I wouldn't trust that figure in the slightest.
As HQHost says, it's essentially impossible to get 100% uptime. At least, it's impossible with a standard dedicated server package (as below, you'll be looking at $x,xxx or $xx,xxx per month for near-100% uptime)
Now it *is* possible to get incredibly high uptime (99.9999%)/near 100% uptime barring any major, trans-state emergencies simultaneously happening. But the cost would be incredibly high.
Firstly, you'd need a cluster of mirror dedicated servers (each with redundant power supplies, and in a high quality DC with all the 'bells and whistles') in two geographically separate locations (call one 'Primary' in.. Texas, and one 'Secondary' in.. CA)
Then, if one of the servers in 'Primary' went down, one of the other servers in the cluster would just 'kick in' and serve up the content (whilst the one that just went down could get fixed). This ensures very high redundancy.
However, the DC itself could have a major power cut or emergency (like The Planet's DC which had an explosion and was shut-off for a while, or another DC a few months ago which was raided and completely shut-off for a while by the FBI, etc). Hence, if this happened, the 'secondary' cluster would kick-in.
So yeah - it *is* possible to get very near 100% uptime, but only if you are wanting to pay $x,xxx or $xx,xxx per month. And even then you can't guarantee 100% uptime 24/7/365.
But to say 100% uptime on a <$100 package? That's impossible sorry