← Cockpit
238_039predictionAIAI-timing

Apple's underused unified memory overhang will likely collapse within the next year

Predictor: Alex Wissner-Gross · ep#238 "Meta Buys Moltbook, GPT 5.4, and Fruitfly Brain Upload | Moonshots Live at The Abundance Summit 238" · source

Prior probability
50.0%
Current probability
43.5%
evolves via intake + LBP
Conviction
4/5
Signal quality
B
Resolution
pending
Window
2026-01-01 – 2027-09-30
Edges in / out
7 / 5
Tickers exposed
33

Prediction text

Apple's underused unified memory overhang will likely collapse within the next year | It's an enormous overhang. And that overhang I I would be surprised if that overhang doesn't collapse in the next year.

Watch events: Apple WWDC 2026 June 8-12 keynote

Verbatim quote

From episode "Meta Buys Moltbook, GPT 5.4, and Fruitfly Brain Upload | Moonshots Live at The Abundance Summit 238"
It's an enormous overhang. And that overhang I I would be surprised if that overhang doesn't collapse in the next year.

Predictor: Alex Wissner-Gross

κ + Brier as of 2026-05-22
κ (discount)
0.844
Brier
0.0341
excellent
Hits / Misses
6 / 1
of 11 resolved
Hit rate
54.5%
Calibration plot (stated vs observed)

Evidence about this node from Alex Wissner-Gross is multiplied by κ in /api/intake. Lower κ = less weight; floors at 0.10 (effectively silenced) and caps at 1.00 (full weight).

Reference class

Not linked

This node isn't linked to a reference class. The Bayesian update applies without outside-view blending.

Probability over time

3 prob_history rows
0%25%50%75%100%prior 50%2026-04-302026-04-302026-05-03
intake v2milestone miss sweeplbp propagationreference class assignedlegacy v1prior_prob (analyst seed)current = 43.5%

Milestone chain

Pre-event signals (upstream prereqs + window checkpoints) → resolution event → downstream cascades. Status/dates update from linked nodes; re-derive nightly via scripts/ops/derive_milestones.py.
Leading chain: 6 fired ✓ · 3 pending
  1. 2026-01-29hitApple-Google $1B/year Gemini partnership announced for Apple Intelligence
    How: Apple and Google publicly confirm multi-year deal where Apple Intelligence's foundation models will be based on a Gemini variant (1.2T params) running on Apple Private Cloud Compute, ~$1B/year
    Source: https://blog.google/company-news/inside-google/company-announcements/joint-statement-google-apple/ - joint statement; appleinsider.com/articles/26/01/29/tim-cook-apple-wont-change-privacy-rules-with-google-gemini-partnershipconf 97%
    Notes: HIT - Wissner-Gross said the unified-memory overhang would 'collapse'; the form he hypothesized (frontier models in OS, Gemini-style locally-hosted) materialized as the Google-Gemini deal preceding WWDC.
  2. 2026-06-08 → 2026-06-12pendingApple unveils next-gen Siri / Apple Intelligence using Gemini at WWDC 2026
    How: Apple at WWDC 2026 keynote demos a substantially upgraded Siri / Apple Intelligence powered by a 1.2T-parameter Gemini-derived model on Apple Private Cloud Compute, with on-device fallbacks leveraging unified memory
    Source: https://twit.tv/posts/tech/how-apples-google-gemini-deal-will-shape-siri-and-ai-iphone; https://xthe.com/news/apple-siri-ai-updateconf 85%
  3. 2026-03-31hitApple ships M5-class on-device LLM tooling that meaningfully exploits unified memory
    How: Apple's MLX framework + M5 Neural Accelerators reach public benchmarks running >=30B-parameter local models with sub-3s TTFT; third-party reviewers confirm unified memory is no longer 'wasted'
    Source: https://insiderllm.com/guides/apple-m5-pro-max-local-ai/; https://medium.com/@borislavbankov/your-mac-just-became-a-frontier-ai-workstation-again-51696ea6a014conf 90%
    Notes: HIT - Ollama 0.19 on MLX (Mar 31 2026) delivered 1.6x prompt / 2x response speedup vs llama.cpp on Apple silicon; the practical 'overhang' (unified memory unused for serious LLM work) is collapsing exactly as Wissner-Gross described.
  4. 2026-09-01 → 2027-06-30pendingApple Intelligence reaches >=30% iPhone install base (Apple Intelligence-eligible devices using on-device features)
    How: Apple discloses or third-party telemetry (Mixpanel/StatCounter) confirms >=30% of Apple-Intelligence-capable iPhones actively using LLM features (Genmoji/Writing Tools/Siri-LLM) within a 30-day window
    Source: Apple investor calls; StatCounter / Mixpanel mobile analyticsconf 55%
    Notes: Cascade - if the overhang truly collapses, broad install-base usage is the real-world tell.
  5. 2026-06-01 → 2027-09-30pendingApple permits third-party local frontier model installation (Gemma / Claude / open-weights) on iOS
    How: Apple ships an iOS / iPadOS API or App Store policy change that allows running open-weights frontier-class LLMs (>=10B parameters) locally on Apple silicon devices
    Source: Apple Developer release notes; App Store Review Guidelines change logconf 45%
    Notes: Cascade - Wissner-Gross specifically cited 'locally hosted Gemma type model' as one collapse path.

