Broadcom is selling for only 18x FY-Oct 2027 earnings, growing at over 50% annually for the next three years, with its profit margins exceeding Nvidia’s. It is valued at just 10x sales growing over 35% for the next three years. That has a significant margin of safety.
For the last few years, Broadcom has been the leading competitor to Nvidia’s GPUs, surging from a measly $12Bn in FY2024 to a massive $115Bn in FY 2027. That’s pretty heady growth indeed and after its earnings call on Sept 2nd, 2026, I’m surprised that the markets aren’t rewarding it enough.
For the 2026 quarter, Broadcom beat consensus revenue estimates by $160Mn to grow 86% to $29. 6Bn. Its AI revenue grew a whopping 221% to $16.7Bn with expected growth of 236% in the current quarter. Going another step further, just like Nvidia (NVDA) last week, it announced heady growth for AI chips sales to $115Bn in FY2027 and$230Bn in 2028, clearly making a mockery of the supposed rule of large numbers slowing growth down. At $115Bn it would be about 25% of Nvidia’s AI revenue for the same period, to give you an idea of how far Broadcom has reached. Next competitor AMD is still miles behind without about 40% of Broadcom’s revenue.
CEO Hock Tan emphasized on the earnings call that demand was far ahead of supply, echoing Jensen Huang’s supply chain constrained lament, and it's showing up in some strong quarterly numbers. Adjusted EPS rose 96% to $3.32, a whisker above consensus estimates of $3.22. $29.6Bn of sales also beat estimates of $29.2Bn growing 86% YoY.
Broadcom’s adjusted operating margin of nearly 68%, beat Nvidia’s OPM of 66.45% - these are insane numbers, and perhaps either the markets are missing the stellar operating metrics or perhaps savvier than the rest of us in saying that this cannot sustain….
I think the tepid response to Broadcom’s earnings was because guidance was just par for the course, with “just” 93% increase in sales $34.8Bn – in line with estimates. Perhaps, another small quibble was the slightly lower OPM of 66% for the next quarter. I’m not complaining, the post market drop has made the stock more attractive.
Broadcom’s largest client is Alphabet (GOOG) which is making a massive push into systems on chips, clearly wanting to reduce its reliance on Nvidia, and its decade long partnership with Broadcom for its Trainium range has been a big boon for both of them.
Besides the minor quibbles, the big risks and challenges are contingent liability from data center buildouts, and competition. The June announcement of a fund from Apollo and Blackstone to finance a large buildout of data centers deploying Broadcom chips seemed to be a positive. But Broadcom could be on the hook for possibly $29Bn in lease obligations should the lessees default. However, as Barron’s points out the upside is asymmetric and by far $750Bn to a possible $29Bn in liability.
If all the data centers envisioned by the agreement materialize, third-party investments could climb as high as three-quarters of a trillion dollars, so Broadcom’s exposure is just a small portion of the chip sales it would have seen in the interim. But still, this got tagged as “circular financing,” lending doubt to the demand story.
Marvell is sniping at Broadcom’s heels as largest client Alphabet added it to its vendor list for networking chips in August. Clearly no one likes to lose exclusivity but I don’t believe that this should dent Broadcom’s pole position in Alphabet’s ecosystem in any way.
Still the valuation looks very promising, and I continue to hold with a plan to accumulate on declines. Broadcom is selling for only 18x FY-Oct 2027 earnings, growing at over 50% annually for the next three years, with its profit margins exceeding Nvidia’s. It is valued at just 10x sales growing over 35% for the next three years. That has a significant margin of safety.
