Recent research on custom hardware to break RSA-1024

In 2003, Adi Shamir and Eran Tromer proposed an ASIC device called TWIRL, which should be able to sieve RSA-1024 using the General Number Field Sieve during a year, “only” using 10-20M$ of investment including NRE and power costs. Now, 17 years later, the cost of such a device should be exponentially lower. Has any research concerning custom ASIC cracking been done for RSA-1024?