Last week we commented on analyst Ming-Chi Kuo’s report that Apple’s modem had failed some internal hurdle. That is big news if it is accurate. Here we want to look at how sell-side analysts collect data through the lens of this latest report.
Sell-side analysts face a dilemma. They need to produce ‘actionable’ information, data that their buy-side clients can make trades on. And the field is competitive, so this information has to be differentiated, not something their clients can hear from ten other analysts or read in the Journal.
At the same time, analysts need to do this without violating any laws or ethical principles. Many countries have rules like the US Reg FD, which require that when companies disclose material information they Fully Disclose that information to everyone at the same time. This eliminated a common practice in the previous century where companies would win over analysts by ‘whispering’ numbers to them ahead of time. And for the most part, most analysts today abide by these rules. We certainly have no reason to think Kuo violated anything in obtaining his information.
So analysts need to come up with interesting, individual information. One solution is to obtain some sort of proprietary data set. Decades ago, there was the story that one analyst’s father-in-law ran operations at TSMC, and so that analyst would show up at client meetings and pull out a print-out of the foundry’s customer orders for the coming months. That is not what we are talking about. Instead, we are looking at more creative, less NDA-violating methods. For instance, one analyst years ago purchased satellite photos showing the parking lots of 100 shopping malls across the US to gauge foot traffic during the Q4 shopping season. Another hedge fund reportedly used satellite photos to measure the shadows of collapsible oil storage tanks and thus measure oil inventory levels. Many analysts conduct some form of consumer survey to gauge all sorts of topics. There is a whole industry now around so-called “Alt Data”. The downfall of this approach is that information spreads quickly, and so today’s proprietary data sets are tomorrow’s “everyone does this”.
Probably the most common form of information gathering is to cultivate human contacts. Companies like to talk about themselves, the marketing teams always have something to say. Speak to all of the companies in a sector and good analysts can sort through the biases to get a real sense of a market. The drawback of this approach is that marketing teams are by their nature going to spin things optimistically, leading to industries where everyone shares 150% of the market.
So analysts need to speak to people beyond the marketing and investor relations teams. Over time, good analysts can develop relationships across the executive teams of the companies they care about. This does not mean a throwback to the 1990’s, in our 20+ years in the markets, we have never heard an executive team divulge truly material information, and only in one instance were we ever asked that kind of question by an investor (and no, we did not provide it). Instead, analysts need to develop a deep understanding of their covered industries with an appreciation for all the nuances. In technology, this means diving deeper into the technology stack and being able to compare across companies.
But why would company management talk to analysts about these things? Ultimately, they are the ones who risk sanction for Reg FD violations. Good analysts need to first obviously build trust with management teams, and a good way to do this is to never ask pointed questions about the current quarter. Instead ask them about topics they want to talk about, most semiconductor CEOs love to talk about their products. Ask these questions enough times and analysts can get a deeper understanding of the underlying portfolio.
The really best analysts make themselves industry experts, so that management teams seek them out for advice and insight. This makes for much more balanced relationships, with a two-way exchange of information. For instance, one analyst we know made use of his Mandarin-speaking associate to gather information about Chinese companies. He then provided that data to the CEO of one of his top companies who were fighting fiercely in China. The CEO came to greatly value his monthly then weekly conversations with the analyst, getting real competitive insight from them. The analyst parlayed that into a multi-year relationship and eventually an executive role at the company.
We imagine that Kuo has done something similar here. He is widely regarded as knowledgeable about the Apple supply chain, and so many executives will want to speak to him. And there is a peculiarity of Apple that points in this direction. For all of Apple’s suppliers, the fruit company is both incredibly important and also very hard to work with. In particular, Apple is terrible about forecasting, or at least the forecasts it provides to its vendors are not reliable. The Apple operations team wants to make sure they have sufficient supply of parts, and so they regularly overestimate their actual demand. The vendor cannot complain about this, but they really do not want to end up with a mountain of unused inventory that Apple will never buy. So they are always searching for validation of the models they get from Apple. This provides an entry for analysts, who can collate data from multiple sources to provide more refined estimates. For the most part, companies can participate in this without violating their ironclad NDAs with Apple.
So how do we get down to a specific part like an Apple modem? Here we are just going to have to speculate. Our best guess is that since the modem is such an important part of a phone, many chip companies downstream of it depend on it for their own design wins. At this stage, there are many companies vying for the RF, passive and memory sockets in the iPhone. And if those companies suddenly turn pessimistic about 2023 it provides a signal. Again, the company does not have to say “we were on the Apple modem reference design and that project has stopped”. Instead, they can say they had pinned their hopes on one platform, but are less certain about that now and so have to change their outlook. It can even be as simple as finding out that certain executives are making their first trips to San Diego in years, because now they have to court Qualcomm, and find a way into their reference designs. Again, we are speculating here, but we can think of a lot of different ways that a good analyst can suss this data out.