I keep buying Broadcom (NASDAQ:AVGO | AVGO Price Prediction) and I am not embarrassed to say it out loud. Every paycheck window that opens, I add. The post-earnings selloff in June only reinforced my conviction, confirming that the market keeps handing long-term holders a discount on the one AI infrastructure name that also pays me to wait.
The core of my thesis is simple. Hock Tan built a company that sells the picks and shovels the hyperscalers cannot buy anywhere else, and he pairs that with software cash flows from VMware and a dividend I can plan a retirement around. That combination is why I keep clicking buy.
The Numbers That Keep Me Coming Back
Start with the AI engine. Q2 FY2026 AI semiconductor revenue hit $10.80 billion, up 143% year-over-year, and management guided Q3 AI revenue to $16.0 billion, growth of over 200%. CEO Hock Tan told shareholders: “The momentum continues and in Q3 we expect semiconductor revenue from AI to grow over 200 percent year-over-year to $16.0 billion.” Total Q2 revenue landed at $22.19 billion, up 47.9%, with non-GAAP EPS of $2.44, the eighth consecutive quarter of beating expectations.
Next comes the cash. Q2 free cash flow was $10.26 billion, or 46% of revenue, with an adjusted EBITDA margin of 69%. Full-year FY2025 free cash flow reached $26.91 billion. Cash generation at that scale speaks for itself.
Then the dividend. Broadcom has raised its payout 15 consecutive years since fiscal 2011, with the last hike lifting the quarterly to $0.65, a 10% increase. That is the kind of streak I plan a retirement withdrawal schedule around.
Why This One, Not NVIDIA or AMD
Every reader lands first on NVIDIA (NASDAQ:NVDA) or Advanced Micro Devices (NASDAQ:AMD) when they think AI chips. I keep landing here instead because Broadcom sells something distinct from NVIDIA’s lineup: custom ASICs designed to spec for individual hyperscalers, plus the networking silicon that stitches those clusters together. The Apple relationship, reported by retail as a $30B+ custom AI chip deal through 2031, is the template. Add a Semiconductor Solutions segment growing 79% year-over-year and an Infrastructure Software segment producing $7.18 billion of high-margin recurring revenue, and I own a business model neither pure GPU maker offers, with a dividend streak neither matches.
The Risk I Refuse to Wave Off
Customer concentration is real. Broadcom itself flags dependence on a limited number of large customers and significant indebtedness. If one hyperscaler cuts its custom ASIC roadmap, a quarter gets ugly fast. I stay because free cash flow of $10.26 billion in a single quarter services the debt with room to spare, and cash on the balance sheet grew to $19.63 billion, up 107.22% year-over-year. That is defense I can live with.
The Line in the Sand
Shares closed at $386.50 on July 21, roughly 6% off the 52-week high of $494.18. The Wall Street consensus target sits at $524.51 with 44 buys against 4 holds and zero sells. The base-case fair value modeled at $404.96 is my accumulation ceiling. Under that number, my finger stays on the buy button. Above it, I let the dividend do the work.
The pivot back in is back on because Broadcom is compounding cash at hyperscaler speed while paying me a raise every year. I will keep buying until the price catches up to the business, and then I will keep holding while the business keeps running ahead.
Contact [email protected] for any questions or corrections.