Tesla's Bitcoin 10-K: Feb 8, 2021 on the Tape

On the morning of February 8, 2021, Tesla disclosed that it had bought $1.5 billion of bitcoin and intended to accept it as payment, not in a press release or a flashy 8-K, but inside the risk-factors-and-everything-else text of its annual report. It was the moment corporate-treasury crypto went from meme to filed fact. This page receipts the whole arc: the 10-K, the session, the crypto-adjacent stocks that moved an order of magnitude more, the options tape that shrugged, and the 2022 disclosure that sent most of the bitcoin back out the door. Every number is a stored query; expand any panel for the SQL.

The filing, receipted

WITH (
    SELECT round(toFloat64(cash_and_equivalents) / 1e9, 1)
    FROM global_markets.stocks_balance_sheets
    WHERE has(tickers, 'TSLA') AND timeframe = 'annual' AND fiscal_year = 2020
    LIMIT 1
) AS yearend_cash_bn
SELECT
    count() AS tesla_filings_that_day,
    countIf(form_type = '10-K') AS annual_reports,
    any(form_type) AS form_type_on_file,
    yearend_cash_bn AS fy2020_cash_bn,
    round(100 * 1.5 / yearend_cash_bn, 1) AS purchase_pct_of_cash
FROM global_markets.stocks_sec_edgar_index
WHERE filing_date = '2021-02-08'
  AND (issuer_name ILIKE '%tesla%' OR cik = '0001318605')

One filing on the date: a 10-K, the annual report. The bitcoin disclosure rode along inside the 10-K's risk factors and subsequent-events text, a detail that matters for how the news traveled: annual reports are hundreds of pages, and the market's reaction began as readers found the paragraph, not at a scheduled headline moment. The SEC filings guide covers what each form is for, and why the 10-K is where the bodies are buried.

The filing also stated the why, in treasury language: Tesla had updated its investment policy in January 2021 seeking "more flexibility to further diversify and maximize returns" on cash it did not need for operating liquidity, and under that policy it "invested an aggregate $1.50 billion in bitcoin", adding that it expected "to begin accepting bitcoin as a form of payment for our products in the near future." One paragraph, two commitments, a treasury position and a payments experiment, each with an aftermath below.

What Tesla actually bought

The 10-K discloses dollars, not coins: $1.50 billion, with no coin count anywhere in the filing. Every "Tesla bought roughly this many bitcoins" figure you have read is an outside estimate, the dollar amount divided by that window's bitcoin price, and since bitcoin's own market sits outside this warehouse's licensed universe, this page will not manufacture one. What the filing record does support is scale: against the $19.4 billion of cash and equivalents on Tesla's year-end 2020 balance sheet (receipted above), the purchase came to 7.7%, a real allocation, nowhere near a bet-the-company number.

