AI bubble talk is cheap until it meets actual financing documents. OpenRouter's release says the company raised $113 million in a CapitalG-led Series B while weekly volume reached 25 trillion tokens, which gives the weekend's optimism a concrete counterparty and a concrete usage claim [2].

That does not make every AI valuation sound. It does mean the bearish argument has to separate companies with only mood from companies with disclosed customers, volume, revenue, or cash. OpenRouter is still a venture bet, but it is not an anonymous one [2].

Lionsgate's release offers a useful non-AI discipline. Its business case rests on library revenue above $1 billion over the trailing twelve months and positive quarterly free cash flow. Those numbers are not comparable to token volume, but they remind readers what mature evidence looks like [1].

Dentro's timeline captures why the bubble debate feels overheated: launches, subscriptions, benchmarks, and funding all arrived in the same week [3]. The right response is not euphoria or reflexive dismissal. It is source sorting. Which companies disclose volume, revenue, contracts, or cash, and which merely borrow the AI multiple. The receipts decide which story survives.

The funding receipt supports caution without settling the bubble question. [2]

-- THEO KAPLAN, San Francisco