Credo: The $1.3 Billion Cable Company Hiding in the AI Build-Out
Credo Technology makes the unglamorous part of an AI data center — the cables, and the tiny chips inside them that rebuild the signal before it breaks down. Revenue more than tripled to over $1.3 billion in fiscal 2026, management guides for 80%+ growth again, and 18 of 19 analysts are bullish with targets up to $350. But one customer was a third of last quarter's revenue, the top four were 87%, the CEO and CFO both sold on July 1 into a net $174M of insider selling — and the stock just fell ~24% in a week. This is the moat, the concentration, and the insider tape.
Credo's fiscal 2026 revenue — more than tripled in a year
$1.3B+
▲ +197 revenue in fiscal 2026, up from $437M a year earlier and under $200M the year before that — with management guiding 80%+ growth again in FY27. The catch is who it comes from: one customer was 34% of last quarter's revenue.
The flashiest trades in AI are the GPUs. One of the quietest is the cable. Credo Technology builds the active electrical cables and retimer chips that keep an AI cluster's thousands of chips talking without the signal decaying — and the numbers are staggering: revenue went from under $200 million to over $1.3 billion in two years, more than tripling in fiscal 2026 alone, with 80%+ growth guided again for FY27 and gross margins near 68%. Wall Street is nearly unanimous: 18 of 19 analysts bullish, price targets raised to as high as $350. So why did the CEO and CFO both sell on July 1? The data underneath the story is where it gets uncomfortable — one customer was 34% of last quarter's revenue and the top four were 87%, insiders sold a net $174M (8 buys against 259 sells), and even after a ~24% weekly drop the stock is not cheap. The moat is real; the concentration and the insider tape are the price of admission.
The contradiction
Credo is a real AI-infrastructure winner growing 80%+ a year — but the revenue leans on a handful of hyperscalers, the CEO and CFO sold into the run on July 1, and even after a 24% weekly drop the stock trades near 100x earnings. The moat is genuine; the concentration and the insider tape are what you are underwriting when you pay the multiple.
What the headline says
The story Wall Street is buying
A hidden AI winner: revenue tripled to $1.3B, 18 of 19 analysts bullish, targets to $350
What the data says
What the data underneath says
One customer = a third of revenue, top four = 87%; insiders sold a net $174M; ~100x earnings — and down 24% in a week
Chapter 02
The catch: one customer is a third of the business
Last quarter (fiscal Q4 2026) the CFO disclosed four customers each above 10% of revenue — 34%, 27%, 16% and 10%. That means one customer was more than a third of the business, and the top four were 87%. In fiscal Q3 the top three alone were 88%.
Here is what the growth rests on. On the fiscal Q4 2026 call, CFO Dan Fleming walked through customer concentration in unusual detail: Credo had “four 10% customers in Q4,” ranging “from a high of 34% down to 10%” — specifically 34% for the largest, 27% for the second, 16% for the third and 10% for the fourth. Add them up and four customers were 87% of revenue; the single largest was more than a third of the entire company. This is not a one-quarter artifact: in fiscal Q3, the CFO said the top three customers were 39%, 32% and 17% — 88% between just three names. Management frames the diversification as improving — it expects three-to-four 10%-plus customers going forward and is adding neoclouds beyond its core hyperscalers — and stresses that “customer mix will vary from quarter to quarter.” That is fair, and the direction of travel is toward more customers, not fewer. But the risk is arithmetic: when a third of your revenue is one buyer, that buyer’s deployment schedule — a pause, a pull-forward, a design change — moves the whole company’s quarter. A hyperscaler is a wonderful customer right up until it is a concentration risk, and today Credo has both at once.
"34% for the largest, 27% for the second largest, 16% for the third." — CFO Dan Fleming on Q4 customer mix
Customer mix, fiscal Q4 2026: share of revenue by 10%+ customer
Source: Credo fiscal Q4 2026 earnings call, CFO customer-mix disclosure
Largest customer
34%more than a third of Q4 revenue
Top four
87%four names, most of the business
Management's view
3–4 big customers"mix will vary quarter to quarter"
Chapter 03
The Street is nearly all-in
Analysts are almost unanimous: 18 of 19 are bullish, one neutral, none bearish. Targets were raised hard through June — Stifel to $350, B of A to $340, Evercore initiated at $325 — with a mean near $279. After the drawdown, the stock trades below every published target.
Wall Street loves it, and it is not subtle about it. The current recommendation split is 18 bullish (four strong-buy, 14 buy) to a single hold, with zero sell ratings. And the targets were marched higher through June as the stock ran: Stifel lifted its target from $250 to $350, Bank of America from $252 to $340, Roth from $200 to $300, Mizuho to $290, and Evercore initiated coverage at $325 with an Outperform. Only Rosenblatt sits off to the side with a Neutral and a $215 target. The mean target is about $279. Here is the part that matters after the last two weeks: with the stock around $196, Credo now trades below every one of those published targets — even the lowest, Rosenblatt’s $215. On its face that is a wall of implied upside (the mean sits ~40% above the price). But it is worth remembering what a target actually is: a bullish desk’s estimate, and most of these were raised while the stock was near its highs, not after it fell. The forward EPS revision breadth backs the enthusiasm — six estimates up versus one down in the last 30 days. So the Street’s message is clear and near-unanimous. The question this piece keeps asking is whether the people closest to the company agree.
