I think apart from the scale, they view labor as a risk, rather than an asset. The ironic part to me is they're hiring "genius" level talent to create these dummy level algorithms.

That's certainly a part of the irony in how much they'd prefer to hire white collar SV labor than blue collar labor anywhere else. For the average salary of just one PhD in "data sciences" in SV cost of living you could often hire an entire floor of a call/support/service center in some Midwest state. (Though few of those consider labor an asset either, that's a more systemic problem in how the current zeitgeist of capitalism views the entire labor market, unfortunately.)