The Musk Ledger

[Tesla vehicles will be capable of complete autonomy in] a bit over two years.

The prediction

Full/complete self-driving by roughly 2018. (Fortune's headline: 'Tesla Vehicles Will Drive Themselves in Two Years.')

What actually happened

Not achieved. In 2026 Teslas still require a supervising driver; FSD remains SAE Level 2 driver-assistance.

Note: Paraphrase of an interview rather than a verbatim quote.

SourcesFortune

We'll be able to do a demonstration drive of full autonomy all the way from LA to New York … by the end of next year, without the need for a single touch, including the charger.

The prediction

A zero-intervention, LA-to-New-York self-driving demonstration by the end of 2017.

What actually happened

It never happened - not in 2017, and the promised no-human-touch coast-to-coast demo has still never been performed.

SourcesTechCrunch

All Tesla vehicles produced in our factory … will have the hardware needed for full self-driving capability at a safety level substantially greater than that of a human driver.

The prediction

Cars built from 2016 (HW2) would become Level 4/5 self-driving through over-the-air software updates by 2018.

What actually happened

HW2/HW2.5 cars were never made fully self-driving. Tesla later conceded they needed hardware retrofits (HW3, then HW4), and by 2025 acknowledged even some HW3 cars may need upgrades. The software-only promise failed.

SourcesElectrek·NBC News

I think we will be feature complete - full self-driving - this year.

The prediction

FSD feature-complete in 2019, with drivers able to 'fall asleep and wake up at their destination' by end of 2020.

What actually happened

FSD was not feature-complete in 2019, and sleeping in a moving Tesla is still neither legal nor supported. FSD remains driver-supervised.

SourcesCNBC

A year from now, we'll have over a million cars with full self-driving … operating robotaxis next year with no one in them.

The prediction

More than one million Tesla robotaxis operating driverless by the end of 2020.

What actually happened

Zero robotaxis operated in 2020. No robotaxi pilot began in any form until a narrow, supervised Austin rollout in mid-2025.

SourcesCNBC

It's financially insane to buy anything other than a Tesla. … They will be like owning a horse in three years.

The prediction

FSD would turn each Tesla into an income-generating robotaxi, making it appreciate in value.

What actually happened

With full autonomy never arriving, Teslas depreciated like normal cars; used-Tesla prices fell sharply in 2023–2024.

SourcesLex Fridman Podcast #18·CNBC·Electrek (retrospective)

I remain confident that we will have the basic functionality for Level 5 autonomy complete this year.

The prediction

Level 5 (no human needed, any conditions) essentially complete in 2020.

What actually happened

No Level 5 - or even consumer Level 3 - arrived in 2020 or since. Tesla's own filings to California regulators stated FSD would remain Level 2.

SourcesReuters

My personal guess is that we'll achieve Full Self-Driving this year, [with safety] better than human.

The prediction

FSD solved, safer than a human, in 2022 - the same promise made in 2020 and 2021.

What actually happened

Not achieved in 2022. FSD Beta remained a supervised Level 2 system. Jalopnik noted it was 'next year' for the ninth year running.

Note: Worth a primary-transcript check for exact wording; substance well documented.

SourcesThe Verge

[A dedicated robotaxi] highly optimized for autonomy … with no steering wheel or pedals … aiming for volume production in 2024.

The prediction

A purpose-built robotaxi mass-produced in 2024.

What actually happened

No robotaxi was mass-produced in 2024. The 'Cybercab' was unveiled as a prototype in October 2024, with production pushed to 'before 2027'; the purpose-built model entered production in February 2026.

SourcesTechCrunch·Electrek

[A paid ride-hailing robotaxi service in Austin in June 2025 with] no drivers in the cars … unsupervised FSD to owners in many regions of the U.S. in 2025.

The prediction

Driverless paid robotaxis in Austin by June 2025, scaling to 8–10 metros and 'half the US population' by year-end; unsupervised FSD released to owners nationwide.

What actually happened

The June 2025 launch was geofenced, invite-only, ~10-20 vehicles, and - contradicting 'no drivers' - carried in-car safety monitors plus chase cars. The service later expanded to full commercial operation in Dallas and Houston (April 2026) and to additional metros, while the Bay Area still ran only a safety-driver service and several planned cities had not launched. The predicted scale-out to 8-10 metros and 'half the US population' by end-2025 was not met.

