Financebotresearch desk研究台

Discovery › screen

Screen value gems software systems ai hardware

Date: 2026-07-13 Criteria: Multi-round net-casting ("every company is a bubble in the ocean — let them surface"), then moat-read the survivors. Hunt better values — beaten-down but healthy — across software, systems, AI, and hardware/infrastructure. Fresh round of hidden gems; names not already on the watchlist. Method: 2 rounds of wide web search → filtered against existing coverage → quick Fundamentals health check (ROIC, margins, revenue growth, valuation, FCF/leverage) on 8 finalists via roic.ai + Alpha Vantage.


The filter this round

The net came back full, but most of what surfaced had already surfaced — either held/watched (ANET, AVGO, AMD, QCOM, SNPS, AMAT, MANH, PTC, CEG, NVTS) or run so hard they're no longer value:

  • Cut for price, not quality: Fabrinet (FN) — 44x fwd P/E, EV/FCF ~540; Vertiv (VRT), Credo (CRDO), Super Micro (SMCI), Seagate (STX) — all momentum, priced for perfection. Great AI-infra exposure, terrible margin of safety today.
  • Cut for being expensive and low-return: Onto Innovation (ONTO) — P/E 95, EV/EBITDA 47, ROIC 5%. Beautiful secular story (advanced-packaging metrology), zero value cushion after the run from $86→$205.
  • Moved to speculative sleeve: Ambarella (AMBA) — unprofitable (-17% net margin), fwd P/E 98. Edge-AI vision pure play, not a value name. Belongs next to your QBTS/EVGO spec bucket, not here.

What's left below actually clears the "beaten and healthy" bar.


Ranked Shortlist

# Company Ticker Sector Mkt Cap Price Off 52w High Fwd P/E EV/EBITDA P/FCF ROIC Gross Mgn Rev Growth (YoY) Verdict
1 Photronics PLAB Semiconductors (photomasks) $1.8B ~$30 −46% 13.3x 3.5x ~7x ROE 13% 34% −0.5% (flat) 💎 Deep value
2 ON Semiconductor ON Semiconductors (power/SiC, edge-AI) $24B $62 −16% ~17x 13.7x 15.5x 8.7% 37.5% trough +8%* 🟢 Cyclical value
3 Varonis VRNS Cybersecurity (data security / DSPM) $2.5B $21 −66% n/m (GAAP loss) n/m 16.5x neg (transition) 78% ~mid-teens ARR ⚠️ Turnaround-value
4 Advanced Energy AEIS Systems (data-center power, semi plasma) $12.3B ~$336 −15% 36x 43x high 8.7%→rising 40% +26% 🔵 Great biz, wait
5 Amkor AMKR Semiconductors (advanced packaging/OSAT) $17.5B ~$70 −27% 37x 13.7x ~40x (capex) 8.7% 14% +27.5% 🟡 Cyclical, thin
6 Kulicke & Soffa KLIC Semi equipment (bonding, TCB) $3.4B $66 −15% ~30x fwd 44x 124x 4.2% 48% trough 🟡 Ran off lows

* ON's reported growth is off a cyclical trough; earnings are depressed, which inflates the trailing P/E.


The gems — why each made the cut

1. Photronics (PLAB) — 💎 the cheapest quality name on the board

The single cleanest "hidden gem" this round. Photronics is one of only three merchant photomask makers on earth (with Toppan and DNP) — the stencils every chip fab needs to pattern wafers. That's a quiet, real oligopoly moat with high switching costs (masks are qualified per-process, per-customer).

  • Valuation is the story: EV/EBITDA 3.5x, trailing P/E 11, P/B 1.4, and a net-cash balance sheet. 18.5% net margin, ROE 13%. You are paying almost nothing over tangible book for a profitable duopoly-adjacent business.
  • The catch (be honest): revenue is flat (−0.5% YoY). The whole question is value vs. value-trap — is flat a pause (China fab buildout, high-end EUV mask mix coming) or a ceiling? At 3.5x EBITDA with net cash, the market is pricing permanent stagnation. If growth so much as inflects, this re-rates hard.
  • Fit: fills a genuine gap — you own chip designers (AMD, AVGO, QCOM) and EDA (SNPS) but nothing in the mask/materials layer.

