"Don't buy Broadcom — buy Credo. Don't buy Nokia — buy Harmonic. Don't buy Corning — buy Clearfield." Three pure-play bets on connectivity. Here's which ones the data actually backs.
Buy the specialist, not the conglomerate. Same exposure, more torque — if the small-cap executes.
The logic is sound in principle: at Broadcom, Nokia, or Corning, the connectivity product you actually want is a single-digit slice of a sprawling business — so even if that niche booms, it barely moves the giant's stock. At a focused small-cap, that same niche is the company, so the boom flows straight to the share price. You trade diversification and safety for leverage and torque. The catch: torque cuts both ways, and the three names are at very different points in their cycles. The thesis isn't equally true for all three.
| Pair | The niche | Why the giant dilutes it |
|---|---|---|
| CRDO vs Broadcom | AI data-center interconnect (AECs + optical) | Broadcom is a ~$1T+ behemoth — AI connectivity is real but buried under its broader chip/software empire |
| HLIT vs Nokia | Virtualized cable broadband (vCMTS / DOCSIS 4.0) | Nokia is a giant carrier-equipment vendor; cable-access software is a footnote to mobile RAN |
| CLFD vs Corning | Fiber-to-the-home management & delivery | Corning spans Gorilla Glass, autos, life sciences — optical fiber is one of many divisions |
If you think AI wiring, cable internet, or home fiber is about to boom, the giant that makes it (Broadcom, Nokia, Corning) won't move much because that product is a tiny part of their huge businesses. The small companies that only do that one thing will move a lot more — up if it goes well, down hard if it doesn't. The bet is on focus over safety.
The AI-wiring play. Best business of the three — at the most demanding price.
Credo makes the high-speed "smart cables" (Active Electrical Cables) and optical connectivity that wire AI GPU clusters together. It's the category leader, and crucially it's profitable: FQ4 revenue $437M (+157% YoY), 68.6% gross margin, ~50% operating margin, $169M GAAP net income, FY26 ~$1.3B (tripled), guiding 80%+ FY27 growth with optical becoming a $500–600M business via the DustPhotonics acquisition.
| Credo (CRDO) | Broadcom (AVGO) | |
|---|---|---|
| What you're buying | Pure AI-connectivity | Diversified chips + software (VMware) |
| Connectivity = % of company | ~100% | A slice of a ~$1T+ empire |
| Revenue growth | +157% | Strong but blended/slower |
| Torque to the AI-wiring theme | Maximum | Muted |
| Valuation | ~96–100x P/E | Rich but more diversified |
| The catch | Priced for perfection; fell ~15% on its own beat | Won't move much on AECs alone |
Verdict — thesis TRUE but expensive. If you want torque to AI interconnect, CRDO is unambiguously the purer bet, and it's the rare profitable hypergrowth name. But the market knows it: ~96x earnings, median analyst targets barely above spot (Goldman's below it), insider selling, and a 15% drop on a beat-and-raise. It's not a monopoly either — Marvell and Broadcom itself can contest AECs. Great company, crowded price. The pair-trade logic works; the entry is the problem.
Credo is the cleanest way to bet on AI data-center wiring, and unlike most AI names it actually makes good money. The only issue is the stock is very expensive after a huge run — so you're paying a premium price for a great company. The thesis is right; you just want a better entry.
The cable-broadband play. Just became a pure-play — and the timing looks right.
Harmonic makes virtualized cable-broadband infrastructure (cOS / vCMTS) — the software-defined gear that lets cable operators deliver gigabit internet and roll out DOCSIS 4.0 without ripping out hardware. The big structural news: Harmonic is selling its Video business (~$145M) to become a pure-play broadband company — exactly the focus the thesis wants. And the broadband numbers are accelerating:
| Harmonic (HLIT) | Nokia (NOK) | |
|---|---|---|
| What you're buying | Pure cable broadband (post-video-sale) | Global carrier equipment giant |
| Broadband growth (Q1'26) | +43% YoY · Rest-of-Market +78% | Low single digits, blended |
| Backlog | $582M · +87% YoY · record | Large but diffuse |
| Profitability | Profitable · raised FY26 EPS $0.57–0.67 | Profitable but slow-growth |
| Guidance | Raised — BB rev $475–495M | Mature |
| The catch | Customer concentration; revenue can be lumpy quarter to quarter | vCMTS is a footnote at Nokia |
Verdict — thesis STRONGEST here. Of the three, HLIT is the cleanest setup: a company actively becoming the pure-play the thesis describes, with accelerating broadband growth (+43%), a record backlog (+87%), customer diversification (Rest-of-Market now >50% of bookings), raised guidance, and it's profitable and buying back stock — all without the nosebleed multiple CRDO carries. The DOCSIS 4.0 upgrade cycle is a multi-year tailwind. The risk is lumpiness and concentration, but the risk/reward is the most balanced of the three. This is where the David-vs-Goliath thesis has the most teeth.
Harmonic makes the software that upgrades cable internet to gigabit speeds. It's selling off its old TV-video business to focus entirely on broadband — exactly the kind of focus this whole thesis is about. Its broadband sales jumped 43%, its order backlog hit a record, it's profitable, and it raised its forecast. Of the three, this one has the best mix of growth, focus, and a reasonable price.
The fiber-to-the-home play. A recovery bet — still in the trough.
