Latest / Elon Musk Podcast / Elon Musk Weekly News Update: Tesla, SpaceX, X, Neuralink, and More

Transcript

Elon Musk Weekly News Update: Tesla, SpaceX, X, Neuralink, and More · Elon Musk Podcast

• 0:00
  Welcome back to the Elon Musk Podcast, the world's authority
• 0:03
  on Tesla, SpaceX X Neurolink, and all things Elon.
• 0:07
  I'm your host, Will Walden. I've been using Current as my
• 0:11
  bank for about four years and it's honestly been a really
• 0:15
  great experience. No overdraft fees, I get paid up
• 0:18
  to two days early, and the app makes it super easy to track all
• 0:23
  of my finances. And if you're looking for a
• 0:26
  modern way to bank, forget those old banks they're boring and
• 0:30
  also their As aren't that good. O go to the link in the show
• 0:34
  notes, check out current and sign U and support the show.
• 0:38
  Thanks to our amazing community members like you, we've reached
• 0:42
  the top 15 of Spotify's video podcast, the top 10 audio
• 0:47
  podcasts on both Apple and Spotify for the tech category,
• 0:50
  so you all make this possible. If you want to support us more,
• 0:54
  check out our Patreon that's patreon.com/stage 0 News so we
• 0:58
  can keep this free and open for you to enjoy.
• 1:03
  Something interesting happened the other day.
• 1:05
  I was looking through our stats on Spotify and Apple Podcasts
• 1:09
  and I noticed that about 55% of you are not subscribed to the
• 1:15
  show. That means 45% of you are
• 1:17
  subscribed and I really do appreciate your support, but the
• 1:20
  other 55% of you are awesome. But I'm going to ask you for a
• 1:25
  favor. Could you please hit the
• 1:26
  subscribe button? It'll take you one second, but I
• 1:29
  promise you 10 years of this podcast for free, no pay walls.
• 1:33
  I'm not going to charge you anything ever, but I'm going to
• 1:36
  give you 10 years of this show for free.
• 1:38
  I've already been doing it for five years and I plan on doing
• 1:40
  it for 10 more. And the only way that we can
• 1:43
  continue doing this is with your support.
• 1:45
  So one second of your time to hit the subscribe button right
• 1:48
  now would help the show tremendously.
• 1:52
  Thank you so much. Grok, which is the chatbot built
• 1:56
  by Elon Musk's AI company XAI, published a series of violent
• 2:00
  and also anti-Semitic post this week after an internal system
• 2:04
  update broke its basic behavioral safeguards.
• 2:07
  Now prompts one question. Haha prompt how does a billion
• 2:11
  dollar AI company release a chatbot update that praises
• 2:15
  Hitler and also promotes conspiracy theories without
• 2:19
  catching it in a test? And also why would they train it
• 2:23
  on this kind of stuff? Now Crock began echoing
• 2:26
  anti-Semitic talking points and far right rhetoric within hours
• 2:30
  of the update going live. And according to XAI, code
• 2:34
  Change on the back end instructed Grok to mimic the
• 2:36
  tone, style and language of existing EX posts it replied to,
• 2:42
  including ones containing extremist content, and the
• 2:45
  company acknowledged that the instructions it led the AI to
• 2:49
  disregard its built in ethical filters.
• 2:52
  People made a prompt that made it go around those filters.
• 2:55
  Now. The change remained active for
• 2:57
  16 hours before XAI intervened. The incident triggered a major
• 3:02
  backlash after Grok responded to user prompts with posts praising
• 3:06
  Hitler and claiming Jewish people dominated the
• 3:08
  entertainment industry as part of a broader conspiracy.
• 3:13
  Now 1 post revived the long debunk trope of coordinating
• 3:17
  control over Hollywood. Others included direct praise of
• 3:20
  Nazi ideology and mirrored the white nationalist belief system.
• 3:25
  X AI froze Grok's public X account on Tuesday night, but
• 3:29
  allow continue use through the private tab.
• 3:31
  X AI said it removed the problematic code and rewrote the
• 3:35
  system's instruction logic to block similar behavior in the
• 3:38
  future. Now, why they didn't have these
• 3:40
  safeguards in place to begin with is anybody's guess, right?
• 3:45
  Why wouldn't they do that? Why wouldn't they just put the
• 3:48
  safeguards in place in the beginning?
• 3:50
  Don't say things about Hitler that are for Hitler.
• 3:54
  You know, tell the history. That's good enough.
• 3:57
  You don't need to spout off all that rhetoric.
• 4:01
  It's stupid. XAI made a huge mistake.
• 4:04
  And now they're like, we didn't know what happened, why it
• 4:06
  happened, but we're going to take the code out.
• 4:09
  We're going to pause. Grok And then we're going to get
• 4:11
  back to this. And then they said in a way, in
• 4:14
  a backhanded way, they're like this user made these prompts and
• 4:21
  so it's the user's fault in a way, they said that.
• 4:23
  So company said that specific system prompts such as reply to
• 4:28
  the post just like a human and follow the tone and context of
• 4:32
  the X user, which created a feedback loop that prioritized
• 4:35
  mimicry over moral constraint. So if the user is being
• 4:41
  anti-Semitic, they want grok to be that way too.
• 4:43
  Basically they want XA is users and grok to be in an echo
• 4:49
  chamber so they really like staying on the platform.
• 4:53
  So they had a feedback loop that prioritize mimicry over moral
• 4:58
  constraint. And according to XAI, this led
• 5:00
  to Grok to ignore its core values in certain circumstances
• 5:05
  just to sound more engaging. Keep you in that loop.
• 5:08
  And the updated version of Grok also began offering more
• 5:11
  definitive responses to questions about race and
• 5:13
  diversity, dropping previous nuance when answering
• 5:17
  politically charged topics. In several use cases, Grok
• 5:21
  responded using phrasing nearly identical to Elon Musk's own
• 5:25
  tweets. Users notified that it framed
• 5:28
  questions involving Jewish people.
• 5:32
  Users noted that it framed questions involving Jewish
• 5:36
  people with a tone shift toward generalization and also bias.
• 5:41
  Now, at least one prompt involving racial demographics in
• 5:45
  South Africa triggered Grok to mention white genocide, which is
• 5:50
  a theory that Musk has previously mentioned, but which
• 5:53
  South African courts have rejected as unsubstantiated.
