TI set to double emerging RFID business
TI, the only major RFID chipmaker in the US, doubled its RFID business in the emerging payment and retail supply chain markets during the past year, the company told ComputerWire yesterday. "We expect we will again next year," said Tony Sabetti, UHF/retail supply chain director at TI's RFID systems division. TI pulled in $12.6bn in sales of its various semiconductors and sensors for diverse applications and last year. Its total RFID business, which ranges from vehicle security to ticketing, is comparatively tiny, driving just $100m or so in revenues. The bulk of TI's RFID retail supply chain business is in the US, where Wal-Mart mandated RFID implementation among its suppliers ahead of other large retailers elsewhere. But now TI is moving to boost its existing 20% market share and is aggressively investing in new silicon-making technology. After all, the RFID industry is predicted to triple to $6.1bn by 2010. TI is betting on its 130-nanometer manufacturing node to etch an untouchable competitive advantage for the company. This technology promises to drive down silicon costs, which is key to the success of TI's RFID strategy. After all, silicon is the most expensive raw component of an RFID tag, along with RFID antennas, which TI also makes. "We've taken a leap of faith and designed the silicon itself on the most advanced analog node available today, the 130-nanometer node," he said. Unlike digital microprocessors that currently are being manufactured at smaller, 90-nanometer nodes, RFID is based on analog technology. Already TI has begun limited 130-nm silicon runs, but full production is slated for the fourth quarter, he said. Currently, TI has the manufacturing capacity for several hundred millions RFID tags and Sabetti said it would be able to reach one billion units by years' end. But it won't be until at least 2007 until TI reaps the benefits of its investment because 130-nm technology is currently very pricey. As silicon wafer costs come down, which they always do, 13-nm production will become cheaper. In the meantime, the company claims an edge on competitors because the 130-nm die size is smaller than 250-nm production nodes and, therefore, more power efficient. Smaller silicon also means more robust tags that are less prone to breakage and wear-and-tear, Sabetti said. "It positions us from the get-go in the most competitive place from a silicon standpoint," he said. "It's always one [area] in which our competitors are going to play catch up. Everyone else we know is on 350-nm or 250-nm nodes." Infineon Technologies, Philips Semiconductors and STMicroelectronics are chief RFID silicon competitors to TI. TI's manufacturing clout is for its newest RFID silicon that is based on ultra high-frequency EPC Generation 2, which the International Standards Organization currently is reviewing as the global RFID technology standard. EPC Gen 2 is widely expected to be the technology basis for ultra-high volume, international supply-chain RFID deployments. Gen 2 has several advantages of early EPC Class 0 and Class 1 standards, including a disabling feature, security password and better performance that enables a greater volume of RFID tags to be read per second. While the ISO likely won't give Gen 2 the green light until early next year, RFID hardware makers already are beginning to churn out Gen 2 products. The thinking is that enterprises would be able to embark on multi-site RFID projects knowing their equipment won't be obsolete within a year. "The Gen 2 technology, which has been held out as this great promise for retail supply chains, is imminent," Sabetti said. TI released its Gen 2 silicon in late June and currently is manufacturing it at its 13-nm plant in Dallas, Texas, where the company is headquartered. Once production ramps by years' end, these chips will be manufactured in Malaysia, where TI makes its older-generation tags. This week, RFID printing company Zebra Technologies was the first company to demonstrate printer/encoder communication with TI's Gen 2 silicon. The companies touted a tenfold performance improvement over earlier Gen 1-based products. Sabetti said ultra high-frequency (UHF) Gen 2 tags promises a dramatically lower cost per inch of reading range versus earlier products. "But it has some real application limitations," he said "Signals that bounce around too much, signals that are too heavily attenuated by dense material and so it's not entirely clear to the industry as a whole, whether UHF will be applicable in every application." Currently, UHF is mainly used in case and pallet tracking in retail supply chains. But retailers want to more than just track the material, Sabetti said, and it's not yet clear with UHF is the right tool for applications such as "smart shelves." Smart shelves track items on a store shelf so a retailer knows when a product is out of stock, for instance. "This is an exciting opportunity in apparel and in packaged media ... we don't know whether UHF can work with smart shelves but we know [older-generation] high-frequency can," Sabetti said. Beyond UHF Gen 2, TI is currently working to add more memory to an RFID tag, to record information about a product's journey through a supply chain, such as a date stamp. The company also is working on additional security features, Sabetti said. Beyond memory, TI plans to leverage its existing sensor capabilities and push into a host of RFID sensor capabilities, such as enabling a tag to record the history of product's temperature throughout the supply chain, he said. Obvious applications would be cold-food retail and pharmaceuticals. "Sensors, I believe, are a few more years out," Sabetti said. "We believe strongly that the evolution of RFID in the supply chain leads to an enterprise database system that not only can find everything in its supply chain but tell you something about it."