Music streaming services, such as Spotify, have the potential to disrupt the social dynamics of music consumption in ways not previously encountered. Not only do these platforms offer anytime, anywhere access to vast catalogues of music at little or no cost, they are seeking to manipulate what and how people consume it. Combining the judgements of music experts, extraordinary volumes of data about people's identities and interactions, and computational techniques designed to extract and predict similarities and differences in musical preferences at scale, music streaming services are adapting what music is selected and presented to individuals on an increasingly personalised basis. Drawing on mixed qualitative methods and working through the case of Spotify, the market leading service in the UK, this thesis explores if and how music streaming services are shaping the part music taste and consumption play in the reproduction of class. In the 1960s, Pierre Bourdieu (1984) demonstrated how patterns in cultural tastes and consumption practices are shaped by class background and - at the same time - serve as a mechanism through which class divisions are reproduced. On the one hand, this thesis demonstrates that Spotify challenges existing class practices by undermining the potential for musical expertise and a capacity to appreciate music as an end in of itself to function as a source of distinction for the middle classes. On the other hand, this thesis argues that Spotify creates opportunities for the middle classes to (re)deploy their cultural assets both 'on' and 'off' platform, through practices such as playlist creation and vinyl music consumption. In doing so, this thesis contributes to debates about the social dynamics of music consumption and the changing nature of the cultural assets underpinning class privilege in the 21st century.