Evidence: Well-supported for the IBM Nature and Science circuit papers. “Smartphone displays” in newspaper ledes are speculation attached to a mixer demo.
RF devices
Wu, Lin, Bol, Jenkins, Xia, Farmer, Zhu, Avouris (IBM), Nature 472, 74 (2011) — top-gated CVD graphene RF FETs on diamond-like carbon, gates to 40 nm, f_T to 155 GHz, weak temperature dependence down to 4.3 K. Different substrate and growth route from the 2010 SiC 100 GHz result. Still cut-off frequency on a discrete FET.
Circuits, not only transistors
June 2011: IBM reports a graphene frequency mixer on silicon (broadband mixer, mixing demonstrated to ~10 GHz) — Science circuit paper, covered contemporaneously by the New York Times. First widely cited circuit, not a SoC. IBM notes large-area film production is still being learned.
Policy
FET Flagship shortlist / proposal work is public enough that universities announce their role. Launch is still January–October 2013.
How to read GHz headlines
f_T scales roughly as 1/gate length in these papers. Silicon at 32 nm logic nodes is a different contest (density, power, complementary logic, foundry PDK). Graphene RF is a real measurement. “Graphene replaces silicon this decade” is not a measurement.