The day, on one row

WITH
    (
        SELECT argMaxIf(toFloat64(close), window_start, (toHour(toTimeZone(window_start, 'America/New_York')) * 60 + toMinute(toTimeZone(window_start, 'America/New_York'))) BETWEEN 570 AND 959)
        FROM global_markets.delayed_stocks_minute_aggs
        WHERE ticker = 'TSLA'
          AND window_start >= toDateTime('2021-02-05 00:00:00') AND window_start < toDateTime('2021-02-08 04:00:00')
    ) AS prior_rth_close
SELECT
    round(prior_rth_close, 2) AS prior_close,
    round(toFloat64(argMinIf(open, window_start, (toHour(toTimeZone(window_start, 'America/New_York')) * 60 + toMinute(toTimeZone(window_start, 'America/New_York'))) BETWEEN 570 AND 959)), 2) AS rth_open,
    round((toFloat64(argMinIf(open, window_start, (toHour(toTimeZone(window_start, 'America/New_York')) * 60 + toMinute(toTimeZone(window_start, 'America/New_York'))) BETWEEN 570 AND 959)) / prior_rth_close - 1) * 100, 1) AS gap_pct,
    round(minIf(toFloat64(low), (toHour(toTimeZone(window_start, 'America/New_York')) * 60 + toMinute(toTimeZone(window_start, 'America/New_York'))) BETWEEN 570 AND 959), 2) AS rth_low,
    formatDateTime(toTimeZone(argMinIf(window_start, (toFloat64(low), toInt64(toUnixTimestamp(window_start))), (toHour(toTimeZone(window_start, 'America/New_York')) * 60 + toMinute(toTimeZone(window_start, 'America/New_York'))) BETWEEN 570 AND 959), 'America/New_York'), '%H:%i') AS low_et,
    round(maxIf(toFloat64(high), (toHour(toTimeZone(window_start, 'America/New_York')) * 60 + toMinute(toTimeZone(window_start, 'America/New_York'))) BETWEEN 570 AND 959), 2) AS rth_high,
    round(toFloat64(argMaxIf(close, window_start, (toHour(toTimeZone(window_start, 'America/New_York')) * 60 + toMinute(toTimeZone(window_start, 'America/New_York'))) BETWEEN 570 AND 959)), 2) AS rth_close,
    round((toFloat64(argMaxIf(close, window_start, (toHour(toTimeZone(window_start, 'America/New_York')) * 60 + toMinute(toTimeZone(window_start, 'America/New_York'))) BETWEEN 570 AND 959)) / prior_rth_close - 1) * 100, 1) AS day_change_pct,
    round((minIf(toFloat64(low), (toHour(toTimeZone(window_start, 'America/New_York')) * 60 + toMinute(toTimeZone(window_start, 'America/New_York'))) BETWEEN 570 AND 959) / prior_rth_close - 1) * 100, 1) AS low_vs_prior_pct,
    round(toFloat64(sum(volume)) / 1e6, 1) AS day_shares_m,
    countIf((toHour(toTimeZone(window_start, 'America/New_York')) * 60 + toMinute(toTimeZone(window_start, 'America/New_York'))) BETWEEN 570 AND 959) AS rth_minute_bars
FROM global_markets.delayed_stocks_minute_aggs
WHERE ticker = 'TSLA'
  AND window_start >= toDateTime('2021-02-08 04:00:00') AND window_start < toDateTime('2021-02-08 23:59:00')

Here is the tape's quiet verdict: TSLA opened at $869.67 (+2% from Friday), touched $877.77, and closed at $863.3, +1.3% on the day, on 19.7 million shares. A famous market moment, and a thoroughly ordinary session for 2021-vintage TSLA, a stock that had spent months moving multiple percent on no news at all. The asset that actually moved violently that day was bitcoin itself, which repriced double digits, but that market lives outside this warehouse's licensed universe, so this page sticks to what our tape can verify.

"Ordinary" is a measurement, not a vibe

That word, ordinary, is a claim, so here is its baseline: every TSLA session from September 2020 (the first full month after its 5-for-1 stock split, so all closes sit on the same as-traded basis) through the Friday before the disclosure.

WITH daily AS (
    SELECT
        toDate(toTimeZone(window_start, 'America/New_York')) AS et_date,
        argMaxIf(toFloat64(close), window_start, (toHour(toTimeZone(window_start, 'America/New_York')) * 60 + toMinute(toTimeZone(window_start, 'America/New_York'))) BETWEEN 570 AND 959) AS close_usd
    FROM global_markets.delayed_stocks_minute_aggs
    WHERE ticker = 'TSLA'
      AND window_start >= toDateTime('2020-08-31 00:00:00') AND window_start < toDateTime('2021-02-09 01:00:00')
    GROUP BY et_date
),
chg AS (
    SELECT
        et_date,
        close_usd,
        lagInFrame(close_usd) OVER (ORDER BY et_date ASC ROWS BETWEEN 1 PRECEDING AND CURRENT ROW) AS prev_close
    FROM daily
)
SELECT
    countIf(et_date >= toDate('2020-09-01') AND et_date < toDate('2021-02-08') AND prev_close > 0) AS baseline_sessions,
    round(quantileDeterministicIf(0.5)(abs(close_usd / prev_close - 1) * 100, toUInt32(et_date), et_date >= toDate('2020-09-01') AND et_date < toDate('2021-02-08') AND prev_close > 0), 1) AS median_abs_move_pct,
    countIf(et_date >= toDate('2020-09-01') AND et_date < toDate('2021-02-08') AND prev_close > 0 AND abs(close_usd / prev_close - 1) > abs((SELECT close_usd / prev_close - 1 FROM chg WHERE et_date = toDate('2021-02-08')))) AS sessions_moving_more,
    round(anyIf((close_usd / prev_close - 1) * 100, et_date = toDate('2021-02-08')), 1) AS feb8_change_pct
FROM chg
HAVING countIf(et_date >= toDate('2020-09-01') AND et_date < toDate('2021-02-08') AND prev_close > 0) > 0

Across those 109 sessions, the median daily move was 2.5% in absolute terms, and 80 of the 109 moved more than the bitcoin day's +1.3%. A $1.5 billion treasury surprise landed as a below-median day on Tesla's own tape.

