OpenAI wants to sell you a box that sits on your counter. The box will run a voice assistant. The voice assistant will call OpenAI's servers. The servers will respond. OpenAI will pay for that compute. You will pay, or you will not, and OpenAI will figure out the math later.

That is the product. Not a revolution. A speaker with a microphone and a network call that lands on infrastructure the company has not yet made profitable.

The smart speaker market is not a growth opportunity. Amazon and Google built it, priced their hardware at cost or below, and still could not turn it into a business. Amazon's Alexa division was reportedly losing over $10 billion per year before the cuts started. That is the market OpenAI is entering.

The Object Gate stops here. No part number. No die. No watt-per-inference figure. No cost-of-goods estimate. No yield number. No carrier deal. Nothing from the working layer, only the slide layer.

What is known: every query the speaker sends costs OpenAI money to answer. What is unknown: what the speaker costs to build, ship, warranty, and support. What is certain: the sum of those numbers must be smaller than the revenue per device, and no one has shown that math.

The machine that runs is interesting. The press release about the machine is not. File again when there is a part number.