As I've already said, I think expectations that Anaconda would become less powerful fundamentally misconstrue the situation. What matters to Microsoft meeting their goals is the total actual cost of the console, and the perceived selling points. Retail price isn't pinned directly to either of those things.
The intended Lockhart consumer was people who care most about price. The intended Anaconda consumer was people who care most about performance. So which scenario makes more sense?
Lockhart: costs $325 to build, sold for $350
Anaconda: costs $600 to build, sold for $500
If you sell in a 70/30 ratio (which is what they projected), you lose ~$12m per million units (mild loss is typical for new platforms, as games/accessories/services make it up).
Lockhart: costs $360 to build, sold for $300
Anaconda: costs $500 to build, sold for $600
A 70/30 ratio here costs you about the same amount. But there's many advantages. First, the build cost gap is only 40% instead of 85%--these two devices can more easily run the same games and services. Second, the majority of the audience are Lockhart buyers. They're getting an absolutely lower price for slightly better hardware, and the perceived savings from the premium choice is doubled. Yes, your enthusiasts get less, but will they even perceive it? It's likely a 17% lowering of cost is not going to require a full 17% dip in performance. You can engage your world-class marketing team to effectively build hype for the USPs the premium experience provides. And even with this "weaker" Anaconda, you're still talking about a build price likely higher than any PS5, with its one-size-fits-all needs. You pay more but you get more, and that's something Microsoft carried through with One X.
But there's an even bigger advantage latent here. And that's in answering the question, "What if the split isn't 70/30?" In the first scenario, Anaconda may be such a good deal that the percentage rises to 35, 40, 45.... And every step, you're losing more and more money on hardware. This makes sales, bundles, and future price cuts harder to do. With the second scenario, you lose less money if your premium console is unexpectedly popular. You could pocket that cash, use it defray the cost of an incentivizing price cut on Lockhart, or even bolster Anaconda with sales or bundles. And what if the ratio goes the other way, say to 80/20? You do lose more money, as with a rising ratio in scenario one. But the way out is to promote sales of your best product, with bundles or specials that don't impact margins too deeply. In addition, it's generally harder for this small amount of "premium interest" to approach the floor than for it to grow (i.e. Xbox has built a dedicated hardcore audience).
I think that's a pretty thorough case that the second cost-vs-price scenario is more likely to be how Microsoft were planning (even if the exact figures aren't correct). Which means if Lockhart is removed, you don't really have to downgrade Anaconda. With only one SKU, it could still cost $500 to build, but you sell it for $500. At worst, you'll be up against a PS5 that also costs and sells at $500...just fight them on services. More likely, you're against a PS4 that costs $450 to build and sells for $400, or even costs $400 to build and sells for $400. Either way, your machine is more expensive but it's also more powerful, maybe significantly so.
This will probably have no effect. First, even in a premium position Anaconda is unlikely to have ever been planned for zero disabled silicon. With such large chips, the portion that come off the wafer without any faults is very small, no matter whether you're planning a lower binned tier to absorb costs. The Hovis Method would help, but it'd be most cost-effective simply to plan for disabling from the get go, as with One X.
Second, there wasn't a lower tier to utilize defective chips. Lockhart was rumored to have a GPU half as powerful as Anaconda, or maybe even as low as a third as powerful. Binning chips between such disparate products defeats the purpose. The Lockhart APU would end up being massively oversized for its active silicon, and thus a waste of money.