The shape of the session

SELECT
    formatDateTime(toStartOfInterval(toTimeZone(window_start, 'America/New_York'), INTERVAL 30 MINUTE), '%H:%i') AS et_time,
    round(toFloat64(argMax(close, window_start)), 2) AS bucket_close,
    round(min(toFloat64(low)), 2) AS bucket_low,
    round(toFloat64(sum(volume)) / 1e6, 1) AS shares_m
FROM global_markets.delayed_stocks_minute_aggs
WHERE ticker = 'TSLA'
  AND window_start >= toDateTime('2021-02-08 04:00:00') AND window_start < toDateTime('2021-02-08 23:59:00')
  AND (toHour(toTimeZone(window_start, 'America/New_York')) * 60 + toMinute(toTimeZone(window_start, 'America/New_York'))) BETWEEN 570 AND 959
GROUP BY et_time
ORDER BY et_time

The strength came early, the largest buckets sit in the morning as the disclosure circulated, and the afternoon drifted sideways in a narrow band. As event tapes go, this is what "priced within the hour" looks like, a useful contrast to the overnight NVDA gap where the market was closed for the reaction, or the GME halt cascade where the reaction kept breaking the market. Same era, three completely different event mechanics.

The sympathy tape: where the move actually went

TSLA was the wrong place to watch this event. The right place was the set of stocks whose treasury or business already lived on bitcoin, the miners, the corporate holders, the payment apps with crypto rails. Same headline, very different tapes:

SELECT
    ticker,
    round((event_close / prior_close - 1) * 100, 1) AS change_pct
FROM (
    SELECT
        ticker,
        argMaxIf(toFloat64(close), window_start, toDate(toTimeZone(window_start, 'America/New_York')) = toDate('2021-02-05') AND (toHour(toTimeZone(window_start, 'America/New_York')) * 60 + toMinute(toTimeZone(window_start, 'America/New_York'))) BETWEEN 570 AND 959) AS prior_close,
        argMaxIf(toFloat64(close), window_start, toDate(toTimeZone(window_start, 'America/New_York')) = toDate('2021-02-08') AND (toHour(toTimeZone(window_start, 'America/New_York')) * 60 + toMinute(toTimeZone(window_start, 'America/New_York'))) BETWEEN 570 AND 959) AS event_close
    FROM global_markets.delayed_stocks_minute_aggs
    WHERE ticker IN ('MSTR', 'RIOT', 'MARA', 'SQ', 'PYPL', 'SPY')
      AND window_start >= toDateTime('2021-02-05 00:00:00') AND window_start < toDateTime('2021-02-09 01:00:00')
    GROUP BY ticker
)
ORDER BY change_pct DESC

MARA closed +42.5% and RIOT +40.2%, the two bitcoin miners in the set, repricing an order of magnitude more than TSLA itself. MSTR, the software company that had turned its treasury into a bitcoin vehicle the prior summer, gained 29.2%. The payment platforms with crypto rails, SQ and PYPL, added single digits, and the broad-market ETF SPY closed +0.7%. The pattern is clean: the session repriced bitcoin exposure, wherever it lived, and Tesla, whose new position was 7.7% of its cash, had the least of it.

The options tape shrugged too

Did the leverage-seeking crowd at least show up in the options? Trade records here reach back to 2014, so the panel below compares February 8 against the prior twenty sessions, a baseline that includes the late-January mania weeks, a demanding yardstick:

WITH daily AS (
    SELECT
        toDate(toTimeZone(sip_timestamp, 'America/New_York')) AS session,
        toFloat64(sum(size)) AS contracts,
        toFloat64(sumIf(size, option_type = 'P')) AS put_contracts,
        toFloat64(sumIf(size, option_type = 'C')) AS call_contracts
    FROM global_markets.options_trades
    WHERE ticker LIKE 'O:TSLA2%'
      AND sip_timestamp >= toDateTime('2021-01-08 00:00:00') AND sip_timestamp < toDateTime('2021-02-09 01:00:00')
    GROUP BY session
)
SELECT
    round(maxIf(contracts, session = toDate('2021-02-08')) / 1e6, 2) AS event_contracts_m,
    round(avgIf(contracts, session < toDate('2021-02-08')) / 1e6, 2) AS prior_avg_contracts_m,
    round(maxIf(contracts, session = toDate('2021-02-08')) / avgIf(contracts, session < toDate('2021-02-08')), 2) AS volume_multiple,
    countIf(session < toDate('2021-02-08')) AS baseline_sessions,
    round(maxIf(put_contracts, session = toDate('2021-02-08')) / maxIf(call_contracts, session = toDate('2021-02-08')), 2) AS event_put_call,
    round(sumIf(put_contracts, session < toDate('2021-02-08')) / sumIf(call_contracts, session < toDate('2021-02-08')), 2) AS baseline_put_call,
    round(sumIf(put_contracts, session < toDate('2021-02-08')) / sumIf(call_contracts, session < toDate('2021-02-08')) - maxIf(put_contracts, session = toDate('2021-02-08')) / maxIf(call_contracts, session = toDate('2021-02-08')), 2) AS callward_shift
FROM daily
HAVING countIf(session = toDate('2021-02-08')) > 0 AND countIf(session < toDate('2021-02-08')) > 0

0.85 million TSLA option contracts traded on the day, 0.78x the prior-month average of 1.09 million. Not an explosion; a below-average day even by contract count. The mix did lean: the put/call ratio (put contracts divided by call contracts, a rough gauge of which direction traders are paying up for) printed 0.52 against 0.7 across the baseline, a call-ward tilt of 0.18. Relatively more upside interest than the prior month's norm, on unremarkable volume. Equity tape, options tape, same verdict: noted, absorbed, moved on.

Did the move hold?

A deep dive owes the reader the next question: was +1.3% the start of something? Here is the disclosure day plus the twenty sessions that followed, each close indexed to February 8:

WITH daily AS (
    SELECT
        toDate(toTimeZone(window_start, 'America/New_York')) AS et_date,
        round(argMaxIf(toFloat64(close), window_start, (toHour(toTimeZone(window_start, 'America/New_York')) * 60 + toMinute(toTimeZone(window_start, 'America/New_York'))) BETWEEN 570 AND 959), 2) AS close_usd
    FROM global_markets.delayed_stocks_minute_aggs
    WHERE ticker = 'TSLA'
      AND window_start >= toDateTime('2021-02-08 00:00:00') AND window_start < toDateTime('2021-03-11 01:00:00')
    GROUP BY et_date
)
SELECT
    toString(et_date) AS session,
    close_usd,
    round((close_usd / (SELECT close_usd FROM daily WHERE et_date = toDate('2021-02-08')) - 1) * 100, 1) AS pct_from_feb8
FROM daily
ORDER BY et_date ASC
LIMIT 21

No. TSLA never closed back above the disclosure-day close inside this window. Five sessions later (2021-02-16) the running change read -7.8%; by 2021-03-08 it read -34.8%, and a sharp bounce on the window's final row still left it at -22%. None of that is a statement about bitcoin, the same stretch was harsh for the era's high-multiple growth names generally, and February 8 sits near the top of TSLA's early-2021 range. A single-day event study inherits whatever regime it lands in; narrating this chart as "the bitcoin buy top-ticked Tesla" reads coincidence as mechanism.

Cars for bitcoin, the U-turn, and the 2022 exit

The 10-K's second commitment, accepting bitcoin as payment, has a complete, dated arc, and it is the part many "Tesla bitcoin" searches are really about:

1. February 8, 2021, the 10-K discloses the $1.5 billion purchase and the intent to accept bitcoin as payment (receipted at the top of this page).
2. March 24, 2021, Musk announces that US customers can buy a Tesla with bitcoin. The experiment goes live.
3. April 26, 2021, with first-quarter results, Tesla discloses that it sold "approximately 10%" of its bitcoin holdings, described at the time as a liquidity demonstration, a figure quoted from the company's own disclosure text, not a warehouse query.
4. May 12, 2021, roughly seven weeks after payments went live, Tesla suspends vehicle purchases with bitcoin, citing the fossil-fuel energy mix of mining. The treasury position stays.
5. July 20, 2022, Tesla's second-quarter update discloses the big exit: it "converted approximately 75% of its bitcoin holdings to fiat currency," adding "$936 million of cash" to the balance sheet, figures quoted from the company's own filing text; this warehouse holds no bitcoin-holdings column to compute them independently.