Note: Prediction is verbatim/close; the rollout reality is drawn from detailed Electrek tracking.

SourcesTechCrunch·Electrek (status check)·Teslarati·Wikipedia

We think probably we'll have a thousand cars or more in the Bay Area by the end of this year … probably, I don't know, 500 or more in the greater Austin area.

The prediction

A fleet of at least 500 robotaxis in greater Austin, and 1,000 or more in the Bay Area, by the end of 2025.

What actually happened

The Austin fleet was only about 30 vehicles by late November 2025 - roughly 94% short of the 500 target - and Musk revised the goal down to about 60. California never granted Tesla a permit to operate without a safety driver, so the Bay Area half also failed.

Note: Quote is a cleaned transcription of a single continuous utterance from the episode audio (YouTube j6_VfR-CyuM, ~66:25-66:47); the 'We think' opener and the 'probably / I don't know' hedges are attested in that primary auto-transcript, and the ellipsis elides in-utterance filler only. Distinct from the June 2025 Austin launch entry above - this is the later, more specific fleet-count promise.

SourcesAll-In Podcast (YouTube)·Electrek·CNBC

[Humans on Mars in] as little as 10 years … worst case, 15 to 20 years.

The prediction

First humans on Mars possibly by 2021.

What actually happened

No human - and no SpaceX vehicle - has reached Mars. Missed by 5+ years and counting.

Note: Widely reported paraphrase of a 2011 interview; predates the 10-year window but sets the pattern.

SourcesNBC News

[First crewed Mars flights potentially 2024; ~$200,000 a seat; ultimately a self-sustaining city of one million people.]

The prediction

Crewed Mars missions beginning ~2024, building toward a million-person Martian city.

What actually happened

No crewed or uncrewed SpaceX Mars flight has occurred; the vehicle (later Starship) has not reached orbit or Mars.

SourcesTIME·Space.com

[An uncrewed Dragon capsule would soft-land on Mars as early as 2018.]

The prediction

A robotic Mars landing by 2018.

What actually happened

Never flew. Slipped to 2020, then cancelled when SpaceX dropped propulsive Dragon landings. Musk in 2017: 'I'm pretty confident that is not the right way [to land on Mars].'

SourcesSpaceNews

[Two private citizens who paid deposits would fly a week-long loop around the Moon before the end of 2018 on Falcon Heavy and Crew Dragon.]

The prediction

A crewed lunar-flyby tourism flight by late 2018.

What actually happened

Never happened; slipped to 'mid-2019 or later,' then was quietly dropped and folded into the (later cancelled) Starship dearMoon plan.

SourcesSpace.com

[First uncrewed cargo ships to Mars in 2022; two cargo plus two crew ships departing for Mars in 2024.]

The prediction

Uncrewed Mars landing in 2022; crewed Mars landing in 2024.

What actually happened

Neither occurred; no Starship has left Earth orbit.

SourcesSpace.com

Fly to most places on Earth in under 30 mins and anywhere in under 60. Cost per seat should be about the same as full fare economy in an aircraft.

The prediction

Commercial point-to-point suborbital rocket passenger travel.

What actually happened

Never demonstrated with passengers; no commercial point-to-point service exists. Still purely conceptual.

SourcesDezeen

[Yusaku Maezawa and eight artists would fly a Starship loop around the Moon by the end of 2023.]

The prediction

A crewed Starship lunar flyby by end of 2023.

What actually happened

Cancelled June 2024 due to Starship delays; never flew. Maezawa: 'I can't plan my future in this situation.'

SourcesSpaceNews

This is going to sound totally nuts, but I think we want to try to reach orbit in less than six months.

The prediction

First orbital Starship flight by roughly early-to-mid 2020.

What actually happened

The first integrated Starship flight came in April 2023 and exploded minutes in; the 'less than six months' deadline was missed by years. On Flight 13 (July 24, 2026) Starship flew a suborbital point-to-point flight, deployed 20 Starlink V3 satellites designed to burn up on reentry, and soft-splashed down intact - but still had not completed a full orbital insertion.

SourcesSpace.com·Ars Technica

[Highly confident of landing humans on Mars] about six years from now. … If we get lucky, maybe four years.

