In what may have been his last keynote address as chief executive officer of Intel Corp., Craig Barrett said last week that U.S. math and science education needs an overhaul. "The biggest hurdle to technology growth we have today is the sorry state of our education system," Barrett said at the Intel Developer Forum here.

"I think one of the great challenges for all countries is to continue to produce a flock of good technologists. The constant influx of new ideas and new technologies is absolutely critical for us," said Barrett, echoing a theme he has been sounding since becoming top boss at the world's largest chip maker in 1988. He will become chairman of Intel in May when president Paul Otellini takes on the CEO title.

"Even our best students don't compete very successfully with their international counterparts," said Tracy Koon, director of corporate affairs at Intel, who helps oversee the $100 million budget in education programs started under Barrett in 1999.

The number of both U.S.-born and foreign-born engineering graduates in the United States is on the decline, said Koon, citing the rising quality of overseas universities and job opportunities. "Asia produces seven times the number of engineers per capita that the U.S. does," she said.

Intel has developed several programs to help stem the tide, including an effort that has provided professional development in math and science education for 2.5 million teachers to date. The company also sponsors a national science fair.

Barrett shared the stage here with Burt Rutan, whose Spaceship One last year won the $10 million Ansari X Prize for commercial manned space flight. "The last 30 years of space flight we have been boring our kids with little progress, and we can't afford to bore our kids," said Rutan.

Barrett also cited the need for investment in research. "Basic R&D in the physical sciences has been flat for more than the last two decades. That's an Achilles' heel for U.S. innovation," he said. "The good ideas are starting to be developed elsewhere."

Meanwhile, process technology continues to advance. "We've been hearing about the death of CMOS in the next 15 years for about 20 years, but smart engineers keep finding ways to keep it alive," he said. "Today we can see how to do this down to the 5-nanometer range. Beyond that, leakage current gets in the way." In the more distant future, "there are plenty of contenders for the next-generation transistor, from nanotubes to quantum dots, but I don't think people have decided which it will be," said Barrett.