“A” versus “an” is all about the next syllable (and its sound), not the next letter—they’re the same word, just with context-dependent pronunciation which happens to be materialised in the spelling.
My favourite way of demonstrating this practically is to observe usage of “a” versus “an” on words that have developed or lost a silent “h”.
For example, in the days of long ago, “herb” had a silent “h”, so it would be spelled “an herb” as a vowel sound followed; and in American English it still is silent, and is consequently still “an herb”: https://books.google.com/ngrams/graph?content=a+herb,an+herb.... But in British English, people started sounding the “h”, so that in the 1870s, “a herb” became more popular than “an herb”, and it has remained so ever since: https://books.google.com/ngrams/graph?content=a+herb,an+herb... (though “an herb” is still much more common there than I’d have expected).
I normally read more recent translations of the Bible, but still sometimes read the KJV aloud, so there are many cases of silent aitches that are no longer silent (a regular expression search for `\ban h` has 711 matches in the KJV, compared with only 11 in the WEBBE, which are for “honour”, “honest”, “hour” and “heir”); I decided it’s better to read “a house” rather than “an house”, given that the original spelling was only chosen because at that time they pronounced it “an ’ouse”—or at least, I think so; the ngrams there aren’t quite so conclusive, I suspect that one might have already been shifting in some parts of England, as is the case with quite a few other such words; the KJV was written in a style that was already overly formal and sometimes slightly archaic when it was new, let alone 400 years later—Tyndale’s translation from the best part of a hundred years earlier reads much more easily once you normalise spelling as was done in the KJV between 1611 and 1769.
My favourite way of demonstrating this practically is to observe usage of “a” versus “an” on words that have developed or lost a silent “h”.
For example, in the days of long ago, “herb” had a silent “h”, so it would be spelled “an herb” as a vowel sound followed; and in American English it still is silent, and is consequently still “an herb”: https://books.google.com/ngrams/graph?content=a+herb,an+herb.... But in British English, people started sounding the “h”, so that in the 1870s, “a herb” became more popular than “an herb”, and it has remained so ever since: https://books.google.com/ngrams/graph?content=a+herb,an+herb... (though “an herb” is still much more common there than I’d have expected).
I normally read more recent translations of the Bible, but still sometimes read the KJV aloud, so there are many cases of silent aitches that are no longer silent (a regular expression search for `\ban h` has 711 matches in the KJV, compared with only 11 in the WEBBE, which are for “honour”, “honest”, “hour” and “heir”); I decided it’s better to read “a house” rather than “an house”, given that the original spelling was only chosen because at that time they pronounced it “an ’ouse”—or at least, I think so; the ngrams there aren’t quite so conclusive, I suspect that one might have already been shifting in some parts of England, as is the case with quite a few other such words; the KJV was written in a style that was already overly formal and sometimes slightly archaic when it was new, let alone 400 years later—Tyndale’s translation from the best part of a hundred years earlier reads much more easily once you normalise spelling as was done in the KJV between 1611 and 1769.
In short: you’ve got an NVIDIA card.