Interface IP year 2015: Winners and Losers

The global Interface IP market is still growing in 2015, no doubt about it. It’s interesting to zoom in the various protocols to check their respective behavior. Which protocol generates an IP business growing more than the average market? Which protocol generates a disappointing IP business? In other words, which are the winners and the losers in 2015 when we review USB 2, USB 3, PCI Express, (LP)DDRn, HDMI, MHL, DisplayPort, MIPI, Ethernet or the VHS SerDes?

Universal Serial Bus USB is ubiquitous, but if you want to understand the market dynamics of USB IP segment, you have to look at USB 2 and USB 3 separately. Unlike with PCIe, certain applications will integrate USB 2 forever and never upgrade to USB 3. Because there is no need for higher bandwidth as provided by USB 3 justifying to pay the over-cost linked with USB 3 IP. Nevertheless, for the first time in 2014, USB 2 IP market has started to decline, when USB 3 IP market was strongly growing. If you take a look at the picture on the top, you can see the cross-point between USB 2 and USB 3 IP business in $ value happening in 2014. Since the end of 2013, USB 3 IP business generates more value than USB 2 and the trend is confirmed and amplified until 2020. USB 2 IP business is slightly declining when USB 3 IP is still growing, and this growth is strong. Now, if we look at USB IP market in term of design starts, there are still more ASIC/ASSP integrating USB 2 IP than USB 3. In 2015, our evaluation is slightly less than 400 USB 2 design starts and about 300 USB 3 design starts for a total of almost 700 USB design starts. Yes, USB is ubiquitous! USB IP segment is a clear winner, generating about $100 million in 2014.

PCI ExpressThe year 2015 looks disappointing as PCIe 4.0 (running at 16 Gbps per lane) has not yet been widely deployed. But the PCIe IP business has grown with a healthy 10% in 2014, after a strong 15% in 2013 and we expect this growth to be confirmed in 2015. If we consider that the sales of more expansive IP licenses (gen-4 Controller and PHY) will only start to climb in 2016, generating upsides sales, PCIe IP segment is very healthy and expected to grow again. If PCI Express protocol is not as ubiquitous as USB, PCIe is clearly pervasive and is used in many applications, many market segments and not only in PC, Data Center or Networking. We have seen a good example of this pervasion in the smartphone segment. PCIe protocol has been initially introduced to interface with WiFi ASSP in mobile (M-PCIe or MIPI M-PHY + PCIe Controller) but we have discovered in October that Apple has also used the PCIe protocol via NVM Express, to interface with Nand Flash memory (and not MIPI UFS) in the last iPhone generation. The effect on iPhone 6s and 6s Plus performances has been dramatic, as benchmark results for SSD sequential Read and Write are 2x better than the competition! Yes PCIe protocol is a winner, with IP sales reaching 80% of USB IP sales in 2014.

DDRn and LPDDRn memory controller This IP segment is growing continuously since 2010 (see graphic) and this segment is probably the most attractive for IP vendors. Two remarks: ARM has exited the DDRn PHY IP market, the number of competitors (ranked in the “others” category) is pretty high. The market size was in 2014 slightly higher than PCIe. The future of the memory controller IP market is interesting. We know that DDR4 will be the last DDRn protocol (may not be true for LPDDRn). This means that new protocols will emerge. We know already two of these, Hybrid Memory Cube (HMC) and High Bandwidth Memory (HBM), each addressing only a part of the market. We expect these new protocols to co-exist with DDR4 and LPDDR4 (maybe LPDDR5 later) to create a complex memory controller IP market. Not the easiest to handle for a mainstream IP vendor, but niche oriented vendor friendly! Will this memory controller IP market continue to grow? The answer is clearly yes, as ASSP vendors tend to use new protocol generation (DDR3 is still widely used in the mainstream) and will target more advanced technology node, leading to more expansive memory controller PHY IP. A winner IP segment again but IP vendors will have to brainstorm when building a winning port-folio, including emerging protocols.