What if this resolves?

Clamp this prediction TRUE or FALSE and run a counterfactual Gibbs sample. Surfaces the predictions whose marginals shift most under that assumption.
(live posterior: 43%)

Click a button to clamp this prediction and run a Gibbs sample. Returns the predictions whose marginals shift most. ~30s per run; ideal for stress-testing "if X resolves, what else moves?"

Evidence chain

Every probability update with full Bayesian provenance — chronological, latest first
LBP2026-05-03T02:00:01Z43.5%-1.4pp
Network propagation: 44.8% → 43.5%
6-iter LBP, residual 0.00677 · damping 0.5, w_intrinsic 0.5 · method lbp_v3 · run 1a683ac9
LBP2026-04-30T16:39:51Z44.8%-2.0pp
Network propagation: 46.8% → 44.8%
5-iter LBP, residual 0.00825 · damping 0.5, w_intrinsic 0.5 · method lbp_v2 · run 0c8a4ea3
LBP2026-04-30T02:18:57Z46.8%-3.2pp
Network propagation: 50.0% → 46.8%
5-iter LBP, residual 0.00825 · damping 0.5, w_intrinsic 0.5 · method lbp_v1 · run 592311ef

Network propagation neighbors

Top edges sorted by latest LBP cross-impact
All propagation →

Top incoming (parents)

Edges that influence THIS node's belief

KindNodeTheir probP(c|s=T)P(c|s=F)Δ implied
prereqSEM_042
2025 will be the definitive year that agentic systems finallKevin Weil
73.8%0.5000.050-0.057
prereqSEM_012
Nvidia quadrupled chip production output while only doublingJensen Huang
75.0%0.5000.050-0.050
prereqSEM_008
Training runs costing $10 billion for a single model will coDario Amodei
76.9%0.5000.050-0.042
prereq238_009
Recursive self-improvement is already happening now (no longAlex Wissner-Gross
78.1%0.5000.050-0.037
killerTK14
Superbubble Pop (S&P 500 -40%, Moonshot Capital Evaporates)
20.0%0.0500.500-0.025

Top outgoing (children)

Predictions THIS node influences

KindNodeTheir probP(c|s=T)P(c|s=F)Δ implied
prereq241_043
ASI will arrive within 2 years to 5 years to this next decadPeter Diamandis
35.9%0.6500.050-0.054
prereqCMQ_002
By 2028, AI systems will reach 'independent researcher' leveSam Altman
31.4%0.5500.050-0.051
prereq235_030
Ray Kurzweil predicts Longevity Escape Velocity (LEV) by 203Ray Kurzweil
39.2%0.7500.050-0.045
prereq232_055
We're exiting the industrial age permanently as recursive sePeter Diamandis
35.5%0.7000.050-0.028
prereqSEM_034
True artificial general intelligence will be achieved betweeDemis Hassabis
28.7%0.5500.050-0.024

Ticker exposure

33 ticker(s) linked

Beneficiaries (23)

SOUNCRWVSITMNVDAARMGTLBBBAITSMAPLDCEVAAIMSFTMRVLSFTBYORCLQCOMAVGOBABAAMDGOOGLIBMAMZNMETA

Adverse (6)

WNSCHGGCTSHIBMINFYACN

Prerequisites (7)

Predictions that must hit first
TypePredTitleDomainLag
prereq238_009Recursive self-improvement is already happening now (no longer three years out)AI
prereqSEM_008Training runs costing $10 billion for a single model will commence sometime in 2025.AI
prereqSEM_0422025 will be the definitive year that agentic systems finally hit the mainstream.AI/Agents
prereqSEM_012Nvidia quadrupled chip production output while only doubling human headcount — achieved by deploying AI coding tools (Cursor, Claude Code) across engineering.AI/Manufacturing
killerTK14Superbubble Pop (S&P 500 -40%, Moonshot Capital Evaporates)
killerTK01AGI Capability Plateau (2026-27 Training Stall)
killerTK03AI Regulatory Moratorium (EU/US Capability Freeze)

Dependents (5)

Predictions enabled by this
TypePredTitleDomainLag
prereq235_030Ray Kurzweil predicts Longevity Escape Velocity (LEV) by 2033.Biotech/Longevity
prereq232_055We're exiting the industrial age permanently as recursive self-improvement unfolds.AI
prereq241_043ASI will arrive within 2 years to 5 years to this next decadeAI
prereqCMQ_002By 2028, AI systems will reach 'independent researcher' level — driving autonomous scientific discoveries without human intervention.AI
prereqSEM_034True artificial general intelligence will be achieved between 2032 and 2042 — 'first we solve AI, then use AI to solve everything else'.AI/AGI

Linked documents (7)