2. ON Semiconductor (ON) — 🟢 buy the trough, not the headline

Classic cyclical value: the optically-scary trailing P/E (43x) is trough earnings, not an expensive business. Look through it — P/FCF 15.5x, EV/EBITDA 13.7x, 37.5% gross margin, 17.5% op margin, ROIC 8.7% even at the bottom of an auto/industrial semi downturn.

  • Moat: #1 in silicon-carbide (SiC) power for EVs and industrial, plus intelligent-sensing. Design wins are multi-year and sticky. The edge-AI angle (inference needs power everywhere, not just in data centers) is free optionality the market isn't paying for.
  • Where it sits: $62, ~16% off highs but well above its $31 trough-low — so you've missed the bottom, but you're buying a quality franchise at mid-cycle-cheap multiples before the auto/industrial cycle turns.
  • Risk: cyclical timing. If the downturn drags, trough earnings get troughier. Position as an accumulate-on-weakness, not an all-in.

3. Varonis (VRNS) — ⚠️ the asymmetric one

Down 66% from its high — the deepest drawdown here. GAAP-unprofitable because it's mid SaaS-transition (subscription conversion depresses reported revenue and margins), but the cash engine is intact: P/FCF 16.5x on 78% gross margins. Real FCF is hiding under GAAP losses.

  • What it is: data-security / DSPM (data security posture management) — it maps and locks down where sensitive data actually lives. That's exactly the problem AI agents and copilots make scarier (every LLM is a new way to over-expose data). Structural tailwind, adjacent to your cyber sleeve (PANW/CRWD/QLYS/RBRK) without overlapping it.
  • The risk: the −66% isn't only the SaaS-transition optics — need to confirm ARR growth is holding (mid-teens) and net-retention hasn't cracked before touching it. This is the highest-risk name on the list; size it like a turnaround, not a compounder.

Watch, don't chase (Tier 2)

  • Advanced Energy (AEIS) — the best business found this round: data-center power (new 800V DC platform for megawatt racks), semi plasma-power at leading-edge nodes, 40% gross margin, +26% revenue, data-center revenue doubling. But it's run from $128→$336 and trades at 36x fwd / 43x EBITDA. No margin of safety. Put it on the watchlist with an entry target well below here and wait for a semi-capex wobble.
  • Amkor (AMKR) — advanced-packaging (OSAT) is essential AI-chip plumbing and PEG is a cheap 0.76, but it's capital-intensive ($2.5–3B capex), thin-margin (6% net), cyclical, beta 2.2, and family-controlled. Cheaper names with better economics exist above.
  • Kulicke & Soffa (KLIC) — thermocompression bonding (TCB) is real advanced-packaging optionality, but the stock already doubled off its $27 low to $66, so you're paying a troughed 44x EBITDA / 124x FCF after the rally. The value window largely closed.

Next steps

Add to Watchlist.md? My recommendation: - PLAB → Semiconductors, ⭐ shortlist candidate (deep-value, net cash) — pending a value-vs-trap /analyze. - ON → Semiconductors, watch (cyclical value, accumulate on weakness). - VRNS → Cybersecurity, watch (turnaround-value, higher risk). - AEIS → Systems/Industrials, watch (quality; entry target below current price).

Run a full /analyze on any of them. My priority order for a deep dive: 1. PLAB — the value-vs-value-trap question (is flat revenue a pause or a ceiling?) is the whole thesis and worth resolving. 2. VRNS — need to verify the −66% is transition-optics, not a broken business, before it earns a real conviction score. 3. ON — moat + cycle-timing read to set an accumulation zone.

Say the word on which to add and which to /analyze first.



Round 3 — Follow-up: Smaller market caps, genuine quality only

Question asked: are there smaller-cap names that are genuinely quality companies (not the speculative micro-cap sleeve) — profitable, high-return, real moat, clean balance sheet?