That last disclosure left the same kind of paper-and-tape trail as the first one, so it gets the same treatment:

WITH
    (
        SELECT round(argMaxIf(toFloat64(close), window_start, (toHour(toTimeZone(window_start, 'America/New_York')) * 60 + toMinute(toTimeZone(window_start, 'America/New_York'))) BETWEEN 570 AND 959), 2)
        FROM global_markets.delayed_stocks_minute_aggs
        WHERE ticker = 'TSLA'
          AND window_start >= toDateTime('2022-07-20 00:00:00') AND window_start < toDateTime('2022-07-21 04:00:00')
    ) AS jul20_close,
    (
        SELECT count()
        FROM global_markets.stocks_sec_edgar_index
        WHERE cik = '0001318605' AND filing_date = '2022-07-20' AND form_type = '8-K'
    ) AS eightk_count
SELECT
    eightk_count AS disclosure_8k_on_file,
    jul20_close AS prior_close,
    round(argMaxIf(toFloat64(close), window_start, (toHour(toTimeZone(window_start, 'America/New_York')) * 60 + toMinute(toTimeZone(window_start, 'America/New_York'))) BETWEEN 570 AND 959), 2) AS jul21_close,
    round((argMaxIf(toFloat64(close), window_start, (toHour(toTimeZone(window_start, 'America/New_York')) * 60 + toMinute(toTimeZone(window_start, 'America/New_York'))) BETWEEN 570 AND 959) / jul20_close - 1) * 100, 1) AS next_session_change_pct,
    round(toFloat64(sum(volume)) / 1e6, 1) AS jul21_shares_m
FROM global_markets.delayed_stocks_minute_aggs
WHERE ticker = 'TSLA'
  AND window_start >= toDateTime('2022-07-21 00:00:00') AND window_start < toDateTime('2022-07-21 23:59:00')

The disclosure 8-K is on file (1 filing dated July 20, 2022 in our EDGAR records), and the release landed after that day's close. The next session TSLA rose 9.8%, from $742.28 to $815.21 on 46.5 million shares. One honest caveat before anyone stores that as "the market cheered the bitcoin sale": the same release carried Tesla's full second-quarter results, and the bitcoin line was one paragraph of it. On neither end does the tape isolate bitcoin, the entrance coincided with a +1.3% session, the exit with a +9.8% session that had an entire earnings report inside it.

What this day teaches

Two mechanics. First, where a disclosure lives shapes how it prices: a 10-K paragraph diffuses through the market over hours as it is read, versus the step-function of a scheduled announcement, one reason the morning buckets carry the move. Second, the significance of an event and the size of the day's stock move are different things: this session's modest percentage belies its consequence, which unfolded over years as the corporate-bitcoin question became a permanent treasury debate, and as the position's own exit, receipted above, closed the loop in the same paper trail. Historic days do not always look historic on the tape; the filing record is often the better receipt.

There is also a scale lesson in the pairing of the two numbers involved: $1.5 billion was 7.7% of Tesla's year-end cash and a rounding error against its market value, yet the disclosure re-anchored a whole policy debate, treasurers, auditors, and boards suddenly needed a bitcoin opinion. Materiality in filings is measured in dollars; materiality in markets is measured in precedent. The gap between those two measures is where this day lives.

Tesla bitcoin FAQ

How did Tesla announce its bitcoin purchase?

Inside its annual report, the 10-K filed February 8, 2021, receipted above from the EDGAR index, disclosing a $1.5 billion purchase and the intent to accept bitcoin as payment.

How much bitcoin did Tesla buy?

$1.5 billion worth. The 10-K disclosed a dollar amount, never a coin count; the widely quoted coin figures are outside estimates that divide those dollars by early-2021 bitcoin prices.

What did Tesla stock do when the bitcoin purchase was announced?

It rose 1.3% that session, a below-median day by Tesla's own prior five months, in which 80 of 109 sessions moved more. The large moves that day belonged to bitcoin-adjacent stocks like the miners, receipted above.

Did Tesla ever accept bitcoin for car purchases?

Yes, briefly. Bitcoin checkout for US customers went live on March 24, 2021, and Tesla suspended it on May 12, 2021 in a statement citing the fossil-fuel energy mix of bitcoin mining, roughly seven weeks of live bitcoin payments.

Does Tesla still hold its bitcoin?

Tesla disclosed on July 20, 2022 that it had converted approximately 75% of its bitcoin into fiat currency, and it has adjusted the remainder since. The current answer lives where the original one did: in the company's most recent 10-Q or 10-K.

Every panel above is a stored, versioned query over the historical tape, expand the SQL to see each measurement. Want to feel this day instead of reading it? It is one of the playable scenarios in the Strasmore Labs trading simulator.