IMO ADRC could be key to Intels ability to increase significant amounts of new data that will drive server sales, this is the No 1 area for Intels future revenue sustainability and growth.
See article below
Intels marketing is everything they are doing as directly or in directly benefitting their datacenter efforts.
It is crucial that this is successful and let's hope we can play a big part in this going forward, seeing that we are apart of this field with our Vital Xense exposure, Intel would of sat up and taken a lot of notice of what we have to help in the sensor and data center market.
IMO If we are what they are looking for to assist them with continued dominance in this area then none of us could guess what $value ADRC or Xped could be to them. All I know is it would be massive!
20, 2016, 5:34 P.M. ET
Intel’s Future Hangs on Data Center, and Cracks Will Appear, Says Bernstein
By Tiernan Ray Intel (INTC) faces “cracks” in its dominance of the server chip market, write Stacy Rasgon and colleagues at Bernstein research today, in the third installment of a research series they are calling “The Future of Processor Architecture.”
The first installment, related here two weeks ago, suggested Intel and arch-rival ARM Holdings(ARMH) would maintain dominance with their respective instruction-set architectures for quite some time, though there is need for some new approaches to computing.
In today’s installment, Rasgon, who has an Underperform rating on Intel shares, and a $26 price targets, leads the team in predicting that by the year 2020, you should start to see cracks appearing in Intel’s server dominance, as a result of a number of factors.
They include the move to cloud computing displacing plain-old enterprise IT server sales, and the encroachment of competition from ARM-based offerings and other alternatives, such as IBM’s (IBM) “Power” processor family:
The trends we are noting now, albeit early days, suggest at least some cracks to Intel’s monopoly position may be forming. The company is probably reasonably safe until ~2020 (when their process technology advantage turns into parity) but 2020 isn’t really that far away (so true pressure appears likely within 3-4 years); in the meantime efforts from the industry to find and fund alternatives is growing significantly, and newsflow and commentary (which has begun to increase) is likely to build even more over time. And new competitive options likely don’t need to achieve significant amounts of market share to cause damage; rather it may be enough if alternative solutions are credible enough to crack Intel’s pricing umbrella.
Rasgon argues that the “long-term structural case on Intel (bull or bear) hinges on the long-term sustainability of the data center group. The business unit has come to represent more and more of revenue, as he shows in this infographic:
But more than that, writes Rasgon, Intel has built all its lines of business around supporting data center:
However, what is new is the role of Intel’s other businesses around DCG; in fact Intel is now marketing everything else they are doing as either directly or indirectly benefitting their datacenter efforts. According to the company, while acknowledging that PCs may be in decline they still fill the fabs, and generate necessary IP (both process and architectural) that supports datacenter. The growth of IoT will (supposedly) generate significant amounts of new data that will drive server sales. Memory efforts revolve around datacenter SSDs and new offerings like 3D X-point. And of course, smartphones also require cloud services, and new networking efforts require a strong radio suite and work on future 5G technologies. Therefore, the long-term structural case on Intel (bull or bear) hinges on the long-term sustainability of DCG.
Among the worrisome signs emerging, Rasgon believes cloud computing is cannibalizing the traditional IT server-chip sales for Intel:
Intel has not been immune. Over the last few years they have met their “15%” growth target only once (Exhibit 16), with enterprise typically the disappointing factor. The company has suggested that the macro has been to blame (and this of course might be part of it). However, cloud and hyperscale growth has remained robust for the company, and we can’t help but wonder if the two trends are in fact coupled to some degree. While one can debate “how deflationary” the move to cloud is (and even among the authors we do not have full agreement), we believe we are arguing over matters of degree; cloud HAS to be having an impact as that is, of course, the entire point of the exercise [...] Overall we believe at best that enterprise server shipments are likely to be flattish; the primary author of today’s piece feels that a declining market is easily within the realm of possibility.
One example of potential threats is the incremental performance imprvoement from the ARM and Power alternatives, notes Rasgon, even though they “as yet mostly remain uncompetitive on performance/Watt and TCO”:
ARM still has a good way to go, but the trend does appear to be in the correct direction; at a minimum we feel comfortable stating that the gap to x86 appears to be narrowing rather than widening [...] IBM has also entered the fray by licensing and opening up their Power architecture through the OpenPower foundation, fur use in servers. OpenPower was formed in Dec 2013 with Google, Mellanox, Tyan, and Nvidia as founding members. And they have also set similarly aggressive targets for potential share (10- 20% of hyperscale and high-performance computing markets). The Power architecture has been reasonably competitive x86 in the word of high performance computing over the last decade or so, holding on to ~10% share in the top-500 supercomputer statistics. However, the performance per watt has been much more problematic so far, limiting appeal for high- volume hyperscale deployments. But IBM has now begun to talk about the next generation (Power9), on the roadmap for 2017. The chip will have 24 cores (up from 12), support for PCI-Express 4.0, and the 2nd generation of their Coherent Accelerator Processor Interface (CAPI) which allows direct attach of accelerators, as well as Nvidia’s NVlink2.0. The chip will also finally be manufactured on leading-edge processes (14nm at GlobalFoundries for IBM’s own products, and through other OpenPower partners on 10nm/7nm in the 2018-2020 timeframe) that should improve power issues markedly. To that end, we note that Google is now working with Rackspace to co-develop Power9 server platforms (called Zaius) which they are planning to make available through the open compute project if accepted. Google has suggested that they would switch to Power over x86 for a 20% improvement to total cost of ownership, and has already made their infrastructure Power-ready (so their software developers can enable it with a simple flag modification in the configuration).
Intel shares today closed up 41 cents, or 1.3%, at $32.17.
