When the solid-state lighting market takes off remains unclear. Some observers say it could drive a 180 billion unit market in 2015 or 2016. But getting the cost down is key, said Dennington. He shows how 8 or 7 Plessey LEDs can be used to make something of equivalent brightness to a 60-watt incandescent bulb. He then compares that with a currently available LED bulb that has 75 low-performance LED die inside.
LED lightbulbs are currently priced at about $30 to $50. If they can be brought below $20 and driven down to $10 per bulb then the market will start to move, said Dennington.
Each six-inch wafer can produce about 14,000 die and Dennington estimates Plessey's production capacity at about 2 million LED die per week. "The MOCVD machine is the constraint, so we plan to buy more," he said.
Plessey's plan is to purchase a further three MOCVD machines to fill out the 6-inch clean room and the company then has the option to make use of an unused testing lab area for another eight MOCVD machines. "The next three machines we expect to install in the next 18 months," Dennington said. The MOCVD machines cost about $2 million each but can help generate annual LED production worth $20 million.
"We've got two options. One is to scale our production as we grow. The other is to find a partner in Asia," said Dennington. Forming a joint venture in Asia might be necessary to engage with some customers who might want a second source of supply and die production that is local to light bulb and luminaire production, said Dennington.
For now Plessey is focused on making blue LEDs optimized to work with a phosphor to produce a broad-spectrum yellowish white light, a good replacement for incandescent bulbs. Meanwhile Dennington and Plessey is already thinking about how to make more of the ICs for smart light bulbs. This could include driver ICs, sensors and microcontrollers for dynamic color control, power management ICs. "We're already in discussion with companies that make PMICs," said Dennington.
Dennington added: "We do have an intent to move from discrete components and ICs to subsystems and even systems. It's a whispered strategy. We start with components first but we see a clear opportunity."
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.