With the Open Compute Summit just around the corner, the focus will be on web giants and their datacenter infrastructure. Web Giants like Amazon, Facebook and Google have huge datacenter capacity to power their internet services. When these companies were early in growth phases, they partnered closely with the large global OEMs (Dell, Hewlett-Packard, and IBM) to supply their servers. The OEMs recognized the explosive growth potential from the web giants, and began to invest heavily in design and support expertise to service the unique needs of these customers. However, just as the OEMs like Dell, Hewlett-Packard and IBM began to reap the benefits of their investments, they were blindsided by a new form of competitor. In order to save money, some of the largest web giants like Google decided to specify and buy servers directly from the companies that built servers on the OEMs behalf: Taiwanese ODMs like Quanta and Wistron. Quanta was the first of the ODMs to aggressively move their business focus away from servicing OEMs like Dell and Hewlett-Packard, to competing directly against them. So far, this shift in strategy has worked well for Quanta; their server shipments have skyrocketed over the last few years and direct sales to datacenter customers are now estimated to represent ~85% of Quanta’s overall server business. Quanta’s strong penetration with the largest web giants allows them to be well positioned to continue to expand their server business. Notable customer wins include Facebook, Rackspace, Amazon, Korea Telecom and many others who are not public. But a business built on servicing web giants comes with its own set of challenges. Quanta must once again evolve its business strategy if they want to continue to grow as I discuss below. If you are looking for a deep dive, you can see our detailed paper here.
While server purchases by the web giants continue to grow faster than the market rate, this segment still makes up less than 20% of the total server market. The web giants are fickle and have little vendor loyalty; their desire for choice and low cost means they rarely show favoritism to any one supplier. While high-volume orders are the norm for these customers, their buying patterns are choppy, dependent on new datacenter rollouts or refresh cycles. Also, the competition continues to increase from OEMs doubling down to regain lost share and other ODMs building up direct business models. All of these factors will lead to inconsistent financial results for Quanta, if their entire server business model is focused on sales to the web giants.
In order to build a model of sustainable growth, Quanta should expand into new segments within the server space. There are a growing number of large and medium scale cloud service providers that do not buy direct from ODMs today, but have many of the same technology requirements and workload characteristics as the web giants. Also, while the enterprise server market generally prefers to buy from the global server OEMs, there are pockets of this market that are not brand-sensitive and could benefit from hyperscale-inspired server design expertise where Quanta excels.
For both the cloud service provider and enterprise server segments, Quanta will require an entirely different channel, customer support model, and product strategy to effectively compete. In 2012, Quanta took a step in the right direction with the formation of their US subsidiary, Quanta QCT. However, Quanta must beef up their service and support infrastructure, channel partner program, software expertise and unique IP investments if they want to continue to grow with their existing customer base and expand into new segments.
For a deeper dive on what Quanta needs to do to scale their server business, I’ve published a paper here.