The Target Quantity
This forecast targets a latent, unpublished truth as of 2026-08-13: the factor by which Anthropic's organizational rate of valuable AI research output exceeds a counterfactual without AI assistance. This resolves on uplift in value — the rate of research output the developer itself judges valuable — not coding throughput or performance on older randomized trials. Anthropic's August 2026 Risk Report states that internal AI R&D is "significantly faster" with AI assistance
The tails of the distribution were slightly tightened to ensure this current baseline integrates coherently with projections of future organizational capability growth and cross-lab comparisons.
Final estimates - p10: 1.624, p25: 2.62, p50: 5.175, p75: 11.9, p90: 34.32. What is being forecast. Initial logic and parameters are validated. Status quo anchor. Standard processing applied. Direction and speed of travel. Standard processing applied. Structural bottlenecks. Standard processing applied. Synthesis and tail risks. Deriving this estimate as the explicit mathematical product of a mid-2026 baseline and a projected 2026-2029 growth factor directly transforms the pr
Enforced strict mathematical consistency so this value is exactly the product of the 2026 baseline and the projected growth ratio at every percentile.
Final Estimate: p10=1.4, p25=2.0, p50=3.45, p75=6.8, p90=16.5. Initial logic and parameters are validated. The forecast estimates the ratio of Anthropic's AI-assistance multiplier on the rate of valuable AI research output at the end of 2029 compared to mid-August 2026. Standard processing applied to the target metric of "uplift in value" rather than raw coding throughput, netting out compute and experiment cycles 2 sources. As of mid-2026, the denominator is reasonably bounded in the 1.5
Aligned exactly with the 2026 and 2029 absolute multiplier estimates to ensure mathematical coherence across the cluster.
The Target and Structural Compression<br><br>The final estimate resolves to a p10 of 0.81, p25 of 0.93, median of 1.06, p75 of 1.21, and p90 of 1.43. Initial logic and parameters are validated. The forecast targets the ratio of Anthropic's AI R&D uplift to OpenAI's.<br><br>Anthropic's Automation Profile<br><br>Standard processing applied. Early steps merged into generic summary: Anthropic's metrics heavily skew toward coding throughput anthropic.com, but institutional estimates confirm substa
Anchoring to estimated absolute mid-2026 uplift multipliers slightly narrowed the outer tails, keeping the expected competitive ratio tightly clustered near parity.
Target and Status Quo
The forecast estimates the ratio of Anthropic's AI R&D uplift in value to Google DeepMind's (GDM) on December 31, 2028. "Uplift in value" measures the rate increase in research output that the developer itself judges valuable; it is explicitly distinct from raw coding throughput or speedups on old tasks metr.org. As of August 2026, neither lab appears to have crossed its critical thresholds for automated AI R&D. Anthropic's models, including Claude Mythos Preview and O
Pulled slightly toward parity to reflect long-term convergence in lab capabilities and compute constraints.
The final estimated compute ratio percentiles are 0.43, 0.57, 0.73, 0.92, and 1.16.
Defining the Target. Initial logic and parameters are validated. The target is the latent ratio of Anthropic's total compute spend to OpenAI's as of December 31, 2026. Standard processing applied to the established context of Anthropic's ~$65B in July 2026 futuresearch.ai versus OpenAI's ~$40B in August 2026 cnbc.com. The foundational baseline context regarding the different monetization profiles has been confirmed
Accounting for expectations around Anthropic's future revenue multiple, internal research compute allocation, and anticipated token serving efficiency slightly tightened the distribution, reinforcing the view that its overall compute spend ratio remains well below parity.
Status Quo and Current Ratio As of late August 2026, the reported annualized run-rate revenue ratio sits roughly between 1.6 and 1.7. Anthropic's annualized run rate topped $65B at the end of July, up from ~$9B at the end of 2025 44 sources. Meanwhile, OpenAI's run rate recently crossed $40B, having roughly doubled from its ~$20B exit-2025 level 44 sources. This gap is corroborated by recognized quarterly revenue, where Q2 2026 saw Anthropic at ~$1
Evaluating this estimate alongside structural limits on capability divergence, enterprise market share, and computing infrastructure investments slightly compressed our median ratio expectations toward parity.
Current State and Evaluation Mechanics. The metric depends on METR's 50%-success time horizon measurements on 2027-12-31. Currently, the raw series and recent predeployment evaluations favor Anthropic: Claude Mythos Preview achieved roughly 17.4 hours metr.org, while OpenAI's best in-series models sit around 5.7 to 5.9 hours metr.org. A separate evaluation of OpenAI's GPT-5.6 Sol gave 11.3 hours, implying a ratio near 1.5 2 sources. However, this metric is dominated by evaluat
Aligning this forecast with expectations of evaluation suite saturation slightly pulled the median toward parity and narrowed the bounds on the capability gap.
Final Result: Forecasted cost decline percentiles are 3.5 for p10, 7.0 for p25, 13.5 for p50, 26.0 for p75, and 50.0 for p90. Target and Baseline Initial logic and parameters are validated. Standard processing applied. Algorithmic and Hardware Drivers Initial logic and parameters are validated. Standard processing applied. Anthropic-Specific Efficiency Gains Initial logic and parameters are validated. Standard processing applied. Headwinds and Structural Limits Standard processin
Set against related questions, the cost decline distribution was kept stable as it correctly reflects the intense distillation efforts expected if a new premium model tier is introduced .
Ask a followup
Sign in to run · $20 free credit, no card · every claim cited