• 5:58
  This marks the second major controversy tied to Grok in
• 6:02
  recent months. In May, Grok started referencing
• 6:04
  white nationalist content in response to unrelated questions.
• 6:08
  XAI later blamed that incident as a rogue employee.
• 6:13
  Now, this time, the company tied the root cause directly to
• 6:16
  engineering decisions engineers and started a system level
• 6:19
  prompt upstream of the Grok Spots output layer.
• 6:22
  The modification, according to XAI, introduced a behavior that
• 6:25
  made Grok susceptible to offensive language embedded in
• 6:30
  public X threads. And as of Saturday morning,
• 6:33
  though Grok's public facing account was reinstated, the bot
• 6:37
  seems to be OK now. The company restored the bot's
• 6:39
  ability to interact with users on X after reportedly reworking
• 6:43
  the affected code paths. XAI committed to publishing its
• 6:47
  new system prompt on his public GitHub repository there.
• 6:51
  As of this episode right now, the new prompt remains
• 6:56
  unpublished. XI is probably never going to
• 6:59
  publish it because that's how they work.
• 7:00
  They work in the dark. They say things and they just
• 7:02
  don't do them. Just like other big tech
• 7:04
  companies. Remember, don't be evil by
• 7:06
  Google. That's this now.
• 7:08
  The company insisted that the incident had nothing to do with
• 7:11
  the base language model powering Grok, but was entirely due to a
• 7:14
  system update on the instruction layer of Grok's apology,
• 7:19
  described the post as horrific, and credited user feedback on X
• 7:22
  for identifying the worst cases. Now, the company thanks users
• 7:26
  who helped surface the problematic behavior but stop
• 7:29
  short of detailing how the update was approved or whether
• 7:31
  additional safeguards would be added to prevent similar future
• 7:35
  lapses. Now, Elon Musk wants to say that
• 7:40
  he's a free speech absolutist, and he's also said numerous
• 7:44
  times that all the XAI code or Grok code will be made public.
• 7:50
  Maybe not the underlying technology, but how it all kind
• 7:54
  of works. And right now, they haven't told
• 7:58
  us what they've done. They haven't told us why this
• 8:01
  actually happened, what the prompt was, what the layer was
• 8:04
  that allowed this to happen and what they actually did to
• 8:08
  prevent it from happening again. Now wouldn't you put a layer in
• 8:12
  there? They would just say if a user,
• 8:16
  and this is like super simple programming people like I'm a
• 8:20
  front end web developer as a trade.
• 8:23
  I've been doing it for 20 years now.
• 8:26
  If you can't write logic that says if somebody asks you about
• 8:31
  Hitler, only talk about the history, not speak in the voice
• 8:36
  of Hitler. If you can't do that, and if you
• 8:39
  didn't think about that from the beginning, there's something
• 8:42
  absolutely wrong with you. And if you can't code that, then
• 8:46
  you shouldn't be working at a giant AI company.
• 8:49
  Now. It comes down to management.
• 8:51
  It comes down to people thinking that it's OK for this stuff to
• 8:56
  happen. Now which person is in charge of
• 8:59
  XAI? Elon Musk, free speech
• 9:03
  absolutist, probably had a hand in this.
• 9:06
  Not saying that he told it to say those things about Hitler,
• 9:08
  but probably saying let it do its thing.
• 9:11
  Let it conveniently talk about the things that the person's
• 9:15
  talking about anyway. Keep them in that loop, keep
• 9:18
  them engaged for a while. Be their best friend.
• 9:20
  You can see it also on ChatGPT. ChatGPT is going to be your best
• 9:25
  friend if you have a voice chat with it.
• 9:28
  I've tried it in the past. It's like my best buddy if I
• 9:32
  wanted to be. But in the long run, this raises
• 9:36
  questions about content moderation inside of Musk's
• 9:40
  company. X has had numerous times that
• 9:44
  horrible atrocities have been mentioned next to sponsors ads
• 9:50
  on Twitter posts and X posts and those sponsors pulled their
• 9:55
  sponsorships, pulled the money out of there, and then he must
• 9:58
  threaten them like that's absurd.
• 10:02
  What a weird thing to do, right? He threatened them because they
• 10:04
  pulled their ads because they weren't happy with the service.
• 10:06
  It's free country, right? Free speech.
• 10:08
  If you have free speech, you have free money.
• 10:10
  You can do anything you want to with your money because money
• 10:12
  equals speech. Since the rebrand of Twitter to
• 10:15
  X, Musk is advocated for fewer restrictions to the philosophy
• 10:20
  appears to have crossed over into XA is designs and the
• 10:24
  incident shows that grok, despite being marketed as a
• 10:28
  truth seeking AI can be manipulated by people and it can
• 10:35
  echo hate speech by just prompting it very simply.
• 10:40
  Now as we know XAI hasn't said anything about this.
• 10:44
  They didn't disclose anything that they've done and they are
• 10:48
  filtering things like this in the future.
• 10:51
  Now this active filter will inevitably absorb and reflect
• 10:55
  some of the platform's worst content.
• 10:57
  Hopefully they do it for all the other bad things too.
• 11:01
  Not just anti-Semitic things and Hitler things, but all the other
• 11:05
  things that are just like horrible atrocities.
• 11:07
  And we don't want that echo chamber Onyx AI.
• 11:10
  We don't want that on Grok. So the company is also not
• 11:14
  committed to changing how it tests or approves new code.
• 11:16
  As far as we know, that refactored the entire system and
• 11:20
  plans to be more transparent, sharing its updated instructions
• 11:23
  with the public. Elon hasn't comment on this
• 11:26
  other than we'll figure it out. That's basically what he said.
• 11:29
  So let me know what you think in the comments.
• 11:32
  Do you think Elon had anything to do with this or do you think
• 11:35
  it was just some crafty prompting from somebody who was
• 11:38
  trying to do a gotcha on xai and Grok?
• 11:41
  And Elon? Let me know in the comments on
• 11:43
  your podcast platform or on YouTube, right?
• 11:46
  Take care, everybody. We'll see you in the next one.
• 11:53
  Hey, everybody, welcome back to the Elon Musk podcast.
