Using final certified 2020 results for Pennsylvania:

• Statewide: Joe Biden 3,461,221 votes vs. Donald Trump 3,379,055, a Biden margin of 82,166 votes. (en.wikipedia.org)
• Philadelphia County: Biden 604,175 vs. Trump 132,870, a Biden margin of 471,305 votes (about “half a million”). (en.wikipedia.org)

If we separate Philadelphia from the rest of the state:

• Rest of Pennsylvania (all other counties combined) gives Trump a margin of 389,139 votes over Biden (3,246,185 – 2,857,046), inferred from the same county-level totals. (en.wikipedia.org)

Now, consider Chamath’s conditional: if Biden nets ~500,000 votes out of the Philadelphia-area count, that margin will be sufficient to carry Pennsylvania. Under the actual rest-of-state margin (Trump +389,139), a 500,000-vote Biden margin in Philadelphia alone would yield a statewide Biden lead of about 110,861 votes (500,000 − 389,139), so it would indeed be sufficient to win the state.

In reality, Biden’s Philadelphia margin ended up slightly below that (≈471k) yet was still large enough, together with his performance elsewhere, to produce the actual statewide win of ~82k votes. (ballotwire.com)

So, both the logic of the threshold (500k in Philadelphia being enough given how the rest of the state voted) and the outcome (Biden winning Pennsylvania with a Philadelphia margin in the same ballpark as Chamath’s “about 500k”) support scoring this prediction as right.