Within 1-2 years we will look back and wonder why training was centralized while inference was decentralized

Within 1-2 years we will look back and wonder why training was centralized while inference was decentralized | I I I suspect a year from now, two years from now, we'll look back and we'll wonder why exactly is it that or maybe royal we other countries may look back and wonder why was training so centralized all the while inference time was so decentralized.

I I I suspect a year from now, two years from now, we'll look back and we'll wonder why exactly is it that or maybe royal we other countries may look back and wonder why was training so centralized all the while inference time was so decentralized.

Beneficiaries (23)

Adverse (6)

{
  "nia": false,
  "qty": "1-2 years",
  "url": "https://www.youtube.com/watch?v=dmtvGKuRE64",
  "mode": "PREDICTION",
  "role": "Host",
  "context": "On the other hand, in in some sort of perverse I I think geopolitical sense, the training time is where all of the values or the majority of the values are ultimately instilled. Training time sort of puts the foundation in place. At inference time, you can put in system prompts. You can put in other guard rails. But I I I suspect a year from now, two years from now, we'll look back and we'll wonder why exactly is it that or maybe royal we other countries may look back and wonder why was training so centralized all the while inference time was so decentralized.",
  "to_year": 2028,
  "verbatim": "I I I suspect a year from now, two years from now, we'll look back and we'll wonder why exactly is it that or maybe royal we other countries may look back and wonder why was training so centralized all the while inference time was so decentralized.",
  "conv_cues": "I suspect",
  "direction": "HAPPEN",
  "from_year": 2027,
  "timeframe": "2027-2028",
  "conv_level": "MEDIUM",
  "milestones": [
    {
      "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": -10,
      "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": -9,
      "source_id": "SEM_008",
      "expected_date": "2026-04-29",
      "observed_date": "2026-04-29"
    },
    {
      "kind": "prereq",
      "label": "Anthropic revenue will cross OpenAI revenue in middle of 2026",
      "status": "hit",
      "weight": 0.5,
      "ordinal": -8,
      "source_id": "234_012",
      "expected_date": "2026-04-29",
      "observed_date": "2026-04-29"
    },
    {
      "kind": "prereq",
      "label": "Anthropic will exceed OpenAI in revenue this year (2026).",
      "status": "hit",
      "weight": 0.5,
      "ordinal": -7,
      "source_id": "235_002",
      "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": "llm_pre_event",
      "label": "First open-source globally distributed RL training run of >=100B parameter model completes successfully",
      "source": "Prime Intellect INTELLECT-2 announcement; Nous Research $50M Series A from Paradigm",
      "status": "pending",
      "weight": 0.4,
      "ordinal": -5,
      "source_id": null,
      "confidence": 0.55,
      "source_url": "https://www.primeintellect.ai/blog/intellect-2",
      "expected_date": "2026-11-29",
      "research_origin": "deep_research",
      "expected_date_range": {
        "to": "2027-06-30",
        "from": "2026-05-01"
      },
      "measurement_criterion": "Prime Intellect, Nous Research, or equivalent publishes weights or technical report for a >=100B-parameter model whose training was geographically distributed across multiple sovereign jurisdictions; INTELLECT-2 (32B) was the prior milestone"
    },
    {
      "kind": "llm_pre_event",
      "label": "Non-US sovereign nation announces national AI training compute initiative (>=10K H100-equivalent GPUs) outside Five Eyes / EU",
      "source": "BCG 2026 Sovereign AI report; UAE-Colossal $60M precedent",
      "status": "pending",
      "weight": 0.4,
      "ordinal": -4,
      "source_id": null,
      "confidence": 0.65,
      "expected_
... (truncated)