Qualcomm wants to buy Intel. Would that be enough to overtake Nvidia?

This story originally appeared in The Technology Letter and is republished here with permission. Shares of Intel jumped 3% Friday as The Wall Street Journal’s Lauren Thomas, Laura Cooper and Asa Fitch reported that Qualcomm has approached Intel about acquiring it for perhaps as much as ninety billion dollars, citing multiple unnamed sources. The “massive” deal, as the authors put it, is financially daunting, as Qualcomm has just thirteen billion in cash and equivalents on its balance sheet against thirteen billion of long-term debt. Even a mostly-stock exchange would require some large debt raise. Intel, moreover, already has nineteen billion dollars of net long-term debt. The deal is much larger than Qualcomm’s attempted acquisition of NXP Semiconductor in 2016 for $38 billion. That was initiated at a time when Qualcomm had an enormous amount of cash trapped overseas prior to 2017’s Tax Cuts and Jobs Act that let Qualcomm bring it back to the U.S. (Qualcomm ultimately spent twenty-two billion of repatriated money on buybacks when the NXP deal was canceled.) The profile of the combination, moreover, would be financially unattractive, as Intel has a 35% gross profit margin to Qualcomm’s 76%. And Qualcomm’s pre-tax operating margin is near 30%, whereas Intel’s is break-even on an adjusted basis, but actually negative by 15% when all costs are factored in. Intel would immediately dilute Qualcomm’s profit profile. Assuming, however, Qualcomm could pull it off financially, what are the synergies, meaning, what is to be gained financially and strategically from such a move? What Qualcomm needs most is diversification, as it is still considered by investors to be a cell-phone chip maker. It still gets about 70% of its chip revenue quarterly from mobile, even though Qualcomm has for several years been selling chips into the “Internet of Things” market and the automotive market. Buying Intel would immediately make the company the top PC microprocessor and server processor vendor, which would certainly change the profile of the company. Intel needs to regain its manufacturing prowess. The positive announcements from the company earlier this week included references to the company having “momentum” in getting its newest chip technology, “18A,” out the door next year. It’s hard to know what that momentum really means, and whether it’s going to restore Intel to greatness. It’s conceivable Intel needs a helping hand. Qualcomm, which has no factories of its own, offers, ostensibly, nothing to help Intel in that regard. While it is possible Qualcomm’s higher-margin products could give Intel a much needed financial lifeline that would advance that momentum, I’m not sure throwing money at the problem is the solution. More cooks in the kitchen is not going to streamline the intricate challenge of enhancing Intel’s factory operations. Moreover, neither company has, individually or collectively, the solution to their mutual problem, Nvidia. While both Intel and Qualcomm have ample artificial intelligence resources, neither has been able to put a dent in Nvidia’s market dominance despite years of trying. It’s possible that Qualcomm’s CEO, Cristiano Amon, sees something deeper. One possibility is that he sees a grand alliance, of sorts, of Intel’s server chips and Qualcomm’s mobile chips that might somehow box Nvidia out of the AI market as AI moves to mobile phones. The phenomenon is called in industry “AI at the edge,” where the server and the handset intelligently apportion the work between them to make the most efficient processing of AI where it makes sense for the available computing resources. One does stuff that’s private and low-resource on the handset, and the really heavy-lifting sorts of AI, in the cloud. Nvidia doesn’t have a mobile chip offering, so that line of argument has a certain appeal. Moreover, Intel’s substantial assets in what’s called chip “packaging,” the ability to combine multiple chips into one giant chip, could allow for new kinds of mobile products beyond what Qualcomm currently builds with the help of Taiwan Semiconductor Manufacturing. A grand alliance to defeat Nvidia still faces the problem that Intel’s “x86” chip architecture, which dominates PCs and servers, is nowhere in the handset business. It’s not clear the combined efforts of the two companies could make it relevant in that regard, with or without lots of AI capabilities. You can imagine lots of other, less-glamorous possibilities. Intel is going to billions of dollars of “CHIPS Act” funding to build U.S. factories, and perhaps Amon sees the possibility there of boosting Qualcomm’s profile by making it an American-made company. It’s also possible that Amon simply sees a stock trading too cheaply. Intel shares fetch two times revenue and nineteen times next year’s expected earnings per share, among the lowest multiples in the industry. For the moment, Qualcomm investors don’t see too much