Executive Perspective: The Beginning of a Virtuous Circle for GLOBALFOUNDRIES FDX™ Technology

By Alain Mutricy

In our highly competitive industry, it is imperative for a company to continually challenge itself to move forward, otherwise it will surely find itself falling behind. At GLOBALFOUNDRIES, this means we are thinking big and embarking on technological paths we have not taken before to provide differentiated value to our customers. Our fully depleted SOI technologies, which we call FDX™, are one example. You’ve heard a lot from us lately about FDX technology. Our 22nm and 12nm FDX processes are ideal for low-power, mobile and highly integrated SoC applications, the sweet spot in the market for many of our customers. We originally developed the 22FDX® technology for IoT systems, which we estimate to be a $50B market opportunity for semiconductors by 2020 (Source: McKinsey & Company,based on volume forecast by Gartner, iSuppli, Strategy Analytics). It delivers the performance of 14/16nm FinFET technology on demand leveraging a software-controlled back-bias technology—yet it also supports ultra-low power systems and an ultra-low-leakage library for battery-operated IoT solutions. The process technology has been designed to integrate high-performance digital libraries with high-performance analog and RF circuits. First IP and silicon implementations have demonstrated the fact that beyond IoT, the 22FDX platform is ideal for low-to mid-range smartphone single-chip integration. Also, with the 22FDX platform we have broken the old paradigm of semiconductor technology development, which went as follows: the most advanced node would be developed for high-performance digital logic implementation. Then, more or less two years later, analog and RF would complement the process toolkit, as leading-edge performance customers were already moving to the next digital node. Finally, add another two years and you may have the ability to integrate non-volatile-memory (NVM). Which means, of course, that systems which would benefit from NVM integration could only integrate logic IP with four-year old performance capabilities. The 22FDX platform enables our customers to break free of this constraint, and to design intelligent, fully integrated (e.g., lowest power and system cost), and connected (e.g. with RF systems on-chip) systems. We see customers working on 28nm today, but also 55nm and 40nm customers (with RF and/or NVM) considering switching directly to 22FDX to leverage this competitive advantage for themselves. We’re already working with scores of customers on FDX technology, and a good number are in the early prototyping phase. While our unique FDX technology is the foundation of our offerings, we have also challenged ourselves to find ways to help customers bring their FDX-based products to market as easily and as quickly as possible. In particular, to make available tools and IP that enable customers to leverage the software-controlled body-bias capabilities of our FDX technology. To do this, we have developed the FDXcelerator Partner Program. It is a collaborative effort that provides customers with the comprehensive support and resources they need to get FDX-based SoCs to market as fast as possible. Think of it as an entire ecosystem of pre-qualified, committed expert partners and suppliers who, along with GF, stand ready to provide whatever customers may need to create and bring to market innovative SoC solutions quickly and cost-effectively. Such strong, strategic partnerships are crucial for any business, large or small. For example, let me make a comparison with something on the personal side. More than 15 years ago, I started a business within Texas Instruments aimed at developing microprocessor solutions for cellular phones, with the early vision that cell phones would become “smart,” have high-level operating systems, the ability to download applications and share media, and larger screens. We called it the OMAP processor. (How this world has changed in these past 15 years!!!) Despite a strong technology advantage versus the competition, with much more performance for multimedia and graphic, at ultra-low-power, OMAP in the early days represented a new type of dual-core solution to program. Our customers were viewing a competitor’s leading microprocessor solution as a safer bet, easier to implement, although the technology could not provide the needed battery life that would make smartphones become a mainstream technology. We launched a similar ecosystem network called the OMAP Technology Center program, to break the design barriers for our customers by bringing the best-of-the-best technology partners with optimized software solutions and tools to OMAP technology. In no time, our partners adapted their resources, skills, tools and their multimedia software or base-port software solutions to support the OMAP platform and take advantage of the emergence of the smart phone market with OMAP, which became quickly the market leader. OMAP customers were all winning in their markets, as they could develop at an accelerated pace. OMAP Technology Centers—our partners—grew very fast and shared in the success. So, back to our topic here: While we estimate that 22FDX SoC design complexity and the level of effort required is much lower than that of FinFET technology, designing with the full leverage of the software-controlled body-bias is a new approach to SOC design. Not complex, but different. (FinFET design complexity is about twice that of 28nm, according to a Gartner study.) We couldn’t have done it so well nor so quickly without expert help, which is precisely the goal of our FDXcelerator Program here. So far, we’ve announced seven world-class program partners and there is a backlog of others who want to join. Each partner we’ve evaluated and selected has committed to offer our customers specific, dedicated resources, and the program now encompasses:

  • Tools (EDA) – Modules that leverage differentiated FD-SOI body-bias features, built into industry-leading design flows
  • Design elements (IP) – Complete libraries, including foundational IP, interfaces and complex IP to enable foundry customers to start designs quickly with validated IP elements
  • Platforms (ASIC)
  • Reference solutions – System-level expertise in emerging application areas to speed time-to-market
  • Packaging and test solutions (OSAT) – Enabling state-of the art SOC delivery
  • Other resources – Design consultation and other services dedicated to FDX technology

This all adds up to unique benefits and added-value that is only available with FDX technology from GF. I see the FDXcelerator Partner Program as a key toolbox for our customers to accelerate their time-to-market. It is the catalyst for a virtuous circle in the use of FDX technology, whereby the support we offer encourages more customers to migrate to FDX technology, and their ideas and participation in turn stimulate the ecosystem to grow further, which in turn draws in more customers, and so on. We are simultaneously creating the opportunity for business success for our partners and time-to-market advantage for our customers, by enabling them to have solutions and resources at hand to increase their design productivity on FDX technology. It’s exciting to be at the beginning of such a transformative effort. I’m proud to be a valued partner with our customers, and I can’t wait to see them win in the marketplace with our technology! Excluding “classic internet devices” such as laptops and smartphones. Also excludes automotive applications. Rough preliminary estimate with indicative split by device type. Integration of simple devices with communication features and memory in SoCs assumed and accounted for in communications category, embedded memory under logic.