It costs energy to flip a single bit in a computer and during our use of
the computer we do flip a large number of bits! If you supercool
a computer down to 3.2 K, it would “only” consume 4.4 * 10^-16 ergs
for setting or clearing a bit.

The annual output of the sum is 1.21 * 10^41 ergs. If you capture all this annual energy and use it to flip bits, then one sun-year has enough energy for this supercooled computer to cycle through all the combinations up to 2^187, otherwise known as cracking a 187 bit symmetric key (actually just setting the bits in the key required to test it).

If you capture the annual energy of the sun for 32 years, you get to
192 bits. If you capture the energy that is released from a supernova you can start breaking a 219-bit symmetric key. That is not even close to a more generally used 256 bits symmetric key.

Finally, this is only changing the bits of the key. It does not even consider running the key through the encryption algorithm and automatically recognizing the calculated result. And of course, there is no time available to cycle a computer through all these changes. In 1995 when the book was written, with a budget of $10T, it would take 10^11 years to break a 128-bit symmetric key. But this budget for crypto will go down a factor of 10 every 5 years and is speeding up rapidly.