Two Words: Arcane Trickster
TLDR: An Arcane Trickster build can get more damage by casting True Strike to trigger a Sneak Attack than by either attacking two rounds, using a damaging cantrip and then attacking, or casting a damaging cantrip both rounds. Also, certain feats can combine with True Strike to trigger Super Advantage (3d20) or even limited Ultra Advantage (4d20) on an attack roll.
Damage Comparison: True Strike vs Attacking
Because Rogues are restricted to using a Finesse Weapon to get their Sneak Attack damage, most make one their weapon of choice. Finesse Weapons (as presented in the PHB) all deal 1d8 or less damage as their base. For the purposes of this example, let's assume our Rogue is using a 1d8 Rapier and has a Dexterity modifier of +3, making their base weapon damage with a non-magical Rapier 1d8+3.
By the time this Rogue can become an Arcane Trickster at level 3, they get an additional 2d6 damage if they can Sneak Attack. If this Arcane Trickster takes True Strike as one of their cantrips, then, in the event that they do not have another condition that will give them an opportunity to use their Sneak Attack, they can potentially trade 2d8+6[14] (assuming 2 successful attacks, one on each turn) for 1d8+2d6+3[13] if neither roll is a critical hit, so they average out to be pretty similar. If either of your attacks is a critical hit, without True Strike granting you the Sneak Attack damage, the trade at level 3 is 3d8+6[18] vs 2d8+4d6+3[23] with it.
Tempted yet? I'm only getting warmed up.
As time goes by, this trade becomes better and better, eventually capping out at level 19 when it would be trading 2d8+6[14] for 1d8+10d6+3[37] without a critical hit and 3d8+6[18] without the spell vs a whopping 2d8+20d6+3[71] with the spell if one of the rolls is a crit!
Double critical hits are rare enough that it's not worth comparing.
Because it takes a while for the return to be worth it casting it often (whenever you can't gain Sneak Attack from another source) it may be best not to pick True Strike as one of your Cantrips at level 3, but the only other chance you will get is when you get your fourth and final Cantrip pick at level 10.
It is also good to take into account that a Rogue is also the most likely character to make an attack roll before initiative is actually rolled. If you attack from hiding, you already get Advantage, and therefore, Sneak Attack, but what if you are in a conversation and decide that someone needs to be stabbed? A Dexterity(Slight of Hand) skill check to hide the fact that you just cast True Strike (no verbal component) could be a great way to still get to open with a Sneak Attack. Depending on your DM, you could probably get Advantage on your attack just by a successful Dexterity(Slight of Hand) roll to conceal that you are readying a dagger, and which of the two checks would be harder will change from DM to DM as well, but using the Slight of Hand check for the spell may be a good way to get to use your rapier instead of a dagger.
Damage Comparison: True Strike vs Damaging Cantrips
The most common argument against True Strike I have seen compares casting a damaging Cantrip instead of attacking the first round, so let's look at that comparison as well, using our Arcane Trickster and the Fire Bolt spell (improves at levels 5, 11, and 17) as our damaging Cantrip.
$$
\begin{array}{r|r|l}
\text{Level} & \text{Fire Bolt [avg. damage]} & \text{True Strike [avg. damage]} \\
\hline
3 & 1\text{d}10+1\text{d}8+3 \; [12] & 1\text{d}8+2\text{d}6+3 \; [13] \; \star \\
5 & \star \; 2\text{d}10+1\text{d}8+3 \; [17] & 1\text{d}8+3\text{d}6+3 \; [16] \\
7 & 2\text{d}10+1\text{d}8+3 \; [17] & 1\text{d}8+4\text{d}6+3 \; [19] \; \star \\
9 & 2\text{d}10+1\text{d}8+3 \; [17] & 1\text{d}8+5\text{d}6+3 \; [22] \; \star \\
11 & 3\text{d}10+1\text{d}8+3 \; [22] & 1\text{d}8+6\text{d}6+3 \; [25] \; \star \\
13 & 3\text{d}10+1\text{d}8+3 \; [22] & 1\text{d}8+7\text{d}6+3 \; [29] \; \star \\
15 & 3\text{d}10+1\text{d}8+3 \; [21] & 1\text{d}8+8\text{d}6+3 \; [31] \; \star \\
17 & 4\text{d}10+1\text{d}8+3 \; [27] & 1\text{d}8+9\text{d}6+3 \; [34] \; \star \\
19 & 4\text{d}10+1\text{d}8+3 \; [27] & 1\text{d}8+10\text{d}6+3 \; [37] \; \star \\
\end{array} \\
\text{(}\star\text{ marks larger damage source for that level)}
$$
As you can see, the True Strike spell will often get more damage per round for a Rogue than the Fire Bolt spell, especially at higher levels, and that doesn't even account for doubling your Sneak Attack with a critical hit!
