AI PCs open for preorders—but will the features shown in the demo arrive at the same time?
The easiest mistake to make when watching a launch event is to assume that everything announced on the same day will all be available that same day.
Microsoft announced on October 7 that the Surface Laptop Ultra, powered by NVIDIA chips, is open for preorder and will go on sale starting October 16. But local model routing on Windows won’t enter experimental preview until later in October, while hybrid AI features for Copilot+ PCs are expected to roll out over the coming months.
These are three separate timelines, and they shouldn’t be collapsed into the claim that “the full experience has already been delivered.” Nor does a preview mean every employee at a company can use it reliably right away.
This affects when the return on investment begins. The cost of the devices can be paid upfront, but the team may not actually save time or reduce rework until software is adapted, tested, and workflows are changed. While waiting, the devices still tie up capital.
My view is that AI purchases should be evaluated by first listing the tasks that can be done now, then separately listing the features that have yet to launch. You can’t use the ideal efficiency shown in an entire presentation to calculate returns as if they’ll start materializing on day one for the first batch of machines.
The same goes for AI-related tokens such as RENDER, FET, and TAO: this is product progress from Microsoft and its hardware partners, not automatically orders or revenue for those projects. Technical excitement and the source of your own returns still need to be checked item by item.
The accompanying photo is an archival photograph of a Microsoft office building.
$RENDER $FET $TAO
Tap my profile picture to view my live trade-copying account
The easiest mistake to make when watching a launch event is to assume that everything announced on the same day will all be available that same day.
Microsoft announced on October 7 that the Surface Laptop Ultra, powered by NVIDIA chips, is open for preorder and will go on sale starting October 16. But local model routing on Windows won’t enter experimental preview until later in October, while hybrid AI features for Copilot+ PCs are expected to roll out over the coming months.
These are three separate timelines, and they shouldn’t be collapsed into the claim that “the full experience has already been delivered.” Nor does a preview mean every employee at a company can use it reliably right away.
This affects when the return on investment begins. The cost of the devices can be paid upfront, but the team may not actually save time or reduce rework until software is adapted, tested, and workflows are changed. While waiting, the devices still tie up capital.
My view is that AI purchases should be evaluated by first listing the tasks that can be done now, then separately listing the features that have yet to launch. You can’t use the ideal efficiency shown in an entire presentation to calculate returns as if they’ll start materializing on day one for the first batch of machines.
The same goes for AI-related tokens such as RENDER, FET, and TAO: this is product progress from Microsoft and its hardware partners, not automatically orders or revenue for those projects. Technical excitement and the source of your own returns still need to be checked item by item.
The accompanying photo is an archival photograph of a Microsoft office building.
$RENDER $FET $TAO
Tap my profile picture to view my live trade-copying account