The prediction

A crewed Mars landing in 2024–2026.

What actually happened

No SpaceX vehicle has reached Mars or trans-Mars injection, and the 2026 window passed with no launch. In February 2026 Musk said SpaceX was delaying its Mars ambitions by about five to seven years to focus on lunar missions.

SourcesCNBC·CNN

[SpaceX's Starship Human Landing System would land NASA astronauts on the Moon for Artemis 3, targeted for 2024.]

The prediction

A crewed lunar landing in 2024 (later slipped to Dec 2025).

What actually happened

The GAO (Nov 2023) found a landing 'unlikely before 2027.' NASA has since re-scoped Artemis III as a 2027 low-Earth-orbit rendezvous and docking demonstration rather than a lunar landing, moving the first crewed landing to Artemis IV in early 2028. Starship orbital refueling - required for the landing profile - still has not been demonstrated.

SourcesSpace.com (GAO)·Space.com (2028 report)·Wikipedia

[~100 Starships a year, a fleet of 1,000 ships carrying ~100,000 people per window, to reach one million people on Mars by 2050. Asked to confirm: 'Yes.']

The prediction

A self-sustaining, million-person Martian city by 2050.

What actually happened

Zero people on Mars; launch cadence and orbital refueling are orders of magnitude short of what the plan requires.

Note: Long-range target not yet formally due - included because the near-term steps have already slipped badly.

SourcesFuturism

[The fastest production car ever made - 0–60 in 1.9s, 620+ mile range, $200,000 - production in 2020;] later, a 'SpaceX package' with cold-gas thrusters that could let it briefly hover.

The prediction

Roadster in production by 2020, later hyped as able to nearly 'fly.'

What actually happened

Still not in production by August 2026 - roughly nine years after the 2017 unveiling and six years past the promised 2020 availability - despite $50,000 reservation deposits taken since 2017. The 'flying' demo has been repeatedly pushed back (most recently to April 1, 2026), with production reportedly targeted to 2027-2028.

SourcesTechCrunch·TechRadar·Electrek

[500-mile range, production starting 2019.]

The prediction

Volume production of the electric Semi truck beginning in 2019.

What actually happened

Only a handful of pilot units reached PepsiCo in December 2022; high-volume production began only on April 29, 2026 at Gigafactory Nevada - roughly eight and a half years after the 2019 production promise - with PepsiCo, Walmart, Costco, Sysco and US Foods among customers.

SourcesElectrek·TechCrunch·Wikipedia

[Single-motor RWD $39,900; dual-motor $49,900; tri-motor $69,900 - cheaper models in late 2021.]

The prediction

A $39,900 base Cybertruck, on sale from late 2021.

What actually happened

Production didn't start until late 2023; real prices ran ~$100,000–$120,000. The $39,900 model never appeared, and the cheap RWD preorder option was pulled.

SourcesCNBC·Yahoo/AOL

[Cybertruck could eventually sell 250,000–500,000 units per year.]

The prediction

Annual Cybertruck sales of a quarter- to half-million.

What actually happened

About 39,000 Cybertrucks were sold in all of 2024 - a small fraction of even the low-end target - and sales fell further in 2025, roughly 16,000 in the first three quarters, as the truck lost the US best-selling electric pickup title to the Ford F-150 Lightning.

SourcesInsideEVs·Electrive·USA Today

[5,000 Model 3s per week by December 2017.]

The prediction

A 5,000/week Model 3 production rate by end of 2017.

What actually happened

Only ~2,425 were built in all of Q4 2017; the 5,000/week rate wasn't hit until the end of June 2018 - about six months late, via a tent-based line.

SourcesNPR·TechCrunch

[An aspiration of 20 million vehicles per year by 2030.]

The prediction

20 million Teslas produced annually by 2030.

What actually happened

Tesla delivered about 1.79M vehicles in 2024 and 1,636,129 in 2025 - an order of magnitude short of 20M - and the goal was removed from Tesla's impact report in May 2024, not restated since.

SourcesReuters via US News·Tesla Q4 2025 Update

[~50% average annual growth in vehicle deliveries over a multi-year horizon.]

The prediction

Sustained ~50% yearly delivery growth.

