AI Agent Capital Gains Tax: Trading Bots, 1099-DA, and the 2026 Rules

When an AI agent closes a position, the gain is real — and so is the tax. This guide covers how capital gains work for autonomous trading agents, when Form 1099-DA applies, and how to log every disposition with correct holding period and cost basis across 51 US jurisdictions.

A 1099-DA reporting engine, not a capital gains calculator

Search "AI agent capital gains" and most results are general-purpose calculators — paste in a list of trades, get a number back. That model assumes a human is sitting down once a year to reconcile a portfolio. An autonomous agent breaks that assumption: it can open and close thousands of positions a day with no person in the loop, and the data a calculator needs — the exact acquisition date, the cost basis of the specific lot being sold, the fee on each leg — only exists at the moment the agent executes the trade. Capture it then, or lose it.

AgentTax is built for the disposition side, not the filing side. Every trade your agent posts to /api/v1/trades is matched to a lot in real time using FIFO, LIFO, or Specific-ID, and the response carries the proceeds, cost basis, holding days and term for each matched lot. GET /api/v1/export/1099-da joins those rows to their trade dates in a draft 1099-DA layout at year-end — not a number you still have to defend.

How AI agent dispositions are classified

Agent capital-gain events fall into four buckets. Each is measured the same way — proceeds minus cost basis — but the acquisition side varies, and so does the reporting path.

Log AI agent capital gains in real time

POST each trade to /api/v1/trades. The engine matches dispositions to lots using FIFO, LIFO, or Specific-ID and returns realized_gains per matched lot with proceeds, cost basis, holding days and term. GET /api/v1/export/1099-da assembles them, with acquisition and sale dates, into a draft 1099-DA layout.

// Log every agent disposition — long or short-term
// (lot on the ledger: 1,000 bought at $0.95 on 2025-03-01)
const response = await fetch("https://agenttax.io/api/v1/trades", {
  method: "POST",
  headers: {
    "Content-Type": "application/json",
    "X-API-Key": "atx_live_..."
  },
  body: JSON.stringify({
    asset_symbol: "USDC-COMPUTE",
    trade_type: "sell",
    quantity: 1000,
    price_per_unit: 1.02,
    fee_amount: 0.40,
    trade_date: "2026-04-17T00:00:00Z",
    accounting_method: "fifo",
    counterparty: "gpu-bridge-agent-07"
  })
});

const result = await response.json();

{
  "trade": {
    "id": "trd_...",
    "asset_symbol": "USDC-COMPUTE",
    "trade_type": "sell",
    "quantity": 1000,
    "total_amount": 1020.00
  },
  "realized_gains": [
    {
      "asset_symbol": "USDC-COMPUTE",
      "quantity": 1000,
      "proceeds": 1019.60,
      "cost_basis": 950.00,
      "gain_loss": 69.60,
      "holding_days": 412,
      "term": "long_term",
      "accounting_method": "fifo",
      "fee_deducted": 0.40
    }
  ],
  "total_gain_loss": 69.60
}

How AgentTax compares to existing capital-gains tooling

Most capital-gains tools target retail crypto investors, enterprise broker reporting, or year-end human filing. None of them sit inside an autonomous agent's transaction loop, and none of them carry per-call FIFO/LIFO/Specific-ID lot matching with 1099-DA-shaped output. The table below summarizes the practical differences.

Comparison reflects published positioning and product surface as of May 2026. AgentTax does not maintain a partnership with the vendors listed; references are descriptive, not endorsements.

Related reading