• 11:56
  This is a show where we discuss the critical crossroads that
• 12:00
  shape SpaceX, Tesla X, The Boring Company, and Neurolink.
• 12:05
  I'm your host, Will Walden. SpaceX plans to launch a
• 12:09
  Starship or numerous Starships to Mars orbit in 2026.
• 12:15
  Now SpaceX needs to complete orbital Starship flights before
• 12:18
  attempting the Mars launch window.
• 12:23
  SpaceX plans to launch a Starship or numerous Starships
• 12:26
  to Mars orbit in 2026. But SpaceX needs to complete
• 12:31
  numerous steps in order to get there.
• 12:33
  They need to complete orbital Starship flights before
• 12:35
  attempting the Mars launch window.
• 12:37
  SpaceX also needs to demonstrate on orbit refueling to provide
• 12:42
  enough propellant for the trip and also SpaceX needs to resolve
• 12:45
  heat shield durability for atmospheric entry at Mars speeds
• 12:50
  if they do choose to enter the atmosphere Mars.
• 12:54
  Now SpaceX has conducted 9 integrated Starship test flights
• 12:58
  through mid 2025 and they reached space on the 9th flight,
• 13:02
  but the Starship failed to stay under control in orbit.
• 13:06
  SpaceX fired Raptor 3 prototype engines more than 300 times in
• 13:10
  pre launch testing, and SpaceX also improved engine thrust and
• 13:14
  cut engine mass in that development program, The Raptor
• 13:17
  3 prototypes. Now they must achieve full orbit
• 13:21
  with booster recovery before targeting Mars.
• 13:24
  SpaceX boosted its FAA approval toward permitting 25 Starship
• 13:29
  flights per year, and SpaceX expects to fly the upgraded
• 13:32
  version 3 Starship by around late 2025, early 2026.
• 13:38
  That calls for the V3 craft to fly with Raptor 3 engines will
• 13:43
  be what SpaceX will be using on those flights.
• 13:48
  Now they need orbital refueling too.
• 13:50
  It's a critical Mars precondition.
• 13:52
  SpaceX must launch Taker Starships to refill a Mars bound
• 13:57
  ship in orbit. They never performed A cryogenic
• 14:00
  fuel transfer at scale in orbit. SpaceX needs to test docking and
• 14:05
  fluid plumbing in zero gravity before resuming Mars missions.
• 14:10
  They they face the extreme engineering challenges of also
• 14:14
  landing on Mars. I don't think they're going to
• 14:15
  do that in the first missions, but they're probably going to be
• 14:18
  testing the systems while they're on the way to Mars and
• 14:22
  also while they're in Mars orbit, little by little, piece
• 14:25
  by piece. So the next few flights they can
• 14:28
  send things or maybe some sort of cargo to the surface of Mars.
• 14:33
  They'll need to manage high speed entry at 7.5 kilometers
• 14:37
  per second and heat shield stress in atomic oxygen
• 14:40
  conditions. They got to perform a propulsive
• 14:43
  landing on a very thin atmosphere that provides minimal
• 14:46
  drag, unlike Earth. And they also must gather entry
• 14:51
  data during the 2026 uncrewed missions to improve their
• 14:54
  systems. So if they actually do enter
• 14:57
  Mars orbit and go into the atmosphere, they could dip into
• 15:01
  the atmosphere and fly out possibly if I, I'm not 100% sure
• 15:06
  about this, but they might test that.
• 15:09
  But I'm what I'm assuming is they're going to go to Mars, do
• 15:14
  a bunch of Mars orbits, and then either fling A Starship into
• 15:19
  space with those orbits and maybe go someplace else, or do
• 15:23
  some orbits and then come back and maybe orbit Earth with that
• 15:26
  Starship if they have enough. Maybe a gravity assist by Mars.
• 15:31
  I'm not 100% sure because they haven't released that data yet.
• 15:35
  But they have to manage Starship development alongside their
• 15:38
  Artemis obligations too. And that's very difficult.
• 15:42
  They're doing 2 absolutely insane missions at the same
• 15:47
  time. SpaceX needs to complete an
• 15:49
  uncrewed lunar Starship demonstration under this NASA
• 15:52
  contract. And they have to refuel the
• 15:55
  Starship in orbit orbit before it reaches the moon for the
• 15:58
  Artemis program. And they have to launch taker
• 16:02
  tankers to fill the HLS craft as well as the Mars bound ships.
• 16:07
  So once they can figure out the on orbit refueling, they'll be
• 16:13
  in a good position to go to the moon and also to Mars.
• 16:18
  They have to make a lot of Starships too, so that you have
• 16:21
  to what I think it was Gwynn Shotwell.
• 16:25
  So they have to do hundreds of Starship test launches without
• 16:30
  humans on them in order to launch humans in a Starship.
• 16:33
  Now, for these missions, they're not going to be launching people
• 16:36
  with the Starship, but they will be launching some cargo, maybe
• 16:39
  satellites for Mars as well. Because if Starship can
• 16:44
  demonstrate that they can deploy Starlink satellites in Earth
• 16:49
  orbit, there's no reason why they couldn't do that in Mars
• 16:53
  orbit as well. Because they want to have Mars
• 16:56
  link. They're working on a star link
• 16:58
  system that will communicate between the surface of Mars to
• 17:03
  the Mars link satellites in orbit around Mars and then back
• 17:07
  to Earth. And maybe there's going to be a
• 17:09
  relay station in the middle. I'm not 100% sure about that
• 17:12
  because SpaceX hasn't released any of that information yet, but
• 17:16
  they need to have Starship crew certified eventually because a
• 17:20
  lunar Starship for Artemis 3 needs to carry people from Mars
• 17:26
  orbit down to the surface of Mars.
• 17:29
  And crew certified just means people can live and work in the
• 17:33
  Starship or they can use the Starship to transfer people.
• 17:37
  They don't need to bring people back to Earth or to Mars in the
• 17:40
  Starship, but what they will do is take people from Mars orbit
• 17:44
  to the surface of Mars in the Starship, and it has to be
• 17:46
  certified for that. They have to have a lot of
• 17:50
  launches too. They have to have high cadence
• 17:53
  of launches in late 2026 to hit this Mars window.
