Credo Means "I Believe." Should You?

Gavin Baker just bought it. His own filings say: not yet, not big.

Credo is Latin for "I believe." The name is either marketing genius or an unintentional confession. Every uncomfortable customer number in this article traces back to the same question. Are you buying a business, or a belief.

I'm Simon Severino. I run the Compounding Portfolio: eight to nine companies I plan to hold for decades. Screened on moats, capital discipline, and the price I pay for both. My only two winners this year, Arista Networks and Amazon, both rode the same AI networking wave. So when Gavin Baker opened a $102 million position in Credo Technology Group this spring, roughly two percent of his book, I went looking for winner number three.

Here is what a modern AI rack actually looks like inside. Nvidia's GB200 NVL72 is the reference design running most frontier training clusters right now. It uses more than five thousand copper cables to connect its GPUs to each other. Five thousand cables, one rack, moving data so fast that a few centimeters of bad signal integrity can stall the entire run. That is the bottleneck Credo sells into.

Its core product is the Active Electrical Cable, a copper cable with a retiming chip built into each end. Passive copper degrades over distance. Active copper corrects the signal and keeps moving. It draws roughly half the power of the optical alternative, in a fraction of the volume of a passive cable. CEO Bill Brennan puts it plainly: "Where you can use copper, you will use copper. It is not an either-or situation." Credo has also built a real optical business through the DustPhotonics acquisition, closed in May, plus a PCIe retimer line playing catch-up to the market leader. A company hedging its own bet on copper is telling you something about how durable it thinks that advantage really is.

The financial results are not in dispute. Fourth-quarter revenue hit $437 million, up 157 percent year over year. Full fiscal 2026 revenue reached $1.335 billion, up roughly 206 percent. Non-GAAP net margin closed the year near 50 percent. Free cash flow over the trailing year ran around $407 million, against a balance sheet holding $1.4 billion in cash and zero debt. Guidance for fiscal 2027 calls for revenue growth above 80 percent. These are not the numbers of a company scraping by.

Now the part that separates a Compounding Portfolio candidate from a hot trade. Credo names its own real competitors in its own 10-K: Broadcom, Marvell, and Astera Labs, each described as a "much larger" incumbent. Broadcom is building SerDes capability directly into its Tomahawk switch silicon, narrowing the standalone opportunity Credo depends on. Astera Labs beat Credo to market in PCIe 6 retimers. Every claim Credo makes about technical superiority, including "up to 1,000 times greater reliability" versus optical modules, comes from Credo itself. I found no independent benchmark anywhere confirming it.

Then there is the number that should slow down anyone reaching for the buy button. Credo's top three customers account for 84 percent of revenue, and the top ten account for roughly 90 percent. Its own filings state plainly that customers can cancel orders on short notice, without penalty. Analysts widely attribute the largest customer to Amazon, though Credo has never named it in a filing. Management says the company now ships in volume to five of six major hyperscalers. That diversification is underway. It is not finished.

Gavin Baker's position deserves an honest accounting too. Atreides held zero Credo shares through the end of 2025. Then, in the first quarter of 2026, it opened a position of 1,091,233 shares worth $102 million. Roughly two percent of his book. Compare that to Astera Labs, which sits at 7.37 percent of Atreides' portfolio, his second-largest holding. Baker is intrigued by Credo. He is not betting the fund on it. The one verified quote tied specifically to Credo, from the DustPhotonics deal, calls the acquired silicon photonics "a natural extension to Credo's existing capabilities." Reasonable. Not evangelical.

So is Credo the third leg of the Arista-Amazon trade. No. Arista and Amazon are proven across cycles, diversified across customers, and priced accordingly. Credo trades near 100 times trailing earnings and 36 times sales. Richer than Amazon. Richer than Arista. Cheaper only next to Astera Labs. Same theme. But 84 percent of revenue sits on contracts either side can walk away from tomorrow. Different risk category entirely.

Five scenarios, three-year horizon, current price $257:

Super Bull. AI capex keeps compounding, Credo wins broader hyperscaler share, optical scales past guidance. Revenue reaches $7.0 billion by fiscal 2029, margin holds near 53 percent, market pays 40 times earnings. Target near $705. Return: roughly +175 percent, about 40 percent a year.

Bull. Guidance holds roughly as issued, growth normalizes gradually. Revenue near $4.5 billion, margin 50 percent, multiple compresses to 32 times. Target near $350. Return: roughly +37 percent, about 11 percent a year.

Base Case. Growth decelerates on a normal semiconductor curve, concentration risk lingers without a crisis, multiple normalizes to peer average near 28 times. Revenue near $3.3 billion, margin 47 percent. Target near $217. Return: roughly -16 percent.

Bear. A major customer cuts or delays orders, a real possibility given the cancel-without-penalty contract terms Credo discloses itself. Revenue near $2.0 billion, margin compresses to 40 percent, multiple falls to 20 times. Target near $82. Return: roughly -68 percent.

Super Bear. The concentrated customer walks entirely, or copper's reach advantage collapses faster than management's own three-year timeline for optical displacement. Revenue near $1.1 billion, margin falls to 30 percent, multiple falls to 15 times. Target near $27. Return: roughly -90 percent.

The Scorecard, out of 100:

1. Essential or nice-to-have: 8. AI clusters genuinely need faster interconnect, though not necessarily this vendor's interconnect.
2. Current moats: 4. Technical claims are real but unverified by anyone outside the company.
3. Moats expanding: 5. The DustPhotonics move hedges a weakness more than it deepens a strength.
4. Balance sheet strength: 9. $1.4 billion cash, zero debt.
5. EPS accelerating: 9. Triple-digit growth, guided to continue.
6. Net margins increasing: 8. Near 50 percent and guided higher.
7. ROIC profile: 7. Strong today, unproven across a downcycle.
8. Reinvestment rate: 6. Growing opex and a fresh acquisition. Sensible. Not yet proven.
9. Capital return: 2. No dividend, no buyback, all reinvestment.
10. Valuation, reverse DCF: 3. Priced for years of flawless execution at 100 times trailing earnings.

Overall Score: 61/100. A real business with real numbers, riding a real wave, priced like the risk has already disappeared. It has not.

The lesson here isn't Credo specifically. It's this: when a name you respect buys something, check the position size before you check the ticker. Two percent of a book is curiosity. Seven percent is conviction.

Credo is asking you to believe. I want proof first. I am watching this one, not buying it, and every position in the Compounding Portfolio gets tracked in the open, before I trade it, not after.

See it live: https://www.strategysprints.com

Happy hunting. Simon & The Sprinters 🐬⚡️🐆