registered adviser, Concinnus Financial, Inc.: ~$30.5M AUM, 158 clients, 2 advisors, $193,037 average account (Form ADV, SEC data verified 13 Mar 2025). The differentiator the brand leans on — a real registered adviser rather than an anonymous account — is genuine but narrower than it sounds, and no performance is published for either entity.
Long-only fundamental research concentrated in one sector: semiconductors, custom silicon, optical interconnect and AI-datacenter infrastructure. Orthodox framework — revenue trend, operating margin, free cash flow, per-share earnings growth, balance sheet. Delivered as written and video commentary behind a $10/month tier, not as a weighted model portfolio with timestamped entries and exits. Free YouTube and podcast funnel sits above the paid research layer.
Assessment
- Registered-adviser status with named, identifiable principals carries an accountability and disclosure regime that anonymous FinTwit and most paid newsletters do not.
- Genuine domain depth in optical interconnect and custom silicon — mapping where AI value migrates past the GPU is a non-obvious, defensible analytical frame.
- Fee model is transparent and cheap: a flat $10/month, no performance fee, no carry, no upsell ladder dressed up as research.
- The media entity and the advisory firm are structurally separate, which is the cleaner arrangement for a creator who also manages client money.
- Zero published performance — no net-of-fee returns, no benchmark, no timestamped entries. The record is structurally unfalsifiable in either direction.
- We could not document a published position list, weights or entry dates, so even the qualitative calls cannot be scored after the fact by a reader.
- Concentration is the category's defining risk: single-sector AI-infrastructure research tends to function as thematic beta versus SOXX/SMH while being presented as stock selection.
- Inherent structural conflict where a creator discusses mid- and small-cap names an audience can move. Disclosure mitigates it; it does not remove it.
- A sector specialist's edge is worth most at a cycle trough and least at a peak — but audience and subscriptions arrive at the peak, when the analysis is least differentiated.
- 2026-07The index-beta framing understates July 2026: SMH -17.6% and SOX -20.6% while NVIDIA held +0.3% and Broadcom +3.1%. Accelerator-versus-optics weighting set the month; the record itself stays unverifiable.
- 2026-07July 2026 ran that frame in reverse: value concentrated back into the accelerators (NVIDIA +0.3%, Broadcom +3.1%) while past-the-GPU names sold off — Marvell -37.0%, Astera Labs -35.6%, Coherent -20.7%.
Record
There is no published track record, and that is the finding. No YTD, one-year or since-inception return — gross or net — and no benchmark comparison is published for the members' research. Part of this may be structural rather than evasive: under the SEC Marketing Rule, a registered adviser advertising performance takes on real compliance burden (net-of-fee presentation, prescribed periods, no cherry-picking), which plausibly explains why a two-person firm declines to publish. The reader's position is identical either way — nothing can be verified or falsified. Attribution is further blocked by the sector's largest recent move: Coherent and Lumentum re-rated hard on NVIDIA's 2 March 2026 announcement of $2B strategic investments in each, plus multibillion-dollar purchase commitments. Any optical thesis looks prescient after a shock like that; without dated, sized positions, foresight and windfall are indistinguishable.
- 2026-07That optical re-rating partly unwound: Coherent's year-to-date gain went from about +80% on 3 Jul 2026 to +28% on 29 Jul, and Lumentum's from +98% to +71%. An unpublished record hides the give-back too.
Risks & fit
- AI-datacenter capex is the single variable underneath the whole category; a hyperscaler spending pause hits semis, custom silicon and optical names together.
- Semiconductor cyclicality is severe, and no hedging, cash discipline or sell framework is disclosed alongside the research.
- Key-person risk is acute — the media brand and the two-person advisory firm rest on the same principal, with no succession structure disclosed.
- Optical component names now carry customer-concentration and strategic-investor dependency on NVIDIA, a re-rating driver that can reverse.
- Content incentives and portfolio incentives diverge: subscription growth rewards conviction and narrative, not position-sizing discipline.
The read changes on a third-party-verified, net-of-fee record with timestamped entries and sizes, showing calls were established materially before the catalysts that validated them — notably NVIDIA's 2 March 2026 optical investments — and that the book beat SOXX/SMH risk-adjusted across a full semiconductor cycle including a drawdown. That would convert "concentrated thematic beta with good storytelling" into demonstrable selection skill. The inverse resolves it too: if the highlighted names fall in line with the sector index through the next semi drawdown, the edge was sector exposure, not stock picking.
Analytically useful to readers wanting a domain-literate map of the AI-infrastructure value chain, particularly optical interconnect and custom silicon, where the framework is genuinely non-obvious. It offers little to anyone evaluating manager skill, since no measurable record exists, or assessing diversification, since the subject matter is one macro variable. Sector education from an identifiable, regulated author — not a validated strategy.
$10/month research tier; the "1,000+ members" figure is self-reported, not audited. No performance fee, no carry. Separately, Concinnus Financial charges advisory fees on ~$30.5M of client assets (Form ADV, verified 13 Mar 2025).