When will any company achieve AGI?

**Stage 1 — Blind analysis (ignore current market price):**

- Context and framing: The question is strictly about a *company announcement* claiming AGI, not about community acceptance, independent verification, or incontrovertible demonstration. That distinction is critical because it changes the event from a technical milestone to a mixed technical/strategic/PR decision. With ~5.5 months until the cutoff, there is limited time for surprising training runs, but plenty of time for high-profile labs to decide to make a claim based on internal metrics, demos, or strategic calculus.

- Base-rate intuition and calibration: Historically, major proclaimed paradigm shifts in machine learning have been announced months or years after key internal milestones; however, proclamations of 'AGI' are socially costly and rare. For a short ~half-year horizon, that suggests a low baseline probability. I start with a conservative short-run baseline (~8–12%) to reflect the historical difficulty and conservative incentives.

- Upward adjustments (why the probability is higher than that baseline): several high-leverage facts push the probability substantially up: - **Strong public claims from frontier leaders** (Altman: "basically built it"; Musk: by end of 2026; Amodei: late 2026–2027) increase the chance of an announcement because they reveal either internal confidence or strategic willingness to claim imminence. - **Incentives to claim**: marketing, investment, hiring, geopolitical positioning, and the desire to set the narrative create real incentives for a firm to assert AGI first — especially private labs that benefit from headlines and fundraising. - **Ambiguous definition**: lack of a community standard means firms can opt for a broad, fuzzy definition and declare AGI in a way designed to be difficult to conclusively refute. - **Precedent of contested claims**: in other tech domains, firms have made aggressive claims that were disputed but counted as announcements (and that is sufficient to satisfy the market question).

- Downward adjustments (reasons to reduce the baseline): - **Reputational and regulatory risk**: an erroneous or premature 'AGI' claim could damage credibility and invite regulatory attention or legal risk; large incumbents (Microsoft, DeepMind) appear risk-averse. - **Technical uncertainty remains**: aggregated forecasts and many leading researchers still place AGI beyond 2027; the median of many formal surveys is well past the cutoff. - **Internal conservatism/tactical silence**: some firms may prefer to show incremental capabilities rather than make a single headline claim.

- Quantitative synthesis to arrive at an independent probability (transparent additive reasoning): - Baseline short-run chance: 10% - Add for frontier public claims and incentives: +20% (because at least two influential CEOs publicly signalling imminent capability materially raises the odds of a firm making a public claim in the short window) - Subtract for reputational/regulatory friction and conservative players: -5% - Net: 25% — then apply a modest upward squeeze for the low-verification nature of an "announcement" (companies can craft ambiguous claims): +10% - Final independent assessment: 35% (rounded; reflects that an announcement is substantially easier than a widely accepted, verified AGI).

- Qualitative judgement summary: I view a ~35% chance as balanced: significantly above negligible because of vocal, confident actors and weak verification standards, but still below coin-flip because of large technical uncertainty and strong disincentives against being wrong.

**Stage 2 — Market calibration (look at current market prices and explain differences):**

- Current market pricing: Yes = 0.10, No = 0.90.

- Why the market may be undervaluing the event (arguments for market mispricing): - The market appears to price this more like a strict technical breakthrough that would withstand broad scrutiny, rather than a public announcement. Because the contract question is only about an announcement, the market likely underweights the possibility of *strategic, ambiguous, or contested* proclamations. - High-profile public statements from Altman and Musk are recent and salient; markets sometimes underreact to rhetoric that signals internal confidence/intent rather than pure technical progress. - Liquidity and concentration effects: with concentrated volume and perhaps informed participants, temporary price suppression could reflect a few large sellers rather than the aggregated informed probability.

- Why the market might be correct or even too optimistic about No (i.e., why my higher probability could be wrong): - The market likely captures the very real reputational/regulatory costs that deter credible firms from making a premature AGI claim. Large incumbent firms (Microsoft, DeepMind) have strong reputational capital and appear unlikely to make an unambiguously false or reckless claim. - Aggregated expert forecasts (Metaculus, community medians) put AGI well after 2027; the market may be incorporating a broad expert consensus more heavily than my calculation.

- Net calibration conclusion and trading implication: The market price of 10% is materially below my independent estimate of 35%. I judge the market is likely underpricing the probability of a company-level announcement because it conflates announcement with community-accepted AGI and underweights the incentives and public statements that make an announcement easier than full acceptance. This suggests a potential mispricing opportunity for those who agree with my view, but it is not an extreme arbitrage — the disagreement centers on how much weight to place on PR incentives versus reputational/regulatory friction.