AGI Countdown Update: April 2, 2026
Three of the world's top AI labs are operating on internal AGI timelines of 2027-2028. Not as a moonshot. As a planning assumption.
🚨 MAJOR NEW ANALYSIS: “INTERNAL MEMO WALL STREET HASN’T SEEN”
Source: Markaicode AGI Timeline Tracker (April 2026)
HEADLINE: Three of the world’s top AI labs are operating on internal AGI timelines of 2027-2028. Not as a moonshot. As a planning assumption.
📊 KEY FINDINGS
1. ARC-AGI BENCHMARK: THE VERTICAL WALL
Performance trajectory:
- 2023: 0%
- Early 2024: 4%
- December 2024: 85.3%
This is not a trend line. This is a vertical wall.
2. BENCHMARK SATURATION ACCELERATING
Time from “AGI-hard” to expert performance:
- 2017-2020 cohort: ~78 months (6.5 years)
- 2024-2025 cohort: under 14 months
We are consuming our measuring sticks faster than we can build new ones.
3. RESEARCHER PREDICTION INTERVALS COLLAPSING
Median predictions for “high-level machine intelligence”:
- 2022 survey: 2059
- 2024 survey: 2047
- Late 2025 informal (frontier lab researchers): 2030 median, 2027-2028 lower bound
The people building AGI think it’s coming faster than the people studying AI from the outside do.
4. THREE MECHANISMS DRIVING PROGRESS
Mechanism 1: Recursive Self-Improvement Signal
- AI now designs better AI training runs
- AI-assisted researcher productivity: 4-7x on subtasks
- If productivity compounds at 2x/year → 2027 output = 8x 2024
Mechanism 2: Capability Concentration
- Top 100 AI researchers at frontier labs: 22% (2019) → 71% (2025)
- Breakthroughs don’t diffuse over years → implemented in weeks
Mechanism 3: Economic Pressure Ratchet
- OpenAI compute spend: $7B+ (2025)
- Anthropic raised $7B in single round
- Google: $75B AI infrastructure investment (2026)
- No rational actor can afford to slow down unilaterally
5. THREE SCENARIOS FOR 2027-2030
| Scenario | Probability | Key Features |
|---|---|---|
| Controlled Emergence | 25% | AGI in 2027-2028, voluntary coordination, phased disruption |
| Race to Deployment (BASE CASE) | 55% | AGI-adjacent systems by late 2027, no coordination, reactive regulation |
| Capability Wall/Safety Event | 20% | Major incident triggers 18-36 month pause, AGI pushed to 2032+ |
6. ANTHROPIC SECURITY INCIDENT (March 31, 2026)
News: Anthropic leaked Claude Code source code (~500,000 lines)
- Second security incident in 5 days
- First incident: “Mythos” model revealed accidentally
- Concerns: Competitors could reverse-engineer agentic harness
- Not directly timeline-relevant, but raises questions about safety practices
📈 UPDATED TIMELINE COMPARISON
| Source | Timeline | Confidence | Notes |
|---|---|---|---|
| Internal Lab Planning | 2027-2028 | High | NEW: Planning assumption |
| Jensen Huang (NVIDIA) | NOW | High | ”We’ve achieved AGI” |
| Stanford 6 | NOW (key tasks) | High | Human-level in key areas |
| Sequoia | 2026 | High | ”This is AGI” |
| Elon Musk | End 2026 | Aggressive | ”Surpass humans” |
| Sam Altman | 2027-2028 | High | Confirmed |
| Dario Amodei | 1-3 years | 90% | Verifiable tasks first |
| Demis Hassabis | 3-5 years | Measured | |
| Scholars | 2040-2050 | Conservative |
🔥 CRITICAL INSIGHT: THE 24-MONTH WINDOW
“If current capability trajectories hold, and if the economic pressure ratchet doesn’t break, by late 2028 we’ll be looking at systems that can perform the majority of current knowledge work at costs approaching zero marginal expense.”
“The only historical precedent for that kind of labor market shock is the Industrial Revolution — and that transition took 80 years. We have, at current trajectory, roughly 24 months.”
✅ HOMEPAGE STATUS
No changes required. Current “~1 year” estimate is aligned with the new analysis showing labs operating on 2027-2028 internal timelines with some capabilities already here.
The Markaicode analysis provides strong supporting evidence for the aggressive timelines we’re already displaying.
📨 SUMMARY FOR CJ
MAJOR NEW ANALYSIS (April 2026):
🎯 “Internal Memo Wall Street Hasn’t Seen” — Markaicode analysis reveals:
- Three top AI labs operating on 2027-2028 AGI timelines as planning assumptions
- ARC-AGI benchmark: 0% → 85.3% in under 2 years (vertical wall)
- Benchmark saturation time: 78 months (2017) → 14 months (2025)
- Frontier lab researchers: 2027-2028 as credible lower bound
🎯 Three scenarios with probabilities:
- 55% probability: “Race to Deployment” — AGI-adjacent systems by late 2027
- 25%: Controlled Emergence (coordinated pause)
- 20%: Safety Event (delay to 2032+)
🎯 Economic pressure ratchet:
- Labs can’t slow down unilaterally without losing the race
- $7B+ annual compute spend creates unstoppable momentum
🎯 24-month warning:
- Current trajectory: knowledge work at near-zero marginal cost by late 2028
- Historical precedent (Industrial Revolution) took 80 years
- We have ~24 months
HOMEPAGE STATUS: No changes needed. The “~1 year” estimate and NVIDIA/Stanford signals align with the new analysis showing 2027-2028 as the planning assumption at top labs.
Generated: April 2, 2026, 6:00 AM NZT Source: Automated AGI Prediction Monitor