After recently reading an excellent article on Mellanox (MLNX), the infiniband solutions provider, it got us thinking about the impact on Intel (INTC). The company plans to expand in the HPC infiniband market next year. Previously we had become rather bearish on Intel, and correctly predicted its fall to the $20 range when many analysts were bullish and recommending buying in the $23-24 range. Although headwinds are strong, we had not considered growth in niche markets, including in the GPGPU space using Xeon-Phi. Most of this growth is scheduled to kick in next year. Together, these new niche markets could make up for some of the revenue loss from the shrinking PC market.
Growth in Niche Markets To Offset Sales Decline?
The purchase of Qlogic was significant in highlighting the various relatively small markets where Intel is expanding. From the specification and the capabilities of Mellanox, it does appear they are greatly under threat with the next generation of their products. Intel is focusing the Qlogic platform on 100GB infiniband. Combined with their size and economy of scale, there will be nothing left for Mellanox, which last year had revenues of only $259m and expected to remain below $500 million. However this is still a small amount of revenue for Intel. In fact it is even small for AMD, which is losing market share to Intel. Valuation wise MLNX is far more expensive than AMD.
Looking at the plans, it appears Intel is also going for Nvidia's high margin lunch, while simultaneously going for the low margin entry-mid level discrete graphics with the Haswell integrated graphics engine of the 2013 CPUs. This could explain why Nvidia's (NVDA) share price is falling. Several reviews and experts are predicting Xeon-Phi will fend off and destroy the GPGPU server market, a market currently cornered by the Nvidia Tesla line of products. If Intel is successful it will mean that they are attempting to eat all of Mellanox's lunch, and nearly 30% of Nvidia's revenue. Nvidia Tesla sales are around the $400 million mark. To defend its share from a fully integrated competitor, Nvidia plans to integrate ARM 64 (ARMH) cores into Tesla, which will allow it to become independent of Intel or AMD (AMD) CPUs. However, assuming Intel wins on price and performance, as indicated by semi-accurate, Nvidia could quickly loose market share. The professional GPGPU market may disappear since programming for Intel's solution, of many and multi-core boxes, is far easier than optimizing algorithms using parallel, Nvidia specific, languages like CUDA or standards like OpenCL. While not game changing for Intel, a total of $0.5-1.5 billion in additional near term revenue can offset the loss from ARM based windows tablets. In other words, Intel may be worth more than our previous simulations indicated.
The Valuation Upside
We approximate that Intel can gain between $0.5 billion to $1.5 billion in revenue in these niche markets. Furthermore, with a small rise in mobile penetration, it can gain some revenue there as well. In total we can see Intel loose 1-2% revenue, or up to $1 billion, to ARM based tablets. So according to these estimates, Intel's net revenue situation could remain constant, or grow slightly as the x86 market shrinks. At zero growth, our previous dividend growth simulation indicates a fair market value of $22-25. Hence Intel may be a buy at $20 if you only expect a modest capital gain in the next 12 months.
Downsides outside of CEO retirement
Lack of industry flexibility and pursuit of new opportunities is one of Intel's biggest drawbacks. It was arrogance when the now retiring CEO dismissed the ARM server threat, while its biggest direct competitor, AMD, was joining the ARM bandwagon. Intel currently lacks the kind of flexibility shown by historicaly adaptive companies, willing to change industry and take risks like Nokia. Such corporations are mostly successful long term, but this kind of flexibility is hard to develop. Other issues with Intel is arrogant price setting which dents demand. Haswell appears to be so highly priced that it may be less of a success than expected, at least according to this artcile.
Nevertheless, Intel has a solid foundation as a semiconductor fab, and houses lots of technology. Notwithstanding other articles focusing on the CEO's resignation, I see it as a non issue. At $20, Intel is definitely worth considering if you want to invest in semiconductors and don't expect unrealistic returns. Intel's as safe as they come at these levels.
Disclosure: I am long AMD.
Additional disclosure: May initiate a long on Intel. my AMD position is pure speculative.