If the machine has enough memory and speed as to compute all states of the Chess game in a reasonable time, can a player with the white pieces - operated by a machine - lose a game?
This is unknown at the time of writing. Further, according to solving chess on Wikipedia, no resolution is expected in the near future.
Let's take alternative chess. The rules are identical to chess, except that White can pass in it's very first move (but Black can't, even if White passed).
Now it's obvious that White has a strategy to never lose. Because if every move except a pass leads to a loss, then White can pass, and we know that whatever move Black does will lead to a loss for black.
In real chess, nobody knows. And nobody knows if there is a forced win for either white or black either. It is also reasonably likely that there is a time and space limited algorithm that can force a draw.
White starts with an advantage - 1/2 tempo. (Why one half tempo? After e3-e5 e4 White has lost one tempo, and White and Black have effectively swapped sides). We may suspect that with perfect game any advantage would tend to increase. But you need a considerably better position than your opponent in order to win and might not be able to ever get that.
Maybe it is worth mentioning why some kind of reasoning like
White has a winning strategy because everything black does, white could have done before.
does not work. You might already be familiar with Zugzwang, i.e. a position where the one who has to move loses (the game or just material) or at least weakens his position. The Trébuchet position shows an example, with reciprocal Zugzwang; here, the player who does not have to move next has a forced win.
It is not known whether the initial position is such a position with Zugzwang for white and a forced mate for black.