2018: A banner year

Intel has a great Q1, but 2019 performance is cloudy

Intel has a long history of generating tremendous amounts of revenue and income. This latest quarter is no exception. Intel has announced record Q1 revenues for this year and they look to continue that trend throughout 2018. AMD released their very positive results yesterday, but their finances are dwarfed by what Intel has brought to market. The company had revenue of $16.1 billion with a net income of $4.5 billion. Compare this to AMD’s $1.625B revenue and $81M net income we see that the massive gulf between these two companies will not be bridged anytime soon with either Intel falling or AMD gaining.

Intel has put its money to good use with a wide variety of products that stretch between the PC market and datacenters. While their low power and ultra-mobile strategies have been scaled back and cancelled in some cases, their core markets are unaffected and they continue to make money hand over fist. The company has always been fundamentally sound in terms of finances and they do not typically spend money recklessly. They continue to feature market leading IPC with their product lines and can address multiple markets with the x86 products they have.

Perhaps the two biggest pieces of news from Intel in the positive column is that they are retaining revenue from the PC market that has been weakening as of late, even with the increased competition from AMD. That group achieved about a 2% growth year on year. The datacenter side is almost matching the PC side in terms of revenue which has long term advantages for the company as marketshare and demand changes are much more stable than the seasonally oriented PC market. Not to mention that it is much harder for competitors to take away significant marketshare from the datacenter.

These are certainly positive signs from the company that they will continue along their path throughout 2018 without major bumps or disruptions. Their 14nm process and derivatives are still performing well and yields are apparently excellent. It has been an excellent moneymaker for them as it has only been relatively recently that the 3rd party foundries have process nodes available that match Intel in terms of transistor performance and dimensions.

Having put those positive things out there, we also know that not all is well at Intel. The elephant in the room has been the 10nm process that Intel has been developing for years now. It is still not up to spec in terms of consistency and yields. Intel claims to be shipping limited amounts of product on it, but those truly are limited and I doubt the average user can get their hands on it. While Samsung, GLOBALFOUNDRIES, and TSMC are busily producing their 14/16nm products to their customers, they are also moving ahead with advanced process nodes that will not only match Intel’s 10nm, but also potentially surpass it.

Intel’s only major competitor in the x86 business has recently released their 12nm based Zen+ processors. While there are no real dimensional improvements from GF’s 14nm to 12nm processes, what has improved is transistor switching speed, power characteristics, and manufacturability. Faster, more efficient parts that have improved yields going along with a mild redesign/optimization has made the new Ryzen 2000 series of CPUs a true competitor for the latest Intel CPUs on the desktop and mobile markets.

10nm for Intel has been a disaster. The company has essentially lost its process advantage over the competition. They are now on what is essentially an even playing field when it comes to manufacturing products throughout the rest of this year. We have seen AMD’s desktop lineup (or so we have been lead to believe) and know what is coming up in terms of the mobile space. The products, from top to bottom, are very competitive between Intel and AMD. Previously Intel could expect several months of near parity in terms of process performance before they release the next generation of product on a new process node and again achieve a tremendous advantage over their competition in terms of performance and die size. The competition is then forced to lower prices to make their products more attractive to buyers all the while Intel reaps the benefits of increased performance, smaller die sizes, and greater efficiency. Now Intel does not have that luxury for anytime in the near future.

During the conference call Intel announced that it would not be reaching volume manufacturing on 10nm until 2019. When pressed further it was mentioned that it would likely be in 2H 2019. This is unknown territory for Intel. In the past when they have encountered real competition from AMD, they still had a good 18+ months of process superiority, sometimes even longer than that. Now Intel needs to rely on their improved 14nm processes that actually give up some die space in return for better power and switching properties.

The foundries are currently shipping some power efficient 10nm parts and are working on getting 7nm out the door. TSMC is in full production with 7nm which can be compared quite favorably to Intel’s 10nm in terms of dimensions and transistor performance. This production is still very new, but parts are flowing out of the fab. TSMC will most likely be relegating much of the line space to smaller customers with smaller ASICs, but will be transitioning to larger designs in greater volume. We already know that AMD has received 7nm Vega based silicon back, but that part is not meant for general consumer usage. AMD apparently is aiming for more compute oriented markets with 7nm Vega and will release NAVI based parts early next year on 7nm as well.

 

« PreviousNext »