It’s been a bumpy old road down to Intel’s 10nm die shrink, the next process node after the seemingly eternal 14nm lithography we’ve been stuck on since dinosaurs were still walking the earth. Intel has repeatedly acknowledged the problems it’s been having, claimed to have sorted them out, and promised to have 10nm product on the shelves before the end of the year (now, where have I heard that before?), but at a recent HPC event in Houston Intel announced that it was actually ahead of where it thought it would be right now.

That’s almost damning itself with faint praise considering the huge delay in the outing of the new process technology, especially with AMD set to deliver its Ryzen 3000 CPUs and Navi GPUs in the second half of the year both using the 7nm node.

But Trish Damkroger, VP and GM of Intel’s Data Center Group, was speaking about Intel’s vision for the future of computing architectures at this month’s Rice Oil and Gas HPC conference in Houston – the same event where AMD suggested its chiplet architecture on its own is a short-term performance fix – and suggested the 10nm silicon might be here sooner than even Intel expected.