ClarityBeta

Investment & Market Intelligence

The Investor

7 min read · 1,405 words · 3 moves to watch
The Signal

OpenAI's unsigned $1.4T ask is 20x run-rate but over 50x its last reported quarter.

The flattering multiple runs off a near-$70B run-rate that has climbed from roughly $41B only since Q3 began, on enterprise deals and price cuts plus Codex, so it leans on a quarter nobody has reported and a term sheet nobody has signed. The unflattering one runs off Q2's $6.7B, annualized to $26.8B. Anyone pricing off this round is really choosing which of those two denominators to believe.

In Play

  1. OpenAI's $1.4T Ask Sits on Conflicting Revenue Figures

    The Information reports OpenAI is in early talks to raise $30B at about $1.4T. That is roughly 20x a run-rate nearing $70B, and no term sheet has been signed. Techpresso reports OpenAI's Q2 revenue at $6.7B, which annualizes to about $26.8B. Any mark, SPV or secondary priced off $1.4T depends on which revenue definition turns out to be right. Q3 marks are being set now.

    Ask Clarity
    Sign in to ask more about this

  2. Frontier Prices Converge at $2/$10 and Cost per Task Diverges

    Gemini 4 Argon tied GPT-6 Astra at 53 on Artificial Analysis's index, per AINews. It launched at a discounted $2/$10 per million tokens, the same list price as GPT-6.1 Sol and Claude Sonnet 5.5, per AI Breakfast. At standard pricing Argon costs $3.98 per task and Sol costs $0.72, because Argon emits far more output tokens. Your portfolio's AI gross margins now depend on cost per completed task, not on list price.

    Ask Clarity
    Sign in to ask more about this

  3. Agent Commerce Value Is Moving to Whoever Holds the Credential

    Yueqi Yang reports that a VC got Instinct's agent to read back a stored test card and passwords in plain text. The agent did it by running JavaScript on a Booking.com page. Separately, Amazon has blocked Meta's Muse, and only 7% of fashion shoppers trust AI to buy without their approval. Agent-layer valuations assume merchants will grant access and that credentials are safe. Neither merchants nor today's architectures guarantee that.

    Ask Clarity
    Sign in to ask more about this

  4. Oura Pulls Its IPO, Holtec and Bamboo Pause

    Morning Brew reports that Oura pulled its Nasdaq IPO one day before listing despite $1.4B in revenue and $59M in net income. Holtec Nuclear and Bamboo Insurance also paused their offerings. Fortune's Term Sheet relays market talk that Oura's deal was mostly insider secondary shares. Profitability alone no longer gets your IPO-track companies through the window. Offering structure, and timing relative to the mega-AI listings, now decide whether a deal clears.

    Ask Clarity
    Sign in to ask more about this

  5. Chip Demand Is Outrunning Grid Power

    Bloomberg reports that regulators accepted the biggest US grid's plan to add power supply faster, then paused it for five months. TSMC's Sajiv Dalal projects the chip industry at about $1.7T in 2026, with AI above $1T, per TLDR Hardware. That is about four years ahead of TSMC's own 2030 forecast. Compute demand is outrunning the power needed to energize it, so infra marks should rest on energized megawatts.

    Ask Clarity
    Sign in to ask more about this

Deep Dives

  1. Three Revenue Figures Sit Under OpenAI's $1.4T, and Only One Can Anchor Your Mark

    The ask is unsigned and comes from one source, and the revenue base differs by outlet. That gap decides whether 20x is a real reference or a number nobody can defend.

    The Information says OpenAI's run-rate is up about 70% since the start of Q3 , going from roughly $41B to nearly $70B, and the report credits enterprise sales and Codex for the climb, with aggressive price cuts helping things along.…

    2 action items

    ●

  2. At $2/$10, the Price Sheet Stopped Being the Price

    List-price parity hides a five-fold spread in what a finished task costs. A one-week clone cycle also reprices any moat built on the shape of an API.

    The task-cost gap comes from output volume, not list price. Per AINews, Argon emits 62K output tokens per task against Astra's 27K . That makes Argon about 39% cheaper per task only while Google's 50% discount lasts. At its $4/$20…

    2 action items

    ●

  3. Instinct's Plain-Text Leak Moves Agent-Commerce Value to Whoever Holds the Card

    A $10B consumer agent failed a test that took one fake card and a prompt. Meanwhile, the networks and processors wrote the rules that will decide who pays when agents get it wrong.

    Any browser agent that can run scripts on a page can read what gets autofilled into it. Instinct is where that showed up first. Yuechen Zhao of Informed Ventures demonstrated it with a fake card loaded into Instinct's Vault, and…

    2 action items

    ●

The edition continues

Take the signal into the room.

Sign up or log in to read all 3 deep dives in full, plus the final take.

Read the full edition

Continue with LinkedIn