You just contradicted yourself.
I intentionally specified "gasoline internal combustion engine" and you said you agree that a certain fuel has an inherent limit to how much energy you can get out of it. That reinforces my original position.
To clarify: There are physical properties of the materials used in a design that establish an upper limit of possible efficiency. In the engine analogy, the fundamental limiting properties are the potential energy of the gasoline and the burn rate. So no, the upper limit of efficiency of the gasoline ICE are not practical limits (as in: configuration, materials, and losses), they are inherent (as in: unchanging physical property of the gasoline). We have not yet reached this upper limit of efficiency.
I think here my engineering background is causing a bit of a communication problem because what you wrote can be interpreted two ways. Please let me know which way you meant. I'll explain:
I initially assumed that when you mentioned 600 HP per liter you were referring to a liter of engine displacement. That's what I based the rest of my post on. There is no inherent theoretical limit to how much power you can get per liter of engine displacement. In theory you can make a 1 cc engine which puts out 100,000 HP given the right design and materials but you're still going to have to figure a way to get huge amounts of fuel per unit time into that engine.
Now if you're referring to 600 HP per liter of gasoline, then the units are wrong and it makes no sense. HP is a unit of power, or
energy output per unit of time. This is
not the same as energy. HP-hour is a unit of energy. HP-hours would be the proper unit to use if that is what you meant, as in you can't get 600 HP-hours per liter of gasoline. 600 HP-hours could be 600 HP for a period of 1 hour, or 300 HP for 2 hours, or 1 HP for 600 hours. It represents a unit of energy, and no matter how clever the engine design you couldn't get more energy out of a liter of gasoline than the potential chemical energy. Point of fact the best engines today get less than 25% of the energy out of gasoline. In theory lots of room for improvement. In practice Carnot's law and the combustion chamber temperature limits due to the melting point of the materials used to build engines will keep the practical efficiency well under 100%. It's an inevitable result of the means by which chemical energy in the fuel is converted to motion. An electric motor on the other hand uses a different process and can in both theory and practice convert 100% of the electrical energy into motion. The best electric motor designs exceed 95%. Batteries in turn convert upwards of 80% of their chemical energy into electricity.
LEDs don't have the practical limits of gasoline engines. In fact, a good analogy is that the LED is to the incandescent lamp as the battery-electric motor is to the gasoline engine. The battery-electric motor combination can easily convert upwards of 80% of the energy stored in the battery into motion because it's not a heat engine. The gasoline engine will be lucky to manage 25% of the energy stored in gasoline for practical reasons. Likewise, the process of photon emission in a semiconductor can in theory be 100% efficient without resorting to the super exotic materials a 100% efficient engine or a 100% efficient incandescent lamp would require. We still won't get there since nothing in this world is 100% efficient, but I think 80% is realistic, 90% a possibility.
In general the efficiency of any process is constrained by both inherent physical limits and the method the process uses. A fuel cell-electric motor can in theory extract more energy from a liter of gasoline than the best piston engine designs because it's using a different process. Same thing with LED versus incandescent. A 6700K filament would be over 90 lm/W but we lack a material which stays solid at that temperature. If we did, and if we coated the lamp with materials to reflect invisible light back to the lamp so we need less power to keep the filament at 6700K, we could in theory have 100% efficient incandescent lamp. Likewise, you could have a 100% efficient engine if you could combust the fuel at millions of degrees, or if your waste heat environment were at absolute zero. In practice none of these possibilities exist, so we look for other means to do the same tasks.