There’s no way Starlink will be better than fiber. You have to pay a 25 ms penalty just to play, and then you have the cost of each hop, of which there are many.
Musk is selling precisely that to NYSE-London traders. Speed of light in a vacuum is considerably (I think 40%) faster than speed of light in glass fiber optics
Like I said, there is a latency barrier to get in and out of the atmosphere due to RF modulation. Only NYC to Tokyo will be close, but that’s generous, as it doesn’t take into account intersatellite routing delays. Whereas, NYC to London by fiber is a single hop.
Further, note that Starlink’s current batch of satellites don’t have laser links, so data can’t hop from satellite to satellite. It’s up and then back down immediately. We’ll have to wait for v2 or v3.
highly recommend you watch these two latency modelings:
These don’t include the latency cost to get through the atmosphere, or cost of routing at each satellite. As such they are fairly worthless.
I should also note that many of the simulated Starlink paths between NYC and London are actually higher latency than the best fiber path, which is less than 60 ms (https://arstechnica.com/tech-policy/2010/09/first-nyclondon-...). The author ignores routing and RF delays in the space network, but compares RTT to the public Internet. A joke.