Honest headline: yes, but only two really clear the bar — and true small-cap quality is rare. Most quality in software/systems/AI/hardware got bid up to mid/large cap years ago. What's left small is usually small for a reason (unprofitable, cyclical, governance-flawed, or a contested niche). Five candidates checked:

Company Ticker Mkt Cap ROE ROIC Gross / Op Mgn P/E EV/EBITDA P/FCF Balance sheet Verdict
Alarm.com ALRM ~$2.1B 26.5% 10.5%* 66% / 13% 16.8x 12.3x 13x Positive equity, low debt Genuine quality
Iridium IRDM ~$2.9B n/m (neg. equity) 12.6%* 72% / 26% 27.9x 10.5x 7.2x Levered (buybacks, net debt ~$1.6B) 🟢 Quality cash machine, but geared
PagerDuty PD ~$0.5B n/m neg. 85% / 5% 3.2x⚠️ 15x 4.5x Small net cash 🟡 Cheap, not quality
ACM Research ACMR ~$2.6B 28% 5.6% 44% / 12% 28x 15.3x 151x⚠️ Cheap on tangible book ⚠️ China-governance risk
Aehr Test AEHR ~$1.1B neg. neg. 31% / −34% loss neg. neg. Net cash ❌ Unprofitable + pricey

* roic.ai return-on-capital measure; ALRM's stricter ROIC screens ~6%, ROE 26.5%. IRDM's ROIC is understated by its negative-equity capital structure.

The two that qualify

🥇 Alarm.com (ALRM) — the cleanest genuine-quality small-cap found. Cloud SaaS platform for smart-home / security / video, sold through a locked-in network of ~11,000 professional installers (dealers) — that dealer channel is the moat; homeowners don't switch platforms, and dealers don't re-tool. 26.5% ROE, 66% gross margin, 12% net margin, positive equity, low debt, ~13x FCF, 16.8x P/E — and it's trading at $43, near its 52-week low ($41.5) vs. a $60.76 high. Recurring SaaS revenue (SaaS is ~85% of the total), profitable for years. This is a real quality compounder at a fair-to-cheap price, and it fills a genuine gap (you own no consumer-IoT / connected-hardware SaaS). Best answer to your question.

🥈 Iridium (IRDM) — quality cash machine, but read the leverage. Owns the only truly global, cross-linked LEO satellite constellation (L-band) — mission-critical for DoD, maritime, aviation, and small-form IoT where Starlink doesn't fully reach. 72% gross margin, 50% EBITDA margin, ~7x FCF, pays a growing dividend, and buys back stock aggressively. The catches: (1) ~$2.9B cap is really mid-cap, not small; (2) the balance sheet carries net debt (~$1.6B) and negative book equity from years of buybacks — the business easily services it, but it's not the "clean balance sheet" you asked for; (3) Starlink Direct-to-Cell is a real long-arc competitive question. Quality and cheap, but geared.

The three that don't

  • PagerDuty (PD) — genuinely cheap (P/FCF 4.5, EV/S 1.15) and finally GAAP-profitable, but the headline P/E of 3x is a tax-benefit artifact (real op margin is only ~5%). Growth has stalled to low single digits and the incident-management niche is contested by Atlassian, ServiceNow, and Splunk. This is a cheap potential takeout target, not a quality compounder. Interesting as deep-value, not as quality.
  • ACM Research (ACMR) — fast-growing (+34%) semi-cleaning-equipment maker at a cheap tangible book, but its operating heart is China-based (ACM Shanghai, customers SMIC/Hua Hong), carrying real US–China export-control / delisting / governance-structure risk, and FCF conversion is poor (P/FCF 151x on heavy capex). Fails the "clean quality" test on governance and cash conversion.
  • Aehr Test (AEHR) — SiC/AI burn-in test niche, but it's unprofitable TTM (revenue fell on customer concentration) and expensive (EV/Sales 24x after running 6x off its low). Speculative, not quality.

Bottom line for you

If you want a genuinely high-quality business at a smaller cap, ALRM is the one — profitable, moaty, fairly priced, near its lows, and it fills a real portfolio gap. IRDM is a strong quality-and-cheap idea if you can accept the leverage and that it's really mid-cap. Everything smaller that surfaced trades cheap for a reason.

Offer: add ALRM (Systems/IoT SaaS, watch) and/or IRDM (Infrastructure/Satellite, watch) to the watchlist? And I'd prioritize a full /analyze on ALRM to pressure-test the dealer moat and set an entry zone.


Data: roic.ai (TTM ratios, ROIC, margins), Alpha Vantage (company overviews, valuation), web research (candidate discovery, catalysts). Screen-tier health check — not a full pipeline. Valuations as of 2026-07-13.