Why is FAT32 limited to just under 2^28 clusters?

With FAT16 the maximum partition size is 2GB when your maximum cluster size is 32K.

This is calculated by multiplying the number of addressable units by the cluster size.

(2¹⁶ Allocation units) * (2¹⁵ bytes/cluster) = 2 GiB

However with FAT32, when I do the same calculation I get a much larger number than the 8 TiB maximum when using 2³² clusters.

(2³² Allocation units) * (cluster size)

If I use a cluster size of 512 bytes, I've already arrived at 2 TiB.

In an XP TechNet article, Microsoft says

The maximum possible number of clusters on a FAT32 volume is 268,435,445, and there is a maximum of 32 KB per cluster, along with the space required for the file allocation table (FAT).

This puts the maximum cluster size at 2²⁸ - 11.

Why is the maximum number of clusters in FAT32 2²⁸-11 and not 2³², given that it was 2¹⁶ in FAT16?

FAT32 only uses 28 bits not 32. Four bits are "reserved for future use".

So, a FAT32 partition has a maximum cluster count of 268,435,455 (2²⁸-1)

Reference

Although VFAT was a clever system, it did not address the limitations of FAT16. As a result, a new file system (and not just better FAT management as was the case with VFAT) appeared with Windows 95 OSR2. This file system, called FAT32 uses 32-bit values for the FAT entries. In fact, only 28 bits are used, as 4 bits are reserved for future use.

With the appearance of the FAT32 file system, the maximum number of clusters per partition went increased from 65535 to 268,435,455 (2²⁸-1). FAT32 thus allows much bigger partitions (up to 8 terabytes). Although the maximum theoretical size of a FAT32 partition is 8 TB, Microsoft has voluntarily limited it to 32 GB on Windows 9x systems to promote NTFS