The fact that you could also crit with your damaging Cantrip is valid, however the comparison becomes difficult since if one of your attack rolls is a crit, you don't get to choose which one it is, so if it was the Fire Bolt roll you could get as much as 8d10+1d8+3[47] vs 2d8+20d6+3[71] with True Strike at level 19, but if the crit was the attack, then it would be 4d10+2d8+3[31]. In both cases the True Strike spell provides the superior damage output.
Finally, it is important to remember that your Arcane Trickster could also spam Fire Bolt both rounds instead of using a Rapier or other finesse weapon on one of the rounds, in which case, at level 19 the trade would be 8d10[40] vs 1d8+10d6+3[37] without a crit or 12d10[60] vs 2d8+20d6+3[71]. As you can see, in this scenario, the Fire Bolt spell is more competitive and when you should choose one over the other is really situational. Luckily, if they want to, an Arcane Trickster can have both of these spells at their disposal and make good use of each one.
It's not all strawberry fields, however, since you still have to worry about the problems that come with delaying your attack a round: the enemy gets to act before taking damage, something could change that makes you not get to attack on your next turn, etc. Also, remember that this spell is only for those times when you can't get your Sneak Attack some other way, such as attacking together with a melee ally or the many other situations that grant Advantage. This spell is for those times when you are in a pinch and can't get your Sneak Attack any other way.
True Strike and Feats: Getting Super/Ultra Advantage
The first feat that you should take for this build is Lucky, unless your DM allows Xanathar's Guide, then your first pick should be Elven Accuracy (assuming you are an elf or half-elf).
Paraphrasing the The Elven Accuracy Feat from Xanathar's Guide:
Increase your Dexterity, Intelligence, Wisdom, or Charisma score
by 1, to a maximum of 20
Whenever you have advantage on an attack roll using one of those
ability scores, you can reroll one of the dice once
Be prepared for some DMs to veto Elven Accuracy though, because it essentially turns Advantage on all attack rolls into Super Advantage (best of 3 rolls) which will scare them.
As a DM, I would absolutely allow this feat, I don't believe it is over powered because it still requires Advantage, and sure, you can get Advantage more often with True Strike, but you also have to sacrifice an Action for that, so I feel like, rather than making it too overpowered, this feat brings True Strike up to the level it should have been on anyway.
If allowed, Elven Accuracy makes True Strike much better by not only increasing your chances to hit (57.81% chance to roll at least a 16) but also increasing your chances to get a critical hit (14.26% with best of 3d20 vs 9.75% with normal Advantage) and getting to double that Sneak Attack damage!
Add the Lucky feat to that and you will potentially get to roll as many as 4d20 on a single roll, granting what I call Ultra Advantage. Anyway, no matter what you call it, with this build, your two rolls over for two normal attacks can turn into 4 rolls for a single Sneak Attack.
With Ultra Advantage from one of your three uses of Lucky, your chances of rolling at least 16 are boosted to 68.36%, and your chances of rolling a 20 are 18.55%.
Again, I would argue that this is not overpowered because in the end, you can still only do this three times per day (and those uses of Lucky need to compete with things like Saving Throws), and only when you have Advantage, which you may have had to sacrifice an Action to set up for yourself. On top of that you are spending 2 Feats toward this build, so if your DM has an aneurism over it they need to chill out.