Display: HDMI, MHL, DisplayPort and derivatives HDMI protocol penetration in HDTV, high end PC and smartphones has reached a maximum. We expect MHL to replace HDMI in smartphones, both protocols being issued by the same entity (HDMI LLC). DisplayPort penetration in notebook is very strong, as well as in display devices, projectors or cables, but null in smartphone and very weak in HDTV. In fact, IP vendors tend to sell combo product: HDMI (controller + PHY) + DP (Controller + PHY) combo IP. HDMI is the only interface IP segment generating (per port) royalties generating per year revenue flow in the $50 million range to HDMI LLC (Silicon Image capturing the largest share). A few IP vendors are selling HDMI or combo IP, but Silicon Image is not active on the mainstream market. This display IP market segment represent less than half of the above mentioned segments, even if the adoption has been very successful for HDMI/MHL in HDTV, PC and smartphone and good for DisplayPort in notebook PC and PC peripherals.

MIPI specifications The complete list of MIPI specifications is very long (!) and only a few specifications have been widely adopted in the wireless mobile segment. These are Display (DSI), Camera (CSI-2) and Universal Flash Specification (UFS). The MIPI Alliance has built partnerships with USB-IF, PCI-SIG and JEDEC for (resp.) HSIC/SSIC, M-PCIe and UFS and these new specifications are generating upside MIPI IP sales. New MIPI specifications like SoundWire or I3C are generating a real interest in the industry, but these are low complexity (low price) functions, at least when compared with high bandwidth interfaces. We think there is good growth potential for MIPI. One growth source is with the emerging (Asian) chip maker targeting wireless mobile. Another source is linked with emerging systems like wearable and with entertainment in automotive. The MIPI IP segment is growing year after year but stays at an average level, higher than Display but not at the level of USB, PCIe or DDRn. Is MIPI successful? Yes, if we consider the high adoption for CSI, DSI or UFS. Will MIPI IP sales be comparable to USB or PCIe IP sales? Probably not…

Interface IP penetration To build an accurate forecast of the interface IP market, we have to take into account various parameters. At first the SoC paradox: SoC design starts count is growing, even if global ASIC/ASSP design starts is declining. The SoC integrator behavior is to externalize IP sourcing, even if this is a long term trend. There are two reasons pushing to buy an IP instead of developing it internally. The first reason is to focus on core competencies and externalize the other functions, but when the IP is already available in another product group, the chip maker may prefer to use internal IP. The other reason is linked with IP complexity. If we review the high bandwidth protocols (USB, PCIe, DDRn, Ethernet) we see that a new release with double data rate is generated every 3 to 4 years. A design team able to develop certain IP running at 5 Gbps in the past may not have the right competencies to design the new release running at 16, 20, 28 or even 56 Gbps. Buying a commercial IP (to an IP vendor) becomes the safest solution. That’s why the Interface IP market has grown fast during 2010-2015 and will continue to grow up to 2020, probably with a lower CAGR than before.

IPnest forecast is below:

Access the Table of Content of the “Interface IP Survey & Forecast 2010-2020” here.

Eric Esteve from IPNEST – IPNEST is the #1 for the IP dedicated surveys, enjoying this long customer list:

Synopsys, (US) Cadence, (US) Rambus, (US) Arasan, (US) Denali, (US) now Cadence Snowbush, (Canada) now Semtech MoSys, (US) Cast, (US) eSilicon, (US) True Circuits, (US) NW Logic, (US) Analog Bits, (US) Open Silicon,(US) Texas Instruments, (US) INTEL, (US) University of Michigan, (US) PLDA, (France) Evatronix,(Poland) HDL DH, (Serbia) STMicroelectronics (France) IMEC, (Belgium) Inventure, (Japan) now Synopsys “Foundry” (Taiwan) UMC, (Taiwan) SMIC, (China) GUC, (Taiwan) KSIA, (Korea) Sony, (Japan) SilabTech, (India) Fabless, (Taiwan)

More articles from Eric…