It makes it hard to place an exact price tag on how much this employee's coverage would have cost the company without knowing the details, but it's clear that it would have been very expensive to them one way or the other - and certainly a lot more than it cost them to have her switch to her husband's insurance instead.
[0] It's just like how your renter's insurance rates will [often] skyrocket immediately if you try to file any claim at all. People aren't used to this in the consumer health insurance market because there's a lot of indirection in the healthcare system, but that's how all insurance models work at a high level[1], so that's how the costs all (eventually) flow (extra emphasis on "eventually").
[1] There's also the added wrinkle that we also use "health insurance" as a political tool to do things that have nothing to do with risk smoothing at all, so the model gets tricky to unpack, but that's a separate matter.
Large companies actually underwrite their own plans (this is known as self-insuring[0]). The plan is managed by a known insurance company (e.g. Aetna, Blue Cross), so from the employee's perspective it's the same, but the actual risk is borne by the company. I would be very surprised if Amazon did not do this, given their size (~200K employees) and company age (over 20 years).
As a rule of thumb, the break-even point is about 1000 lives covered. Once a company gets to that point, it starts to be in their best interest to underwrite the insurance plans they provide.
Source: Founded a health-tech company that sold to insurers (which meant selling to companies that self-funded their plans as well, as they were they real risk-bearing entity).
[0] https://en.wikipedia.org/wiki/Self-funded_health_care