Data/information is like language in that it is socially constructed. Much like how we have arbitrarily assigned meaning to certain arrangements of letters, we have also assigned meaning to certain arrangements of electrons, or grooves in a record, or holes in a punch card, or varying length beeps in Morse code.
Only some information is meaningful though. Consider a word document with a random sequence of characters, a la infinite monkeys at typewriters. It's literally gibberish and not intelligible in any language. Colloquially, you would say there is no information in this document, but this is not strictly true. It has a bunch of symbols in a specific order, and it would be different from a document with the same symbols but in a different order. What the document lacks is meaning. This is how the colloquial and academic definitions of the term information are different.
In short, information is just symbols + the arrangement of symbols, and meaning is the significance of them.
Encoding information in physical objects
Let's go back to language for a bit. When you have an idea and want to save it for later, you write it down. But what are you actually doing to the paper? You're not embedding the idea in it; you're just putting a bunch of graphite on it. Where the meaning comes from is the fact that you know how to interpret the graphite as an idea.
In electronics, information is stored in binary: a series of zeros and ones. (These ones and zeros have meaning because we have collectively agreed that, for example, in a text document, 01000001 means the letter A.) However, you cannot put the concept of 0 or 1 in a rock, so you do things to the rock and interpret it as a 0 or 1.
Encoding information in microchips
What do they do to the rock? They make billions and billions of tiny holes and fill them with electrons. If the hole has a lot of electrons that's a 1 and if it's empty it's a 0. Each hole has a spigot to fill it with electrons, a drain to empty it, and a sensor to read how full the hole is. (There are other configurations of holes, spigots, drains, and sensors for different types of memory; this is just one of them.)
If you wanted to store the letter A (more specifically: the information that encodes the letter A) in the holes, you pick 8 consecutive holes and fill them or drain them accordingly. To read data you turn on the sensors and see if what holes are full or not, and then the computer translates the zeros and ones to symbols that humans can recognize and displays the human-readable symbols.
Consider the pigeon guided bomb, which is exactly what it sounds like. Pigeons were taught to identify targets on a screen and tap them with their beak, and the computers on board would adjust fins and thrusters in that direction. There is certainly a significance to their actions (people will die) but the pigeon doesn't know that. It only knows that if they tap the picture of a city on the screen they get treats.
We have tricked rocks into thinking because, like the pigeon, they are just following a sequence of instructions that looks like intelligent thought, but they don't actually know what they're doing. All they do is shuffle around information.
holes with electrons -> binary (information to information)
binary -> human-readable symbols (information to information)
human-readable symbols -> ideas (information to meaning)
Computers can turn information into other types of information, but only humans can get meaning out of it.
If you're interested in any of these topics, I would highly recommend John Searle's philosophy of mind lecture series, which is free on YouTube. It also covers computers and artificial intelligence within this context.