This is not investment advice. The author has no position in any of the stocks mentioned. Wccftech.com has a disclosure and ethics policy.

A few days back, Intel posted its earnings result for the third quarter of 2017 beating not only on revenue but on earnings as well. The company reported a (non-GAAP) EPS of $1.01 a share on back of $16.15 Billion in revenue. This was easily greater than analysts expectation of 80 cents and $15.73 Million respectively (as provided by Thomson Reuters). Not surprisingly, the company's stock price gained a solid 6.6% to date following the earnings beat and momentum remains generally bullish. This analysis will aim to dig deep into the earnings posted by Intel and dissect the material changes in the company's fundamentals that have happened this quarter.

Intel NASDAQ:INTC Q3 2017 Earnings Slide Deck

2 of 9

The company reported net income of $4.52 billion which is a 34 percent increase year-over-year. In the year-ago quarter, Intel reported adjusted earnings of 80 cents a share on $15.78 billion in revenue. Intel said it now expects full-year earnings of $3.25 per share on revenue of $62 billion. That's up from its previous forecast for earnings of $3.00 per share on $61.3 billion in revenue. The company has instead highlighted its opportunities in new markets such as artificial intelligence and cloud infrastructure.

"We executed well in the third quarter with strong results across the business, and we’re on track to a record year,” said Brian Krzanich, Intel CEO. “I’m excited about our progress and our future. Intel’s product line-up is the strongest it has ever been with more innovation on the way for artificial intelligence, autonomous driving and more.”

“In the third quarter, we delivered record earnings, exceeded our EPS expectations, and increased our profit expectations for the full year,” said Bob Swan, Intel CFO. “We feel great about Intel's transformation and where we are nine months into our three year plan.”

Intel Q3 2017 Business Unit Trends

All in all this has been a good quarter for Intel with heavy diversification away from the PC market and into AI and autonomous driving sectors. The company has also revised its outlook for the full year as well as given the following outlook for Q4 2017:

One small step for autonomous driving, one giant leap for Intel

Goldman Sachs projected the market for advanced driver assistance systems and autonomous vehicles would grow from about $3 billion in 2015 to $96 billion in 2025 and $290 billion in 2035. So this is one of the most important markets for Intel to diversify into and will definitely be the big leagues of the future. Here was the problem though, Intel's recent Mobileye acquisition had set it on a path that was tangent to what we can call the 'NVIDIA approach'. While the former uses a serial approach to dealing with autonomous driving, NVIDIA employed GPGPUs to bruteforce the problem - an entirely parallel approach. This meant that Intel was exposing itself to serious risk if the market decided that the serial approach was not the right way to go after all.

To give you an example, Mobileye plans to introduce the EyeQ4 chip by 2018 – a truly worthy upgrade (to the EyeQ3) with 14 computing cores, out of which 10 are accelerators. You will have your usual 4 cores as well as accelerators of three different kinds: PMA (Programmable Macro Array), VMP (Vector Microcode Processors) and MPCs (Multithreaded Processing Clusters). The actual count will be 4 Cores, 6 VMPs, 2 PMAs and 2 MPCs for the EyeQ4. This will result in approximately 1.26 Billion MAC/s or 2.5 Tflops. All of this will be done while staying within a TDP Target of less than 5 Watts.

The EyeQ4 processor will have (in roughly comparable terms) a performance rating of 1.5-2.5 TFLOPs where as NVIDIA (NASDAQ:NVDA 201.86 0.00%) is equipping Tesla cars with the Drive PX 2 set which can churn out 24 TFLOPs. I have seen some arguments which point out that the cuDNN equivalent utilization is roughly 0.3 for GPUs and even if we assume that Tesla has made no in-house optimizations, you are still looking at a bare bones performance of 8 TFLOPs. So NVIDIA’s worst case scenario is 3.2x faster than the best Mobileye chip (which hasn’t even fully landed yet) right now. NVIDIA has also released the Pegasus board which is capable of 320 DL TOPs, orders of magnitude faster than the EyeQ4 approach - and capable of Autonomous Level 5.

This is where the Nervana acquisition comes in and why it is so very important for Intel. By acquiring Nervana and rolling out a highly parallel (dare I call it a GPGPU based?) approach to AI, the company has now diversified itself into the only other option in the AI industry - fortifying itself against a binary loss condition. The Nervana 'Neural network Processor' uses a prallel approach to computing and is built pretty much like a normal GPU. It has 32 GB of HBM2 memory dedicated in 4 different 8 GB HBM2 stacks, all of which is connected to 12 processing clusters which contain further cores (the exact count is unknown at this point). An interposer has been used to full effect and Intel's homegrown interconnect seals the deal.

Nervana itself might or might not make it big in the autonomous driving segment with the head start that NVIDIA has but Intel has made sure that it has a fighter in the AI department and its toes in the DNN training industry. We cannot however speculate on the exact nature of the impact this chip might make since for 1) Intel has not disclosed the full specifications of the Nervana chip and 2) no benhcmarks have been revealed so far to compare it against a homegrown GPGPU network running CuDNN.

Taking serious heat from AMD on the computing side and other potential caveats

On the other hand however, Intel actually is entering into murky waters for what appears to be the first time in years. Shortly after the launch of Skylake, it came under heavy fire from AMD with the launch of its Zen series of products. The result was that Intel's Skylake parts were very quickly dethroned with AMD providing a much higher core count (yes, with comparable performance per core) at a cheaper price. In fact, its Threadripper parts beat Intel to market and we have heard preliminary reports that Intel lost CPU market share to AMD because of this disruption.

The company was very quickly able to respond with the Core-X series and Coffee Lake - which put the ball back into AMD's court but not before significant damage had been done. In fact even now, Intel's Coffee Lake parts are facing supply issues and the HEDT Core-X parts are priced too high for the mainstream consumer. The reason for this is that Intel has always built monolithic dies which are very expensive but AMD has gone with the MCM approach which allows them to build relatively cheaper CPUs.

Since computing is one of the most important business of Intel, this remains a rather serious concern for Intel. AMD has already forced it to release the 18-core parts (we have it on good authority, and slides, that the company was only planning to release 10-core parts before the AMD Threadripper surprise) and wrong footed it in terms of pricing premium. Intel parts are still completely superior to AMD parts clock for clock and core for core, but the premium can go as high as 80% for as little as a 20% performance delta over similar AMD parts.

It will need to be very careful with navigating this particular market in the coming quarters, or risk loosing market share to AMD.