The U.S., by many measures, is not a democracy, but more closely operates as an oligarchy [1].
If Bill Gates donates $100 million to a set of Congressional candidates who then use that money to run campaigns, the additional money will have a noticeable impact in the result. Assuming nobody is on the other side making even larger donations, a higher percentage of Bill Gates' preferred candidates will get elected than would have otherwise.
If it was just Bill Gates doing this, it probably wouldn't be that big of a deal. But since the economic and political interests of a large percentage of America's elite are highly correlated, you end up with a large amount of money directed to candidates who support policies that benefit those elites.
In theory, your critique is correct. If every non-elite voted against the elites, all of this money would be wasted. But in practice a lot of that money gets used to convince the non-elites that their own interests will be served by the same candidates who are receiving the money from the elites. It would be called propaganda if it was done by the government, but since it's done by private economic entities it's just considered a form of marketing.
The average American voter is not very informed beyond a few specific issues (and often not even that). The result is that they can be easily overwhelmed with information related to issues they are unfamiliar with. Political campaigns in the U.S. like to portray these issues in dire terms. "My opponnent is soft on crime! Why? Because he voted for this obscure bill, HR834109241027308, that you've never heard of and couldn't interpret if you tried!" The implication is straighforward, but never directly said: "You and your family will be safer by choosing me!"
You don't have time to research all of these claims and debunk them, even if you have a good intuition for which ones are bogus and which are reasonable. The fact that both major parties bullshit equally means that even if you did, you'd still be trying to pick the most truthful liar out of the group. Some people recognize this and give up on the process entirely. A lot of money is spent trying to convince these people that this candidate is different and is worth your money, vote, and trust.
So you are correct that trying to out-money the monied interests is not a very promising strategy for political reform in the U.S. But simply trying to out-number them won't work since the average vote is so uninformed, subject to pathos, and fundamentally cheap. Over a large population, the average vote is all that matters. For this reason, small local elections in the U.S. tend to function in a much more democratic and intelligent fashion than state or national elections. The issues are less numerous, the stakes are lower (implying lower ROI for political "investors"), and the voters are more informed about what is happening in their own area.
Democracy, as discovered by the Greeks, is a fantastic political system for local governance. But it doesn't scale. From a certain perspective, it is impressive that the U.S. has scaled it as far as it has, given such a numerous, heterogeneous, and geographically distributed population.
[1] http://www.bbc.com/news/blogs-echochambers-27074746