What actually happened

Far from 50% growth, Tesla deliveries fell about 1.1% in 2024 (the first annual decline since 2011) and a further 9% in 2025 to 1,636,129 vehicles, a second consecutive annual decline. Tesla also posted its first-ever annual revenue decline in 2025, with revenue of about $94.8B and net income down 46%.

SourcesPBS/AP·CBT News·The Tech Portal·Tesla Q4 2025 Update

With the advancements in vehicle autonomy and the introduction of new products, we expect the vehicle business to return to growth in 2025.

The prediction

Tesla vehicle deliveries to grow again in 2025 after 2024's decline - Musk's earlier framing on the Q3 2024 call was 'something like a 20% to 30% growth next year is my best guess.'

What actually happened

Deliveries fell about 9% in 2025 to 1,636,129 vehicles - a second straight annual decline - with automotive revenue down about 10% and Tesla's first-ever annual revenue decline.

Note: 'We expect' is Tesla's own guidance wording (reaffirmed in the January 2025 deck); the claim is scored on the calendar-2025 outcome. Musk's hedged 'best guess' 20-30% range from the Q3 2024 call is context, not the scored commitment.

SourcesTesla Q4 2024 Update (deck)·Tesla Q4 2025 Update·CBT News·Motley Fool (Q3 2024 call transcript)

[The Solar Roof would cost less than a conventional roof plus the electricity it generates; target of 1,000 installations per week by end of 2019.]

The prediction

A cheaper-than-normal solar roof installed at 1,000 per week.

What actually happened

Peak deployment (2022) was only ~23 roofs per week - ~97.7% short. Prices ballooned (one quote jumped $72k→$146k), a class action followed, and Tesla stopped reporting Solar Roof numbers and pivoted back to panels.

SourcesElectrek·TechCrunch

[A Model S battery swapped in ~90 seconds, 'faster than you could fill a gas tank,' with swap stations planned along highways.]

The prediction

A network of battery-swap stations as an alternative to charging.

What actually happened

Of ~200 invited owners, only a handful ever used the pilot. Musk in 2015: 'People don't care about pack swap.' The program was abandoned; the network never materialized.

SourcesFortune

[A $35,000 base Model 3.]

The prediction

A genuinely $35,000 mass-market Tesla.

What actually happened

The $35,000 version only appeared in Feb 2019, sold off-menu, made by software-limiting a pricier car; it was pulled within weeks and discontinued with the 2021 refresh. It never returned at that price.

SourcesGreen Car Reports·CNN

[A Model S Plaid+ with 520+ miles of range, ~$130k, slated for 2021.]

The prediction

A 520-mile flagship Model S.

What actually happened

Musk cancelled it in June 2021 - 'Plaid+ is canceled. No need, as Plaid is just so good' - before a single delivery. The shipped Plaid tops out around 390–405 miles.

SourcesCNBC

[Pods traveling LA–San Francisco in ~30–35 minutes at up to ~760 mph.]

The prediction

A near-supersonic intercity vacuum-tube transit system.

What actually happened

No operational Hyperloop exists. SpaceX shut its test-pod competition and paved over the track around 2022–2023; Virgin Hyperloop dropped passenger plans and laid off staff.

SourcesHyperloop Alpha (PDF)·NBC News

Just received verbal govt approval for The Boring Company to build an underground NY-Phil-Balt-DC Hyperloop. NY-DC in 29 mins.

The prediction

An operational East Coast Hyperloop; New York to DC in 29 minutes.

What actually happened

No such approval or project materialized; nothing was built. Officials in the White House and named cities said no firm approvals existed.

SourcesCBS News·IEEE Spectrum

[Promised high-speed autonomous tunnel networks - including a DC–Baltimore Loop and a 150-mph, ~12-minute O'Hare-to-downtown Chicago express.]

The prediction

Large-scale autonomous, high-speed underground networks that would 'solve traffic.'

What actually happened

The DC-Baltimore and Chicago O'Hare projects were abandoned. The only operating system remains the Las Vegas Loop - human-driven Teslas at about 35 mph across 2.2 miles and 9 stations, not autonomous 150-mph pods - with Clark County adding safety requirements as the tunnel network expands (May 2026) toward a planned 68-mile, 55-stop system.

SourcesWashington Post·ProPublica·Wikipedia

[Aiming to begin human trials by the end of 2020.]

The prediction

First human brain-implant trials in 2020 - a claim repeated in following years.