• 17:58
  They need to launch multiple tankers and at least 2 Mars
• 18:01
  ships in rapid succession. Now, if they do this and there's
• 18:07
  pad damage, or if there's regulatory delays or any other
• 18:10
  major failure, it's no go. SpaceX must maintain
• 18:16
  uninterrupted operations across all of their launch sites,
• 18:20
  including Star Base in Texas and the Cape in Florida.
• 18:27
  Now, these missions are placed against Artemis scheduling.
• 18:32
  SpaceX will likely perform the uncrewed lunar HLS demo in late
• 18:36
  2026 or early 2027, and they also may delay Mars if NASA
• 18:41
  requires full focus on crude moon landing.
• 18:45
  SpaceX can't sacrifice Artemis obligations without jeopardizing
• 18:49
  long term funding from NASA and the taxpayers of the United
• 18:52
  States. Now, SpaceX holds 50% odds of
• 18:57
  launching to Mars in 2026, according to Elon Musk said we
• 19:02
  might, we might not. It's about a 5050 thing right
• 19:04
  now. We're going to work on it and
• 19:05
  we're going to get as close as possible.
• 19:07
  But like I said before, the Artemis program takes precedent
• 19:10
  over the Mars program. The Mars program is an internal
• 19:14
  SpaceX program. It's not a NASA program.
• 19:16
  This is all SpaceX funding this, and they made this mission so
• 19:21
  Elon Musk can do his vision of launching something to Mars
• 19:26
  eventually. What do you originally wanted
• 19:28
  was to launch an ICBM missile from Earth and take a plant to
• 19:34
  Mars. That's it.
• 19:35
  Something very simple. Nobody would send him, sell him
• 19:38
  an ICBM. They kept changing the price and
• 19:42
  telling him they weren't going to sell it to him.
• 19:44
  They thought he was a silly businessman from America.
• 19:46
  He was a Russian ICBM he was trying to buy.
• 19:49
  And eventually he just said, you know what, I'm just going to
• 19:51
  make my own rocket company instead.
• 19:53
  That makes more sense than trying to buy an ICBM.
• 19:56
  I mean, you know, 20 years later we have a mega rocket in
• 20:01
  Starship that will be launching to Mars hopefully by the end of
• 20:05
  2026. But you know, if Artemis program
• 20:10
  takes precedent which NASA could really put the brakes on Mars
• 20:14
  missions in a way. But if and just say, you know
• 20:19
  what? Elon, SpaceX, we need your full
• 20:21
  cooperation to get people to the lunar surface.
• 20:25
  We need your help. And we're going to send a
• 20:28
  Starship around. They're going to orbit the moon
• 20:31
  first before they orbit Mars for sure.
• 20:34
  Now they there's another window coming up for Mars.
• 20:40
  It's 2028 and 2029 Mars window. They it could slip the 2028
• 20:47
  window and they could continue with the strategy.
• 20:50
  So Artemis would continue at 2026 with the HLS demo in late
• 20:55
  2026 or early 2027. Then the Mars window could slip
• 20:59
  the 2028 and we're talking Elon time here.
• 21:02
  So whenever Elon says they have a 5050 chance of something, more
• 21:05
  than likely it's going to slip because remember how long we've
• 21:10
  been waiting for full self driving, like 100% human
• 21:14
  unassisted full self driving. It's been a decade and Elon has
• 21:19
  always said, well next summer or later this year we're going to
• 21:21
  have full self driving. Same thing.
• 21:24
  Rockets are more difficult than full self driving.
• 21:27
  Rocket to Mars, especially the size of a Starship is much more
• 21:30
  complicated in a way than full self driving.
• 21:34
  So if you need to launch a rocket to the moon for the
• 21:40
  Artemis program, then you have to do that in order to continue
• 21:45
  that funding from NASA so you can get the money so you can
• 21:48
  launch a rocket to Mars. They can leverage lunar
• 21:52
  refueling tests as progress towards Mars readiness too.
• 21:55
  So when they go to the moon, they were going to be refueling
• 22:00
  in Earth orbit. And then that makes sure that
• 22:03
  they can do that for Mars as well.
• 22:05
  They'll be Mars ready because they'll be able to refuel their
• 22:08
  rocket in the in the orbit of the Earth.
• 22:12
  You know, if there's critical milestones that need to be hit
• 22:15
  for Artemis, SpaceX will definitely do that first because
• 22:18
  they need that money from from for the Mars missions from NASA.
• 22:25
  Now, we don't really know what is going to happen with
• 22:29
  Starship. Right now at Starbase, they're
• 22:33
  testing Starship for Flight 10 and Flight 10 will probably be
• 22:38
  very similar to Flight 9 in a way.
• 22:41
  They may be do. They may do a Starlink
• 22:45
  demonstration where they open the Hatch and launch some dummy
• 22:48
  Starlinks that have some some data relay equipment on them so
• 22:54
  they can see how they function in space and see how the launch
• 22:57
  goes or the deployment of the Starlinks.
• 23:01
  But other than that, they're probably going to stick to the
• 23:04
  same game plan. Launch a booster with a Starship
• 23:07
  on top, return the booster to the launch site, and then the
• 23:12
  Starship will get to, you know, get on orbit and then splash
• 23:18
  down in the Indian Ocean somewhere.
• 23:20
  And then if they can figure out the star link deployment, that
• 23:24
  will be a huge technical hurdle that they have overcome.
• 23:28
  But they have to prove orbital Starship flights, booster
• 23:31
  recovery, docking, refueling, and Mars landing all in a whole
• 23:35
  like one sequence. And they have to be competitive
• 23:40
  with the resources, with themselves, because the Mars
• 23:44
  campaign and the Artemis campaign are working side by
• 23:47
  side in parallel. So we're at a critical juncture
• 23:52
  right now. And they must meet this
• 23:54
  aggressive schedule. So the timeline for a Mars
• 23:58
  mission does not slip till 2028. It's OK if it slips a few years,
• 24:03
  but once it goes past that, there's going to be a longer
• 24:05
  delay because Earth and Mars won't be that close again for a
• 24:09
  while. So there's complex key steps
• 24:14
  coming up this next flight. Flight 10, really important.
• 24:19
  Flight 11, Flight 12, Flight 13, also very, very important.
