Your browser is not supported. Please update it.

21 September 2014

Israel set to drive Intel’s push to 10nm and IoT

Intel is to make the biggest ever investment by a foreign firm in Israel, upgrading its fabrication plant there at the cost of $6bn, to ready itself for new processes – 10 nanometers and below – which will push to new frontiers in mobile and IoT chips.

There have been many months of speculation about where Intel would center its 10nm manufacturing, with commercial chips promised from 2016. This geometry will help drive down the size and power consumption of chips to enable new types of ultra-low power devices and sensors, particularly for the IoT (internet of things), at new price points.

The shift is strategically critical for Intel, which has largely missed the boat in the first generations of mobile processors and SoCs. It now has a credible Atom-based offering for handsets, but the growth is moving rapidly towards new mobile and embedded devices such as wearables and a new wave of M2M systems. Embedded devices are upgrading to broadband connectivity and greater processing power, in many cases signalling a transition from simple microcontrollers to full processors, if these can be produced at sufficiently low cost and power.

This is where Intel must seize the opportunity to lead an important market again, and fend off the challenge from processors and controllers based on ARM or MIPs architectures. The US giant has always pushed the boundaries of manufacturing technology to help it stay ahead of rivals, many of which – like Qualcomm and Broadcom – are fabless and are relying on their partners to upgrade their processes effectively.

The decision to spend $6bn on upgrading the Kiryat Gat plant in Israel almost certainly means this will be the first location for 10nm manufacturing, and it ensures that the Israeli location will remain an important element of Intel’s strategy.

Had the firm chosen to drive the 10nm transition from US or (as often reported) Ireland, Kiryat Gat would have been effectively sidelined. For this reason, the Israeli government has fought hard to keep Intel’s support, offering grants of up to $600m and a significant discount on corporate tax rates for the next decade. Intel has also promised to hire an additional 1,000 employees at the plant. Israel’s finance minister, Yair Lapid, said Intel would more than make up for its reduced taxes with the enhanced economic activity it would generate.

In its statement, Intel did not actually mention 10nm, but said the upgrade would transform the plant with “the world’s most advanced technology for the production of micro-electronic components”. Mooly Eden, CEO of Intel Israel, acknowledged earlier in the year that his government was battling with that of Ireland to secure the upgrade, commenting to reporters that both countries “”know how the game is played. In today’s world, this is how governments bring jobs in”.

Eden commented on the new deal: “Israel gets back a lot more from Intel that it does from other multinationals. Over the years Intel has invested $10.8bn in Israel. Taking into account all of the services and outside contractors we use, Intel’s activities in Israel are responsible, in our estimation, for some 30,000 jobs in the Israeli economy.” Intel has received a total of $1.3bn in grants and incentives from Israel. The Irish plant at Leixlip remains significant, and Intel has spent $5bn over the past few years on updating it.

Intel is expected to announce details of its 10nm roadmap next month. It has recently moved from 22nm to 14nm for some products, notably the new Core M processors, whose initial achievement is to enable fan-free tablets and laptops. The 14nm process will also help to boost Intel’s offerings in mobile and embedded, but 10nm is widely seen as the milestone that will transform competition in this space, especially the emerging IoT.

However, the 10nm transition will be a technically challenging one. At this level, the properties of silicon begin to break down, so new materials may have to be introduced too. Even with the introduction of 14nm, Intel has faced issues which delayed shipment of its first chip at this geometry, Broadwell. Mobile processors based on Broadwell are appearing this fall while desktop chips will follow in 2015, along with another 14nm design, Skylake. If Skylake goes smoothly, it will be followed by a 10nm version – Intel’s first 10nm offering – called Cannonlake, in 2016.