While interesting this approach is not any faster than the Ge based photodiodes that we have currently. State of the art is ~-110GHz in Germanium Si photonics using standard fabs. This is only 50GHz.
"the signal state transitions were established at less than 20 ps, which is comparable to the bandwidth limit of the oscilloscope utilized at 50 GHz. [...] the initial rise time constant may be shorter than the measurement limits of the oscilloscope."
producing raw material does not mean you can process it.
The American defense branches have been searching for Ge-free alternatives for thermal optics, etc since during Biden's presidency. The trade war is significant, but does not change the calculus here.
I got the impression that "faster" refers to potential bandwidth, as the device responds to shorter wavelengths (300nm to 1.6nm) than other photodetectors.
Shorter wavelengths means higher frequencies, "number go up" = more data per second.
This is press release language though, it could mean anything.
Sure for InGaAs. The paper is claiming faster for AI so the relevant metric is to compare to what can be shipped today for such applications. To scale for AI we need devices that integrate with standard CMOS processes. More exotic materials can be integrated but they are not very practical for highly integrated systems. NVIDIA’s recent announcement for their copackaged optical switch is in a silicon only ecosystem.
"the signal state transitions were established at less than 20 ps, which is comparable to the bandwidth limit of the oscilloscope utilized at 50 GHz. [...] the initial rise time constant may be shorter than the measurement limits of the oscilloscope."