"18 of 19 analysts bullish, none bearish — and after the drop, the stock is below every target."
Analyst price targets vs the price after the drop
Source: MarketDecode analyst snapshot, July 17, 2026 (recent published targets)
Ratings
18 bullish / 1 hold / 0 sellnear-unanimous
Mean target
$279~40% above the price
Targets raised
to $340–$350Stifel, B of A, in June
Chapter 04
But the insiders went the other way
While the Street raised targets, Credo insiders sold a net $174M — 8 buys against 259 sells — and the signal reads bearish. The CEO and CFO both sold on July 1, near the stock's highs. Insider sales are never a timing signal alone, but the scale and the timing lean the same way as the concentration risk.
Now the tape from inside the building. Over the trailing window, Credo insiders were net sellers to the tune of $174 million — 259 sales against just eight buys — which the scanner flags as a bearish signal. This is not anonymous rank-and-file selling: on July 1, the two most important insiders both sold — President and CEO William Brennan and CFO Daniel Fleming — alongside another senior insider, into a stock that had roughly tripled off its 52-week low of $85 and was trading near its highs. Insider selling is genuinely noisy: executives sell for taxes, diversification, pre-set 10b5-1 plans, and a hundred personal reasons, and a single sale proves nothing. But this is not a single sale; it is a lopsided body of evidence — an 8-to-259 buy/sell ratio and $174M net — stacked on top of a stock at a triple-digit earnings multiple and a business where one customer is a third of revenue. When the CEO, the CFO, and Wall Street disagree this sharply about the same stock on the same week, one side is early. The insider tape does not tell you who — but it tells you the people with the most information were reducing, not adding, right at the top.
"The Street raised targets. The CEO and CFO sold. One side is early."
Insider transactions, trailing window: buys vs sells
Source: MarketDecode insider snapshot, July 2026 (buy/sell counts; net value)
Net insider selling
−$174M8 buys vs 259 sells
Who sold July 1
CEO + CFOBrennan & Fleming, near highs
Signal
Bearishlopsided buy/sell ratio
Chapter 05
The price cracked — and the verdict
The stock fell about 24% in seven sessions, from $266 to $196, down six of seven days and underperforming the S&P by roughly the same margin. It sits ~34% below its 52-week high — and even after the drop, at ~100x trailing earnings, it is not cheap. Real hidden winner, real risks, watchful setup.
The tape has started to agree with the insiders, at least for now. Over the last seven sessions Credo fell about 24%, from $266 on July 9 to $196 on July 17 — down six of seven days, while the S&P was roughly flat, so nearly all of that was Credo-specific, not the market. The stock is now about 34% below its 52-week high of $309 and has slipped below its 20-day average. And here is the uncomfortable part for the bull case: even after a 24% haircut, this is not a cheap stock. On the last clean snapshot it carried a trailing price-to-earnings multiple around 100x and a price-to-sales ratio in the mid-30s; a beta above three means it moves violently in both directions. So the verdict is not “avoid Credo” — the moat is real, the growth is real, and a company compounding revenue 80%+ with 68% gross margins in the AI build-out deserves a premium. The thesis is narrower and it is the recurring MarketDecode one: the quality of the business and the quality of the entry price are two different questions. Today Credo is a genuine hidden winner whose revenue is dangerously concentrated, whose insiders sold into the top, and whose stock is still expensive after a hard drop. The graders are concrete. Watch the next print (fiscal Q1 2027, expected around early September, unconfirmed) for whether the top-customer share falls and the 80%+ growth holds; watch whether insider selling stops or accelerates; and watch whether the stock stabilizes above its recent lows or keeps unwinding toward the analyst-target band it just fell out of.
"The quality of the business and the quality of the entry price are two different questions."
Credo daily close: a ~24% drop in seven sessions
Source: MarketDecode price tape, July 17, 2026 (daily closes)
Last 7 sessions
−24%$266 → $196, 6 of 7 red
From 52-wk high
−34%high was $309
Still not cheap
~100x earningsbeta > 3, P/S in the 30s
Resolution window — 3 months
What would confirm or invalidate this read
Confirmation
By ~2026-10-15: (a) the next print (fiscal Q1 2027, ~early September) holds the 80%+ FY27 revenue-growth guide; (b) the largest-customer share falls below ~34% and top-four concentration eases below 87% as neoclouds/new hyperscalers ramp; (c) insider net selling stops or reverses from the -$174M / 8-buys-vs-259-sells pace; (d) the stock stabilizes above its recent low (~$196) and analysts hold or raise targets.
Invalidation
By ~2026-10-15: (a) a top customer pauses or pulls back and a quarter misses on the concentration risk; (b) the FY27 growth guide is cut below 80%; (c) insider selling accelerates beyond the -$174M pace; (d) the stock breaks and holds below ~$196 and keeps unwinding toward the $215+ analyst-target band it fell out of, with targets cut to chase it down.