AOI’s major hyperscale data center customers are undoubtedly sitting back and enjoying the vendor’s frantic activity in making heroic efforts to cut its manufacturing costs to the bone. Here is a company that announced substantially more vertical integration in real time with its last quarterly earnings conference report. Fewer than two weeks later, there was a major powwow with various investors, and slides were released publicly, which gave further details on the dramatic extent of its impressive expenditures. Naturally, regular buyers of its gear, including Microsoft and Amazon, will be the principal beneficiaries of such actions, as they will certainly be in a position to eventually demand much lower prices from AOI for intra-data center gear, which will possibly set a new bar for the other players in the space as well. They will be able to confidently insist on such new pricing because they have undoubtedly brought in the necessary expertise in-house, formerly employed at large system houses to comprehensively understand any optical component suppliers’ costs and margins, which will be further bolstered by the data offered in AOI’s new slides.
Even if the naysayers on Amazon building its own 100G gear are correct, or even if the gigantic retailer changes its mind, the threat alone has certainly been instrumental in AOI’s frenzied and desperate acts. As we have discussed in the past, these are low-end, highly undifferentiated 100G transceivers, with multiple vendors offering them, including just the intimidation factor alone from Intel, along with the various imaginative ways of these buyers to force prices down even further, and so it is inevitable that margins will ultimately suffer at AOI, despite its best efforts.
Unfortunately for AOI, factors it is advocating in its favor such as bandwidth growth, manufacturing differentiation, greater insourcing, R&D skills, historic growth levels, strong balance sheet, CAPEX growth, and 100G cost reductions will not help it for long on CWDM4 and PSM4. Rather than higher end speeds, we remain more bullish on the company for its 4WDM-10 MSA offering for short-term niche opportunities, and 40G in the much longer term. Nonetheless, we believe that maintaining gross margins of a lengthy period of time in the low 40s is a pipedream.
We partially pointed out previously that AOI’s biggest cost advantage overall is the large percentage of employees (close to 90 percent) in Asia. According to an investor who listened to the most recent webcast, the "headcount in the Sugarland, TX, fab will only rise 50%" during this calendar year. (However, in our opinion, his outlook in general on the company should be taken with a grain of salt, partly because of his lack of pertinent information, especially on the competitive landscape.)
In addition, our earlier conclusion that the degree to which vertical integration gives the supplier an edge being overstated has been backed by its currently large investments in its manufacturing plants to engage in less outsourcing. We also think AOI calling the data center space non-cyclical is at least a bit of a stretch in that spending slowdowns can occur for a variety of reasons including less than expected economic performance by the company as a whole or by taking a break to test the waters on pricing of alternative solutions.
As fibeReality discussed a few weeks ago: “In our opinion, another big problem for AOI is categorically asserting during the [last earnings] call "that in the long term, and in the short term, we don't see any cost advantage to [the MACOM-Fabrinet alliance] model." AOI has to be careful in talking about its large customer.
[written by Mark Lutkowitz]