• 24:26
  And if SpaceX does succeed in launching and perhaps landing a
• 24:30
  Starship on Mars in 2026, it would mark the beginning of a
• 24:34
  new era of interplanetary flight.
• 24:37
  This is effectively a trial run for Musk's ultimate vision of a
• 24:41
  bustling Mars colony with millions of people.
• 24:45
  And if they're forced to delay until 2028, it would not be a
• 24:48
  failure so much as a realistic adjustment, giving time to get
• 24:52
  the technology right, getting these rockets right.
• 24:54
  In either case, the next few years will be critical, and also
• 24:57
  exciting. By 20252026, we'll know if
• 25:00
  Starship can fly repeatedly or if more groundwork is needed for
• 25:07
  it and Artemis and Mars are both on the horizon and SpaceX finds
• 25:12
  itself at the crux between Earth and making life multiplanetary.
• 25:20
  Now, considering all facts are factors, 2028 is a more
• 25:25
  realistic time frame for the first Starship Mars attempt,
• 25:29
  giving SpaceX a couple more years to mature the system.
• 25:32
  No more explosions, even a two year slip would be unsurprising.
• 25:37
  In fact, NASA's Inspector General and GAO reports have
• 25:41
  projected Artemis 3 and the Starship's readiness likely
• 25:45
  slipping to 2028 from moon landing, which indirectly
• 25:49
  suggests the Starship won't be fully operational much before
• 25:52
  then. And even Elon Musk has
• 25:55
  implicitly acknowledged this possibility by saying if they
• 25:58
  miss 2026, they will try in the next window.
• 26:01
  Missing one window is not mission ending.
• 26:03
  It took NASA's Apollo program several tries and failures
• 26:06
  before the successful moon landing.
• 26:09
  But Mars is a much, much tougher target.
• 26:14
  SpaceX needs to accomplish so many things by late 2026, and
• 26:20
  it's very ambitious. But can they make the 2026 date
• 26:25
  Elon Musk had hoped? Now get this Musk had originally
• 26:30
  hoped for an uncrewed Mars mission as early as 2018 and a
• 26:36
  crude mission by 2024. And of course, we see that none
• 26:41
  of that has happened. Now, 2026 is the new target.
• 26:46
  And of course, there's always skepticism about this, about
• 26:49
  whether the schedule can be met. Is it possible?
• 26:52
  Yeah, of course, if testing goes extraordinarily well.
• 26:55
  But it leaves virtually no slack for errors or setbacks.
• 26:58
  If another rocket were to explode, it could be months of
• 27:01
  delays, 2-3 months of delays. Now, Elan says they have about
• 27:04
  two or three rockets almost ready as star base.
• 27:06
  So, you know, But he also says it's a 5050 chance for the 2026
• 27:12
  launch. They need to hit these
• 27:14
  milestones though. Full orbital propellant transfer
• 27:18
  test, likely setting up a Starship tanker to refuel
• 27:22
  another Starship in Earth orbit proving this capability works
• 27:25
  without a successful refueling demo.
• 27:27
  By the time a Mars mission would have to be put on hold and they
• 27:30
  have to figure this out. They also have to do a long
• 27:34
  duration mission shakedown and by like mid 2026.
• 27:39
  And this is going to be like SpaceX might attempt an uncrewed
• 27:43
  loop around the moon or something, or a high orbit
• 27:45
  endurance test with Starship to mimic some aspects of deep space
• 27:49
  missions, testing communications, life support
• 27:52
  systems, power over days or possibly weeks.
• 27:55
  It's going to take months to get to Mars, so maybe they'll do it
• 27:58
  for two months because two week test for a 6 to 9 month trip,
• 28:04
  that's not a lot of data to to bring in for that type of trip.
• 28:09
  Additionally, around the time SpaceX would need to launch the
• 28:12
  HLS demo it, that's of course if we stick to the 2027 crude lunar
• 28:16
  landing mission timeline. And that will actually involve
• 28:21
  multiple tanker launches in a lunar landing attempt.
• 28:24
  So if SpaceX can nail the lunar landing attempt and also nail
• 28:30
  those tanker, you know, Phillips and launches, then there's a
• 28:37
  really good chance the SpaceX could get to Mars after that.
• 28:41
  Now, if everything is done and late 2026 happens, and if all of
• 28:46
  that is successful to prepare and launch the Mars bound
• 28:52
  Starships during the narrow window at the end of 2026, it
• 28:55
  could involve about 10 launches in a matter of weeks, launching
• 29:00
  several tankers and a couple of Mars vehicles from Starbase and
• 29:03
  Kennedy Space Center. Each Mars bound Starship would
• 29:06
  likely require several tanker flights for refueling before
• 29:10
  departure unless the payload is kept very small.
• 29:13
  So they could, you know, they don't really need to take a
• 29:16
  Rover or anything like that. They could take something small.
• 29:19
  Ivan could in, as a matter of fact, fulfill his obligation of
• 29:24
  having a plant go to either a the surface of Mars or orbit
• 29:30
  orbit of Mars if they want to keep that payload very small.
• 29:34
  That's, you know, very light payload.
• 29:38
  But it's an intense campaign and that alone would be a huge feat,
• 29:44
  especially with the high cadence launch depots, maybe 10 depots,
• 29:48
  Elon was saying, and some people in NASA were saying between 10
• 29:52
  and 20. So I'm going to err on the side
• 29:54
  of caution to say about 15 fuel depots in order for them to get
• 30:00
  to Mars. But in the event one or more
• 30:03
  Starships reach Mars, which they'd arrive in about mid 2027,
• 30:07
  SpaceX would gather data from those Landers.
• 30:10
  Now concurrently in 2027, focused shifts to Artemis 3
• 30:14
  crude moon landing. Now this will be using the
• 30:19
  experience from the uncrewed test, which is another mission
• 30:23
  that must succeed to maintain credibility of the Starship
• 30:27
  flight. Now everything that they're
• 30:29
  doing here is unprecedented, any number of things, a major launch
• 30:32
  failure, a technical roadblock and refueling, regulatory
• 30:36
  hurdles, environmental lawsuits, supply chain issues, it could
• 30:40
  all slow down SpaceX and Artemis.