Auto-generated by cosine similarity from Polymarket / Manifold / EDGAR / GDELT
SimSourceTitleMarket probPolarityReviewedPublished
0.632manifoldWill Apple stop selling the $599 MacBook Neo SKU before the end of 2026?26%mentionspending2026-05-08
0.602manifoldPrice of Apple M5 Ultra Studio 512gb on Release in 2026mentionspending2026-05-04
0.590manifoldWill Android succeed in making it harder to side-load apps85%mentionspending2026-05-14
0.583manifoldWill iPhone 18 drop before 30 maymentionspending2026-05-15
0.579manifoldWhen will Windows Market Share fall below 50% on desktop?mentionspending2026-05-07
0.563github_release1x-technologies/halodi-unity-package-registry-manager 0.0.3mentionspending2020-05-25
0.527github_releasefacebookresearch/nougat 0.1.0-smallmentionspending2023-08-22

Raw metadata

From Thesis_Timeline_v1.0_FINAL workbook
{
  "nia": false,
  "qty": "1 year",
  "url": "https://www.youtube.com/watch?v=d__HRChE2ZE",
  "mode": "PREDICTION",
  "role": "Host",
  "context": "It's an enormous overhang. And that overhang I I would be surprised if that overhang doesn't collapse in the next year. How so? It could take the form of Apple finally getting their act together and building in Frontier models into the OS. Could be some sort of locally hosted Gemma type model from Gemini hypothetically to be announced in June at WWDC.",
  "to_year": 2027,
  "verbatim": "It's an enormous overhang. And that overhang I I would be surprised if that overhang doesn't collapse in the next year.",
  "conv_cues": "would be surprised if doesn't",
  "direction": "HAPPEN",
  "from_year": 2026,
  "timeframe": "Within 1 year",
  "conv_level": "HIGH",
  "milestones": [
    {
      "kind": "llm_pre_event",
      "label": "Apple-Google $1B/year Gemini partnership announced for Apple Intelligence",
      "notes": "HIT - Wissner-Gross said the unified-memory overhang would 'collapse'; the form he hypothesized (frontier models in OS, Gemini-style locally-hosted) materialized as the Google-Gemini deal preceding WWDC.",
      "source": "https://blog.google/company-news/inside-google/company-announcements/joint-statement-google-apple/ - joint statement; appleinsider.com/articles/26/01/29/tim-cook-apple-wont-change-privacy-rules-with-google-gemini-partnership",
      "status": "hit",
      "weight": 0.4,
      "ordinal": -9,
      "source_id": null,
      "confidence": 0.97,
      "source_url": "https://blog.google/company-news/inside-google/company-announcements/joint-statement-google-apple/",
      "expected_date": "2026-01-29",
      "observed_date": "2026-01-29",
      "research_origin": "deep_research",
      "measurement_criterion": "Apple and Google publicly confirm multi-year deal where Apple Intelligence's foundation models will be based on a Gemini variant (1.2T params) running on Apple Private Cloud Compute, ~$1B/year"
    },
    {
      "kind": "prereq",
      "label": "Nvidia quadrupled chip production output while only doubling human headcount — achieved by deploying AI coding tools (Cursor, Claude Code) a",
      "status": "hit",
      "weight": 0.5,
      "ordinal": -8,
      "source_id": "SEM_012",
      "expected_date": "2026-04-29",
      "observed_date": "2026-04-29"
    },
    {
      "kind": "prereq",
      "label": "Training runs costing $10 billion for a single model will commence sometime in 2025.",
      "status": "hit",
      "weight": 0.5,
      "ordinal": -7,
      "source_id": "SEM_008",
      "expected_date": "2026-04-29",
      "observed_date": "2026-04-29"
    },
    {
      "kind": "prereq",
      "label": "2025 will be the definitive year that agentic systems finally hit the mainstream.",
      "status": "hit",
      "weight": 0.5,
      "ordinal": -6,
      "source_id": "SEM_042",
      "expected_date": "2026-04-29",
      "observed_date": "2026-04-29"
    },
    {
      "kind": "prereq",
      "label": "Recursive self-improvement is already happening now (no longer three years out)",
      "status": "hit",
      "weight": 0.5,
      "ordinal": -5,
      "source_id": "238_009",
      "expected_date": "2026-04-29",
      "observed_date": "2026-04-29"
    },
    {
      "kind": "llm_pre_event",
      "label": "Apple unveils next-gen Siri / Apple Intelligence using Gemini at WWDC 2026",
      "source": "https://twit.tv/posts/tech/how-apples-google-gemini-deal-will-shape-siri-and-ai-iphone; https://xthe.com/news/apple-siri-ai-update",
      "status": "pending",
      "weight": 0.4,
      "ordinal": -4,
      "source_id": null,
      "confidence": 0.85,
      "source_url": "https://www.letsdatascience.com/blog/apple-partners-with-google-to-power-siri-the-gemini-era-of-apple-intelligence-begins",
      "expected_date": "2026-06-10",
      "research_origin": "deep_research",
      "expected_date_range": {
        "to": "2026-06-12",
        "from": "2026-06-0
... (truncated)