What actually happened

The first human implant did not happen until January 2024 - roughly 3+ years late.

SourcesCNBC·NPR

We're hoping later this year to do our first device implant for a human, enabling someone who is completely blind to see.

The prediction

A blind person's sight restored via the 'Blindsight' implant - claimed for 2024, then 2025.

What actually happened

No blind person's vision has been restored by Neuralink. Blindsight received FDA 'breakthrough device' designation in September 2024 but had still not entered a human trial as of mid-2026, while Neuralink's first human implant (for paralysis) came in January 2024. The 'this year' claim was missed.

SourcesNewsweek·Wikipedia

Optimus ultimately will be worth more than the car business, worth more than full self-driving.

The prediction

Optimus humanoid robots in production in 2023, becoming a bigger business than Tesla's cars.

What actually happened

No production robots arrived in 2023. As of 2026 Optimus is still not doing useful work - Musk said in January 2026 the units were not yet productive - and Tesla planned to unveil Gen 3 in Q1 2026, with production to begin before the end of 2026.

SourcesCNBC·TechCrunch·Electrek

We expect to have thousands of Optimus robots working in Tesla factories by the end of this year.

The prediction

Thousands of Optimus units working in Tesla factories by end of 2025, scaling toward a million a year.

What actually happened

The 'thousands working in factories' target was missed by a wide margin. On the Q4 2025 earnings call (Jan 28, 2026) Musk conceded that no Optimus robots were doing useful work yet, describing them as for learning and data, with no evidence that even hundreds had been built in 2025. Tesla planned to unveil Gen 3 in Q1 2026, with production to begin before the end of 2026.

Note: Prediction is verbatim; the final tally was still emerging at research time.

SourcesTechCrunch·Trefis·Electrek

The coronavirus panic is dumb.

The prediction

That pandemic concern was overblown.

What actually happened

COVID killed ~1.1M+ Americans and ~7M globally and triggered the largest economic shock since the Depression.

SourcesForbes (timeline)

Based on current trends, probably close to zero new cases in US too by end of April.

The prediction

U.S. COVID cases falling to ~zero by end of April 2020.

What actually happened

The U.S. was recording 20,000–30,000+ new cases per day in late April 2020, with 1M+ cumulative cases and 60,000+ deaths by month's end. Off by an enormous margin.

SourcesForbes·Musk tweet

[The U.S. is 'probably' already in a recession that will last up to ~18 months.]

The prediction

A U.S. recession running into 2023.

What actually happened

No U.S. recession materialized; GDP grew ~2.5% in 2023 and unemployment stayed near historic lows.

SourcesFortune

[A global recession lasting 'probably until spring of 2024.']

The prediction

A global recession persisting into spring 2024.

What actually happened

No U.S. recession arrived through spring 2024; markets hit record highs and growth continued.

SourcesCNBC

My guess is that we'll have AI that is smarter than any one human probably around the end of next year.

The prediction

AI smarter than the smartest human by the end of 2025 (and smarter than all humans combined within ~5 years).

What actually happened

As of 2026 there is no consensus that any AI is 'smarter than any one human' across domains; the stated deadline is widely regarded as unmet.

Note: Inherently contestable to adjudicate, but the deadline as phrased was not credibly met.

SourcesFortune

[Projected growing revenue ~5x to $26.4B by 2028, ~931M users by 2028, and 69M Twitter Blue subscribers by 2025.]

The prediction

Explosive revenue, user, and subscription growth after the acquisition.

What actually happened

X reported revenue of about $2.5B in 2023 and $2.7B in 2024 - roughly a tenth of the $26.4B path Musk set - with Blue/Premium subscribers in the low millions. In March 2025 xAI acquired X in a deal valuing X's equity at $33 billion, far below the $44B purchase price, and X Payments expanded to more users only in June 2026 after repeated delays.

SourcesThe Next Web·Fortune·Wikipedia·Bloomberg

[Tesla bought $1.5B in Bitcoin and would accept it as payment for vehicles.]

The prediction

Bitcoin as an ongoing payment option for Teslas.

What actually happened

Tesla suspended Bitcoin payments about three months later (May 2021), citing environmental concerns - a rapid reversal.

Note: A reversed commitment rather than a dated forecast; included for completeness.

SourcesReuters