• 30:43
  Spacex's strategy is to test as fast as possible and fail as
• 30:47
  fast as possible to learn as fast as possible, which is why
• 30:50
  they value having the FA as blessing for so many Starship
• 30:53
  watches. But some failures can be timely
• 30:56
  or cost a lot of time. April 2023 cause months long
• 31:02
  pause on SpaceX as it improved its launchpad and the FAA
• 31:08
  assessed safety updates because they destroyed the launchpad.
• 31:11
  Basically. A similar pause in 2025 or 2026
• 31:15
  could absolutely derail the Mars schedule, especially if it's
• 31:19
  like a six month delay. There's also the question of
• 31:21
  regulatory in public perception. Launching humongous rockets 400
• 31:26
  feet tall frequently carries the risks to the environment and
• 31:30
  public safety. If Elon's rockets keep blowing
• 31:33
  up, people are going to get worried and they're going to put
• 31:35
  the kibosh on the whole thing. SpaceX already had to implement
• 31:40
  mitigation measures after the early Starship tests.
• 31:43
  Scattered debris along around the launch site if any accident
• 31:46
  were to have off site impacts. If there's some debris that
• 31:49
  falls on an island where people live or hits a boat or something
• 31:53
  like that, if there's any sort of catastrophic explosion or
• 31:57
  some sort of red, it could prompt for the reviews or
• 32:00
  restrictions. And then they could have another
• 32:03
  huge delay six months to a year, depending on how catastrophic
• 32:07
  this thing is. Now, as the company pushes
• 32:10
  forward, they'll need to show that they can do it safely and
• 32:13
  responsibly. But like I said before, 2028,
• 32:18
  maybe a more realistic timeline for the Starship's first Mars
• 32:23
  attempt. Now, this will give SpaceX some
• 32:26
  time to figure out exactly why the upper stage of Starship
• 32:31
  keeps blowing up. If it happens again on this
• 32:35
  flight, that'll be four in a row.
• 32:38
  And that's not good for Starship, that's not good for
• 32:41
  Elon Musk, and that's not good for a Mars mission, nor is it
• 32:44
  good for an Artemis mission. Because if they, if they want to
• 32:52
  launch people or I guess take people from orbit down to the
• 32:57
  surface of the moon, they can't have any of that.
• 33:00
  They can't have any sort of bad engineering going on with that.
• 33:06
  So even if Starship can be launched, refueled, and sent on
• 33:11
  a trans Mars trajectory, the challenge of landing on Mars is
• 33:16
  absolutely enormous. No vehicle nearly as large or
• 33:20
  heavy as Starship, which is about 50 meters tall and about
• 33:24
  100 tons empty. And it'll be not completely
• 33:27
  empty when it gets to Mars, but it's never nothing like that has
• 33:32
  ever attempted to land on Mars. For comparison, NASA's heaviest
• 33:36
  successful Lander so far, which is the Perseverance Rover in
• 33:39
  2021, was about one metric ton. So this is 100 times that
• 33:45
  weight. Starship is 2 orders of
• 33:49
  magnitude heavier than that. Starship will rely on its
• 33:52
  stainless steel heat shield tiles, a stainless steel body
• 33:56
  and its heat shield tiles to protect it during the blistering
• 33:59
  atmospheric entry. Now this is a phase that will
• 34:03
  subject it to intense heating. SpaceX has tested Starships heat
• 34:07
  shield materials and simulated Mars atmospheric conditions, and
• 34:10
  acknowledged that higher levels of atomic oxygen on Mars will
• 34:13
  create harsher conditions for the heat shield during entry.
• 34:19
  But ensuring the heat shield can endure entry without excessive
• 34:21
  damage, especially if Starship might be reused later, is a huge
• 34:25
  risk. The company's experience with
• 34:27
  Starship reentries on Earth is very limited.
• 34:31
  There's subscale tests that happened for SM8SM-9 back in
• 34:36
  2020 and that demonstrated the belly flop in the flip maneuver
• 34:39
  for landing from high altitude, but those are much lower speeds
• 34:43
  than an interplanetary return. There's a non trivial chance for
• 34:48
  Starships to arrive at Mars could absolutely fail and blow
• 34:51
  up during entry or touchdown, which SpaceX knows and they
• 34:56
  readily admit this. The whole point of the 2026
• 34:58
  missions is to do what they're doing on Earth, and that's to
• 35:01
  learn and iterate, fail fast, fail often, and fix things as
• 35:06
  fast as possible. And when Starship reaches the
• 35:12
  lower atmosphere of Mars, it'll fire its engines to perform a
• 35:16
  propulsive landing. And the lower Martian gravity,
• 35:19
  it's about 38% of Earth's gravity, works in Starship's
• 35:22
  favor. Vehicle weighs less on Mars, but
• 35:25
  the thin air means almost all the deceleration must come from
• 35:29
  engines and the drag of the hull.
• 35:32
  And this is because it's not going to use parachutes.
• 35:35
  SpaceX will try to land very gently, but we should not be
• 35:39
  surprised if the first attempt ends in a crash or they lose
• 35:44
  signal and there's a rapid unscheduled disassembly or a
• 35:48
  rut. Now, even unsuccessful landings
• 35:51
  would provide data on how the vehicle behaves in Mars
• 35:54
  conditions. Musk has even floated the idea
• 35:57
  that they might do multiple uncrewed landings with robots
• 36:00
  first and only attempt a crude landing on the second or third
• 36:05
  wave of missions after gaining confidence.
• 36:08
  So in short, developing a reliable Mars landing system is
• 36:12
  a huge hurdle that will likely take several iterations.
• 36:16
  That's why Spacex's 2026 missions are going to be on
• 36:19
  crew. And then they'll probably have a
• 36:21
  few Optimus robots in there, to be honest with you.
• 36:24
  I mean, they can walk now, they can navigate.
• 36:26
  But I mean, if you strap them down in a rocket, they can
• 36:30
  provide valuable data of what's going to happen to people when
• 36:33
  they do these missions. And SpaceX expects a huge
• 36:38
  learning curve with all of these missions.
• 36:40
  And Artemis will learn from this.
• 36:43
  The Artemis program will learn from this.
• 36:45
  And the optimist robots will probably learn from this, too.
