Broadcom Inc. (NASDAQ:AVGO) DO NOT BUY
A do-not-buy at $388.57 — and not because the business is weak. Broadcom is a genuinely elite semiconductor-plus-software franchise, but its stock is priced at a valuation extreme, about 65 times clean earnings, just as the macro tape turned against the whole AI-capex cohort. This is a price-and-risk call, not a quality call, and it stands across all three horizons.
Broadcom is one of the cleanest large-cap plays on the AI build-out — custom accelerators for hyperscalers, the networking silicon that wires data centres together, and the VMware software annuity on top. The company is excellent. The question is the price you pay for it today, and the live risk sitting underneath it.
The business is not the problem — it's elite
Let me be clear up front: this is a high-quality company. Revenue grew 47.9 percent year-on-year in the latest quarter to $22.19 billion — a genuine acceleration driven by AI custom-silicon, not a mix trick. Gross margin runs at 67 percent, operating margin 43.7 percent, return on equity 33.4 percent, and free cash flow is around 27 billion dollars a year. On our scorecard quality scores 86 and the underlying AI driver scores 78 — a strong tailwind. Broadcom is one of the few picks-and-shovels giants selling into both the AI hardware build-out and the enterprise software stack at once. None of that is in doubt.
Rev +47.9% YoY · Quality 86 · Driver 78 · ROE 33.4%
The price is the problem — 65x, 2.46x warranted
Here is where it breaks. On clean trailing earnings the stock trades at 64.8 times — that is 2.46 times the roughly 26 times multiple our discounted-cash-flow anchor says the business warrants, and it sits above the 28 times guardrail we hold semiconductors to outright. Either test alone caps the name; both fire. And it got worse, not better: the ten-year Treasury yield rose to 4.67 percent, lifting our discount rate to 9.2 percent and trimming the warranted multiple, so on a near-flat price Broadcom is more expensive than it was a fortnight ago. The forward P/E looks tame only because the Street models earnings roughly tripling — the exact optimism the anchor is built to discount. Valuation scores 25.
Clean P/E 64.8x · 2.46x warranted · 28x guardrail · Valuation 25
A live, armed de-rating catalyst
Expensive on its own would cap the name at hold. What pushes it to do-not-buy is a live catalyst. Our do-not-buy Trigger 2(b) fires when a name is deep in the expensive band AND a material systemic de-rating catalyst is armed. Both are true. The 30th of July macro report re-confirmed the S&P 500 concentration and AI earnings-quality tail as armed — the top ten names are about 41 percent of the index and breadth is narrowing. Broadcom belongs squarely in that cohort: its multiple and its revenue acceleration are directly levered to the hyperscaler AI-capex trade. Its earnings are clean, so the earnings-quality half of the tail doesn't apply — but the AI-capex-leverage half does, and one arm is enough.
Trigger 2(b) fired · Tail: armed · DNB all 3 horizons · Timing 47
What could go wrong
Now the honest other side — because for a do-not-buy the real risk is that we're wrong on the upside. What makes this call wrong is simple: the AI-capex melt-up just keeps going. Our own base case is $455, about 17 percent above today, and the bull case is $560, roughly 44 percent up — both sit above the current $388.57. The probability-weighted fair value works out around $421, also above spot. So on the modal path the stock drifts higher, and this is deliberately not a prediction that it falls. It is a risk-adjusted call: sitting against that upside is an armed bear case at $215, down about 45 percent, that can erase two-plus years of gains in a single re-rating. When a fat, live left tail is that severe, chasing an obviously expensive name is poor risk-reward even when the mean is positive.
Risk vs Reward
The three paths, all 12-month. Bull $560, 25 percent odds: the AI-capex supercycle sustains and broadens, custom-ASIC wins accelerate, and the market keeps paying a premium as earnings compound into it — about plus 44 percent. Base $455, 50 percent and the most probable: the franchise compounds but the multiple grinds lower as the ten-year stays elevated, earnings growth roughly offsetting a modest de-rating — about plus 17 percent. Bear $215, 25 percent and mandatory here because it is armed: the concentration tail fires, the whole AI-capex cohort re-rates, and Broadcom's 64.8 times compresses toward the low-20s, with a Marvell or in-house-silicon share scare adding a company-specific leg — about minus 45 percent. That live bear is precisely why the signal is do-not-buy rather than hold.
The verdict
So: do not buy, on the short, medium and long horizons alike. This is not a knock on Broadcom — the business is elite, the AI tailwind is real, and quality scores 86. It is a price-and-risk verdict. At about 65 times clean earnings, 2.46 times what the business warrants, with a live and armed systemic de-rating catalyst sitting underneath the entire AI-capex cohort, the risk-adjusted entry today is poor. The conviction is Wait. We don't chase a name that is obviously overpriced and carrying a live puncture risk — however good the company is.
That's my read on Broadcom. Financial Freedom. Together.
Read the full report on donatien.ca →