Clearfield makes fiber management, protection, and delivery products for community/rural broadband — the physical fiber plumbing that connects homes. The thesis fit is clean (Corning buries fiber inside Gorilla Glass, autos, and life sciences), but unlike the other two, Clearfield is still climbing out of a downturn: fiscal Q2'26 revenue $34.4M, −15% YoY, a non-GAAP loss of $(0.04), six-month revenue down ~2%. It's a cycle-trough story, not a growth story — yet.
| Clearfield (CLFD) | Corning (GLW) | |
|---|---|---|
| What you're buying | Pure fiber-to-the-home | Glass/ceramics conglomerate |
| Revenue trend | −15% YoY · in a trough | Steady, diversified |
| Profitability | Loss-making (non-GAAP −$0.04) | Profitable |
| The signal | Backlog +39% QoQ · mgmt sees stabilizing demand | — |
| The catalyst | Rural broadband (BEAD) funding cycle + data-center fiber optionality | Broad, muted |
| FY26 est. growth | +52% implied (heavy H2 ramp) | Low single digits |
Verdict — thesis TRUE but highest risk. Clearfield is the deepest-value, highest-variance leg. The pure-play exposure is real and the setup is improving — backlog up 39% sequentially, management calling a demand bottom, a year-to-date 5% rise in Community Broadband, plus early (unquantified) data-center fiber optionality and a return-to-profitability guide. But it's still loss-making with revenue down 15%, and the bull case leans on a big H2 ramp that depends on federal broadband (BEAD) funding actually flowing and rural operators spending again. It's a recovery bet on a cycle turning, not a business already winning. Most upside if the cycle inflects; most downside if the ramp slips.
Clearfield makes the fiber gear that connects homes to the internet. It's the purest fiber bet, but it's still in a slump — sales down 15%, losing money. The hopeful signs: orders jumped 39%, management thinks demand has bottomed, and government broadband money should eventually flow. It's a bet that the worst is over and a recovery is coming — higher risk, but potentially the biggest rebound if it works.
The three, side by side. Different cycles, different risk shapes — not interchangeable.
| Metric | CRDO | HLIT | CLFD |
|---|---|---|---|
| End market | AI data center | Cable broadband | Fiber-to-the-home |
| Growth | +157% | +43% BB | −15% (trough) |
| Profitable? | Yes | Yes | No (recovering) |
| Backlog signal | strong | +87% record | +39% QoQ (off lows) |
| Valuation | ~96x · expensive | reasonable | no E (loss); P/S low |
| Catalyst | optical ramp FY27 | DOCSIS 4.0 + video-sale focus | BEAD/rural cycle turn |
| Risk shape | multiple compression | lumpiness/concentration | cycle timing |
| Thesis strength | true, pricey | strongest | true, riskiest |
The scores rank current setup, not business quality — CRDO is the best company, but HLIT has the best setup right now because it pairs real acceleration and a structural focus-improving catalyst (the video sale) with a sane valuation. CLFD has the most upside if the cycle turns, but you're paid to wait for proof. Note all three share one macro risk: they're capex-cycle names dependent on operator spending (hyperscalers for CRDO, cable MSOs for HLIT, rural telcos for CLFD) — when budgets tighten, focused small-caps feel it first and hardest. That's the flip side of the torque.
All three are good versions of the "buy the specialist" idea, but they're not the same bet. Harmonic has the best setup today — growing fast, getting more focused, fairly priced. Credo is the best company but too expensive right now. Clearfield is the cheapest with the biggest potential rebound, but it's still losing money and you're betting on a recovery that hasn't arrived. Don't treat them as one trade.
John's read. The thesis is right — but rank them, don't lump them.
- The core idea is sound: at Broadcom/Nokia/Corning the connectivity niche is a rounding error; at Credo/Harmonic/Clearfield it's the whole thesis. You get torque. But torque without timing is just volatility — so rank them.
- HLIT is the one I'd lead with. It's literally becoming the pure-play the thesis wants by selling its video unit, broadband is +43% with a record +87% backlog, guidance got raised, it's profitable and buying back stock — and it doesn't carry CRDO's nosebleed multiple. Best risk/reward of the three.
- CRDO is the best business, and I still wouldn't chase it. +157%, profitable, category leader — but ~96x, falling on its own beats, insiders selling. Watchlist it for a pullback; the company earns the premium, the entry doesn't.
- CLFD is the deep-value swing. Purest fiber play, real recovery signals (backlog +39%, demand bottoming, data-center optionality), but still loss-making and levered to the BEAD/rural cycle actually turning. Smaller size, key it to the H2 ramp showing up.
- How I'd weight it: if I had to express the whole thesis in one name, it's HLIT today. CRDO on a dip. CLFD as a smaller recovery call. Same idea, three different entries — not one basket.
Want my entry levels on all three — and which I'd buy first?
Join the Discord to find out! →Credo FQ4 FY26 8-K & transcript (Jun 1, 2026) + DustPhotonics 8-K (Apr 2026) · Harmonic Q1 2026 8-K (May 11 — Broadband +43%, backlog $582M +87%, raised guide, video-sale ~$145M) & Q4'25 8-K · Clearfield fiscal Q2'26 8-K & 10-Q (May 6 — $34.4M, −15% YoY, backlog +39% QoQ, return-to-profit guide) · analyst targets via MarketBeat/Public · ChartMill/Motley Fool/StockTitan transcripts & coverage. Figures from SEC filings & transcripts; not live IBKR pricing.