Cost of AI tokens per student ($10k/year) will come down by a factor of 10 and move on-device.

Cost of AI tokens per student ($10k/year) will come down by a factor of 10 and move on-device. | 10 grand a kid is a lot of token usage, though. That'll come down a factor of 10. But it's impressive that you're >> we're going to get it down to on device.

10 grand a kid is a lot of token usage, though. That'll come down a factor of 10. But it's impressive that you're >> we're going to get it down to on device.

Beneficiaries (23)

Adverse (6)

{
  "nia": false,
  "qty": "10x reduction",
  "url": "https://www.youtube.com/watch?v=X94eBT-VZnc",
  "mode": "FORECAST",
  "role": "Guest-CEO",
  "context": "10 grand a kid is a lot of token usage, though. That'll come down a factor of 10. But it's impressive that you're >> we're going to get it down to on device. But that is our And to be honest, like 3 years ago when we started when I became principal, we literally had humans reviewing the video at night annotating",
  "to_year": 2030,
  "verbatim": "10 grand a kid is a lot of token usage, though. That'll come down a factor of 10. But it's impressive that you're >> we're going to get it down to on device.",
  "conv_cues": "that'll come down; we're going to get it down",
  "direction": "DOWN",
  "from_year": 2026,
  "timeframe": "future (unspecified)",
  "conv_level": "HIGH",
  "milestones": [
    {
      "kind": "llm_pre_event",
      "label": "Inference cost per million tokens drops 50x for GPT-4-class capability (2022-2026)",
      "notes": "HIT — 50x decline already realized for GPT-4-class. Liemandt's 10x reduction target already exceeded for capability-matched inference.",
      "source": "https://blogs.nvidia.com/blog/lowest-token-cost-ai-factories/",
      "status": "hit",
      "weight": 0.4,
      "ordinal": -10,
      "source_id": null,
      "confidence": 0.95,
      "source_url": "https://blogs.nvidia.com/blog/lowest-token-cost-ai-factories/",
      "expected_date": "2026-03-31",
      "observed_date": "2026-03-31",
      "research_origin": "deep_research",
      "measurement_criterion": "GPT-4-equivalent capability cost falls from $20/1M tokens (2022) to $0.40/1M tokens (2026) per Stanford AI Index"
    },
    {
      "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": -9,
      "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": -8,
      "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": -7,
      "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": -6,
      "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": -5,
      "source_id": "SEM_042",
      "expected_date": "2026-04-29",
      "observed_date": "2026-04-29"
    },
    {
      "kind": "llm_pre_event",
      "label": "First credible on-device education-grade LLM (sub-7B params) achieves curriculum-tutor quality",
      "source": "https://www.spheron.network/blog/ai-inference-cost-economics-2026/",
      "status": "pending",
      "weight": 0.4,
      "ordinal": -4,
      "source_id": null,
      "confidence": 0.7,
      "source_url": "https://www.spheron.network/blog/ai-inference-cost-economics-2026/",
      "expected_date": "2026-12-15",
      "research_origin": "deep_research",
      "expected_date_range": {
        "to": "2027-06-30",
        "from": "2026-06-01"
      },
      "measurement_criterion": "Sub-7B-parameter LLM running on commodity device (Mac, iPhone, Pixel, Snapdragon) demonstrates K-12 tut
... (truncated)