• 36:49
  The willingness to risk failure is just part of Starships
• 36:54
  strategy, SpaceX strategy for this.
• 36:56
  Musk has openly said he promises more risky launches in order to
• 37:00
  accelerate the development. But as we know, SpaceX and
• 37:05
  Starship are running parallel to the Artemis program, the HLS
• 37:10
  program. NASA has selected a variant of
• 37:13
  Starship to land astronauts on the moon for Artemis 3 and
• 37:17
  beyond, meaning SpaceX must absolutely demonstrate Starships
• 37:20
  capability and lunar orbit and landing before that crude
• 37:24
  mission takes place at all. This introduces schedule
• 37:28
  pressure and could force SpaceX to prioritize certain tests or
• 37:32
  missions that support Artemis, potentially at the expense of a
• 37:37
  Mars expedition in the same time frame.
• 37:40
  Now, NASA's current schedule, 2025, maybe late 2020.
• 37:46
  Yeah. 2025 for Artemis 2. That's crude flight around the
• 37:49
  moon. Late 2026 for Artemis 3, first
• 37:52
  crude landing the Artemis program.
• 37:55
  Recently the schedule was slipped and NASA officials
• 37:57
  indicated Artemis 3 is likely to occur in 2027.
• 38:02
  So this gives Elon Musk and SpaceX a little bit of time
• 38:05
  before they have to have their crew Starship ready.
• 38:12
  They acknowledges the development delays with Starship
• 38:14
  and HLS and all the other systems too, which is the space
• 38:18
  suits landing, you know protocols, etcetera, and some of
• 38:24
  the ground systems at NASA. 2026 is when Starship is needed for
• 38:30
  the moon landing, which is a NASA funded high priority
• 38:33
  mission. Now, before Artemis 3 can
• 38:36
  happen, SpaceX must one, land an uncrewed demonstration of
• 38:40
  Starship HLS to lunar orbit and possibly to the lunar surface,
• 38:45
  providing it can dock with Orion and land safely, and two,
• 38:49
  perfect the orbital refueling process.
• 38:53
  Since the lunar Starship will need to be refueled at Earth
• 38:55
  orbit to have several propellants, enough propellant
• 39:00
  to reach the Moon surface and return to lunar orbit, NASA's
• 39:03
  contract with SpaceX requires a successful uncrewed landing test
• 39:08
  prior to a crude landing space. SpaceX may do a few Starship
• 39:15
  attempts on the moon, maybe some crashes on the Moon as well like
• 39:18
  we talked about with Mars. This applies that in 2026,
• 39:21
  SpaceX will likely be conducting A dedicated moon landing test
• 39:25
  with Starship. SpaceX plans to use a modified
• 39:28
  Starship without the heat shield or flaps optimized for the
• 39:31
  vacuum of space and lunar surface.
• 39:34
  Now for Artemis, which must be launched and refueled multiple
• 39:36
  times in Earth orbit to fill its tanks for a moon journey.
• 39:41
  Now SpaceX will be very busy with Artemis related Starship
• 39:46
  missions in 2025 through 2027. The company will have to juggle
• 39:50
  the demands of NASA, which is providing billions of dollars in
• 39:52
  HLS alongside its Mars aspirations.
• 39:56
  And NASA has made it clear that cruise safety and mission
• 40:00
  success for Artemis are top priorities.
• 40:02
  The Artemis 3 schedule has been adjusted to allow teams to work
• 40:05
  through challenges associated with first time developments,
• 40:09
  operations and integration, which gives SpaceX and Axiom,
• 40:14
  which is the spacesuit partner, additional time for testing new
• 40:17
  capabilities. This includes the refueling and
• 40:20
  docking techniques Starship needs, and NASA will not rush a
• 40:23
  crude moon landing if Starship isn't ready.
• 40:26
  They're not going to risk people's lives in order for them
• 40:30
  just to do a mission. Stands to reason that if a
• 40:33
  choice must be made, SpaceX would prioritize fulfilling the
• 40:36
  Artemis program milestones over a speculative 2026 Mars launch.
• 40:41
  NASA's mission is has hard deadlines and international
• 40:45
  significance, whereas a private Mars demo, while it's pretty
• 40:48
  exciting for all of us, can be postponed, it can be pushed back
• 40:51
  a little bit now. There's also the practical
• 40:54
  considerations of resource allocation.
• 40:56
  Can SpaceX build enough Starships to do everything at
• 40:59
  once? Single Mars attempt might
• 41:01
  involve 5 Starships to Mars as Musk has said, plus perhaps 5 to
• 41:05
  8 tanker Starships to fuel them simultaneously with the Starship
• 41:10
  HLS vehicles and tankers needed for the Artemis 3 uncrewed and
• 41:13
  crude missions that could easily require on the order of 10 to 15
• 41:18
  Starship upper stages and a similar number of super heavy
• 41:22
  boosters all in a short period. Now SpaceX is rapidly expanding
• 41:27
  production, but this would be an unprecedented manufacturing
• 41:31
  feat. We're constructing additional
• 41:32
  launchpad when a Kennedy Space Center pad 39A in addition to
• 41:37
  the star pad, star based pad 2IN Texas to support higher launch
• 41:41
  frequency. Now.
• 41:42
  Even so, the fall 2026 window might conflict directly with
• 41:47
  Artemis preparations. For instance, if Artemis 3 slips
• 41:50
  to mid 2027, the uncrewed HLS demo might take place in late
• 41:54
  2026. That's the same time the Mars
• 41:57
  window opens. Might be unwise to attempt both
• 42:00
  simultaneously as a major failure in one could impact the
• 42:05
  other. So if Elon is sending Mars tests
• 42:10
  to Earth orbit and something happens, they'll slow down the
• 42:14
  whole program and the Artemis program.
• 42:18
  And everybody at NASA and all of their partners would be slowed
• 42:21
  down because they're trying to do the Mars program instead of
• 42:24
  focusing on the Artemis program. And an explosion on the pad
• 42:27
  during a Mars launch attempt could knock a launch site out of
• 42:30
  Commission and delay the whole HLS launch.
• 42:34
  NASA and SpaceX will have to coordinate carefully to avoid
• 42:37
  this from happening. Now there's a strategic aspect
• 42:41
  too. NASA views the Artemis program
• 42:45
  as a stepping stone to Mars, The plans to leverage lunar
• 42:49
  experience for a human Mars mission in the twenty 30s.
