The tax charge cannot reach the operating line, and that is the part of the bull case that survives contact with it. A noncash charge tied to an intellectual property transfer sits below operating income [3], so it inflates the trailing net income comparison [2] while leaving the GAAP operating margin sequence untouched [8]. Add the $4.5bn back to the year-ago trailing window and net income growth falls from 127% to roughly 68%, which against 32% revenue growth is a spread of about 2.1 times rather than 3.9 [1]. Broadcom says as much itself, in fewer words [4].
The more interesting flattery is the kind nobody had to book. Of the ten-point move in GAAP operating margin over the year, about four points came from deal amortization holding at roughly $2bn a quarter [9] while the denominator grew, so a charge that took more than 13% of revenue a year ago took about 9% in fiscal Q2 [10][6]. At the guided $29.4bn it takes 6.8% [2], handing over two more points before anyone is hired or anything is priced. That lands in reported profit, certainly, and it is still not cost discipline, because a percentage walking toward zero eventually stops paying. The 67% margin in the guide is measured after stripping stock compensation and that same amortization [20], so none of the denominator effect shows up in it, which makes it the cleaner of the two tests.
The number Wednesday settles, then, is the expense base. Hold it at the $4.6bn of the quarter just reported and operating expenses drop from 20.7% of revenue to 15.6% [3] on a sequential step-up of about $7bn [22]. R&D grew 11% against 48% revenue growth [7], and the $16bn of AI revenue Hock Tan expects in the quarter, up more than 200% [13], is not obviously a business that ships on last year's headcount. The software segment turned 9% revenue growth into 13% profit growth because its costs fell [17], which is the same trick at one-tenth the scale.
Sixty times trailing earnings on $29.3bn of trailing net income is about $1.76tn of market value [4], for a stock a quarter below its 52-week high of $495 [14]. Hold 68% profit growth for a year and the 60 becomes 36 [5]. That division is the entire bull case, and it is doing a lot of work.
This is probably too neat, but the bear case worth holding is not the AI mix that then-CFO Kirsten Spears flagged on the June 3 call [12]; it is that the two cheapest sources of margin, amortization shrinking against a bigger number and an SG&A line that fell outright [7], are close to spent, which leaves gross margin, already guided down [12], to carry the next leg. I would be wrong if operating expenses print near flat, the non-GAAP operating margin lands at 67% [11], and semiconductor margin holds the 62% it reached in fiscal Q2 [16].
