: Is this right? 142 km?

: I can buy that Halo is so huge, because Halo doesn't move and have to propel
: itself. The laws of physics are the laws of physics, and creating
: something this big that can move around would require a metric crap ton of
: energy.

The Forerunner abort baby universes for energy. We can assume that their power sources can keep up with demand.

The real question is, how do Forerunner ships propel themselves? Any child in primary school can tell you that no object changes velocity unless it is acted upon by an outside force. For that outside force, you need mass. Reaction mass. Could be a planet, could be ionized exhaust, or it could be something more... exotic.
The only problem with carrying your own reaction mass with you is that it gets expensive. The Tsiolkovsky rocket equation shows us that ruinous amounts of fuel must be carried simply to accelerate the fuel. This can be mitigated with a very high exhaust velocity, but the most the UNSC has ever generated with a magnetically coupled fusion rocket is 34,000 kmph. A rocket (or in this case, an experimental engine for the M58K Archer Missile) would need to carry 34% of its mass in fuel in order to accelerate to the minimum 10,000 kmph(R) considered suitable for space engagements. Thirty four percent is fine for a missile, but fairly impractical for warships and spacecraft that will do considerably more maneuvering.

To get these warships moving, you have to break the Tsiolkovsky equation.

In what nuclear physicists assure us is "Not the weirdest thing we've ever done with a fusion reactor", modern 'rocket' 'engines' function as supercolliders, mashing hafnium atoms with protons to create quark-gluon plasma millions of times per second. Pardon me if I'm dumbing things down to Popular Science levels, but custom elementary particles are created from the plasma and bound into supermassive protons with stronger-than-normal electric charges. It doesn't take long for the universe to catch on to what you're doing, about 1200 milliseconds. After that, it decays into a spray of photons and neutrinos. Just long enough for supermassive protons to be shunted into a rocket engine for thrust. Engineers call it "Paying Conservation of Momentum in fairy gold."

The upshot: The fuel zooming out the back of your warship masses a million times as much as it did sitting in the fuel tanks. Also, Jon's Law is averted. Because most of the exhaust decays into nothing ~15m from the engines, warships are free to maneuver around space docks and orbital stations without cutting habitats into hot ribbons.

The Covenant, of course, have done away with the need to carry reaction mass with them. Turns out, there's a fifth (Maybe more) fundamental force that interacts only with dark matter and, weakly, with exotic forms of baryonic matter. Anything with an impulse drive, from Banshees to Seraphs to supercarriers, use jets of dark matter for reaction mass.

The Covenant reverse-engineered the system from Forerunner Sentinels, and it shows. While Forerunner AI units hover and glide without any visible means of propulsion, Covenant impulse drives are fairly inefficient. The smallest impulse drives glow a bright blue and generate only about 28 kilonewtons of thrust. Larger drives cannot radiate waste energy fast enough, and require active cooling. That's what the 'engines' on larger ships do; what appears to be exhaust is simply waste radiation and denatured coolant, far too gentle of a stream to be useful as propellant. When the exhaust ports are damaged or destroyed, the Covenant must run their impulse drives at reduced power or be cooked alive inside their ships. The boys and girls in the Navy call this strategy "Clipping the wings."

Now back to the Forerunner. How do they propel their large ships? The answer is "We don't know". Earth's orbital defense network was running hot when the Mantle's Approach punched through it, and not one station detected an impulse drive. Our scientists are hard at work on the subject, but even with vessels taken in joint UNSC-Sangheili operations, it's doubtful we will know the answer for thousands of years.