I agree, significant buffs for such units is unable to lead to fun and stable balance. However, the short period after they are buffed may be more fun than usual while people scramble to figure out the best way to exploit the changes. The units may need small buffs, but the issue here is that it will take a long time to notice the changes.
I've also been thinking about two bigger changes relating to frontline constructors. The changes are fairly crazy as they may break everything. One is to make repair more expensive, and the other is to make reclaim faster. If you want some global changes to shake up the whole meta, these are decent changes to consider.
Expensive repair is motivated by two observations. The first observation is that low alpha artillery is significantly worse than burst artillery because constant damage is easy to repair away. See the power levels of Sling and Badger (when used as artillery). The second observation is that it feels like a victory when a big unit gets to retreat and repair after killing almost any number of smaller enemy units. On the receiving end, it is not fun to have the last 1% of damage on an enemy strider be the only important piece of damage. The second observation is more impactful, and tends to dominate mid-late game with one (probable) effect being reduced unit diversity. I am unaware whether it is possible to make repair cost metal, but if we are limited to energy, then a high enough energy drain corresponds to a loss of overdrive metal. Significant energy drain would also make repair a 'high tech' ability as a larger economy has a better conversion rate of metal to energy, for the purposes of overdrive lost due to energy spent on repair.
Repairing a unit from 0HP to full currently requires 50% of its original energy cost. It takes 50% as long as it would to build the unit, unless the unit has taken damage in the last 10 seconds, in which case it takes 200% as long. This 10 second repair disruption rule was added to keep combat repair in check. Without this rule, constructors would be taken into battle much more frequently and non-burst artillery would be even more terrible.
Faster reclaim is motivated by the fact that a reclaiming constructor is barely better than a mex. Constructors take about 30 seconds of reclaim to pay for themselves, whereas a +2 mex with no overdrive takes 37.5 seconds. Obviously, mexes are limited and are built by constructors, so it is not a perfect comparison. This idea is not as well thought out, and probably more dangerous.