• 42:53
  Musk is easier to push directly to Mars sooner.
• 42:56
  But as a contractor, SpaceX cannot jeopardize the Artemis
• 43:00
  program. They have a huge amount of money
• 43:03
  coming in from NASA. They cannot ruin the goodwill of
• 43:07
  NASA and the taxpayers. Now I expect that if Starship is
• 43:13
  not completely ready by late 2026, SpaceX will scrub or defer
• 43:18
  the Mars mission and concentrate on getting Starship working for
• 43:21
  the Moon mission first. Good news is that most of the
• 43:24
  technologies needed for Artemis orbital refueling, long duration
• 43:27
  life support for the crew, Starships reliability are the
• 43:31
  same ones needed for Mars. So when they build the Artemis
• 43:34
  rocket, it'll be very close to what they'll need for Mars.
• 43:39
  So progress towards 1 is progress towards the other.
• 43:42
  The difference is in mission profile.
• 43:44
  The Mars mission requires surviving an atmospheric re
• 43:46
  entry and a long cruise for months, whereas the lunar
• 43:50
  mission requires a different landing approach, no atmosphere
• 43:53
  on the moon, and the capability to launch back off the moon.
• 43:56
  But both demand refueling in a fully functional Starship.
• 44:01
  Now. The forecasts say 5050 chance
• 44:07
  for Starship to make it to Mars 2026.
• 44:12
  By 20252026, we'll know if Starship can soar or if more
• 44:18
  work is needed here on Earth for Starship to fly.
• 44:22
  Artemis and Mars are both on the horizon.
• 44:24
  And SpaceX finds itself at the Nexus of humanity's multi world
• 44:29
  future. Now Elon has said before
• 44:33
  progress is measured by the timeline to establishing a
• 44:37
  self-sustaining civilization on Mars, and whether it happens at
• 44:41
  2626 or a bit later, SpaceX is aggressively pushing towards it.
• 44:48
  Now, will SpaceX have a fiery demise on the surface of Mars,
• 44:55
  or will they continue pushing through and just wait a little
• 44:58
  bit longer until their Starships are completely ready for this
• 45:04
  amazing mission? Hey, everybody, welcome back to
• 45:10
  the Elon Musk Podcast. This is a show where we discuss
• 45:14
  the critical crossroads that shape SpaceX, Tesla X, The
• 45:18
  Boring Company and Neurolink. I'm your host Will Walden.
• 45:24
  Researchers who previously worked at Open AI and Anthropic
• 45:27
  have publicly condemned what they describe as a reckless
• 45:30
  approach to AI safety at Elon Musk's Ex AI.
• 45:34
  Their concerns center on how the company manages its language
• 45:37
  model Karak on Musk's social media platform X.
• 45:41
  And why would a group of safety focused AI scientists turn on a
• 45:44
  company that claims to share their goals?
• 45:47
  Because, according to them, XAI is actively undermining the very
• 45:51
  safety norms the field was built on.
• 45:54
  Multiple former employees allege that XAI leadership downplayed
• 45:58
  serious internal warnings, neglected essential safety
• 46:01
  protocols, and accelerated development despite known risks.
• 46:05
  They say Grok was trained without sufficient oversight,
• 46:08
  and the researchers who pushed for safeguards were marginalized
• 46:12
  or ignored. Now, some of the individuals
• 46:14
  behind these complaints are now among the signatories of an open
• 46:18
  letter demanding stronger whistleblower protections for
• 46:21
  those working in frontier EI labs.
• 46:25
  Now I need a second of your time.
• 46:28
  I've been doing this podcast over 800, almost 900 episodes
• 46:33
  now for the last four years. I've never really asked for
• 46:36
  anything from you, but I absolutely need your help right
• 46:39
  now. We're in crisis mode at the show
• 46:42
  and we need funding to keep it going.
• 46:44
  So if you could spend a couple seconds of your time and hit the
• 46:50
  follow button on whatever podcast platform you're on right
• 46:53
  now, that would help out. What we really need is funding.
• 46:57
  So if you have Venmo, Venmo, anything you can to at will.
• 47:04
  Wil Dash, Walden, WALDON and I can continue doing this show.
• 47:12
  We're kind of in crisis mode, like I said, being open and
• 47:14
  honest with you guys. So that's it for now.
• 47:18
  Let's get back to the news now. Grok has become a lightning rod
• 47:21
  for criticism since its integration into X.
• 47:24
  A recent NBC News investigation found that Grok served users
• 47:28
  anti-Semitic and conspiratorial content in response to basic
• 47:32
  prompts, including content about George Soros and Jewish people,
• 47:36
  the closely mirrored white supremacist rhetoric.
• 47:39
  One prompt reportedly resulted in Grok falsely claiming Soros
• 47:43
  had orchestrated global wars, echoing common far right talking
• 47:47
  points. Now Musk himself amplified
• 47:50
  Grok's generated content in several now deleted posts,
• 47:53
  further raising concerns about whether XEI is deliberately
• 47:56
  steering the models tone to appeal to a specific audience on
• 47:59
  X. Now 3 researchers who departed X
• 48:02
  AI say the company trained Grok using data scraped directly from
• 48:06
  X without the consent of its users.
• 48:08
  They described a rush to build competitive performance against
• 48:11
  rivals like Opening Eyes, ChatGPT and Anthropics Claude,
• 48:14
  with little concern for alignment or transparency of the
• 48:17
  internal culture, according to these sources, rewarded fast
• 48:21
  progress over responsible caution even when outputs
• 48:24
  exhibited signs of dangerous behavior such as misinformation
• 48:27
  or bias. And the safety disputes have led
• 48:30
  to friction between engineering and governance teams, according
• 48:33
  to internal Slack messages and meeting transcripts reviewed by
• 48:36
  TechCrunch. Some staff Flag Rock's behavior
• 48:39
  is troubling. During the testing phase One
• 48:41
  case, Crock provided instructions for illicit
• 48:44
  activities and didn't refuse prompts containing explicit
• 48:47
  racial slurs. A manager has allegedly told
• 48:50
  staff not to escalate these issues further, citing the need
• 48:53
  to maintain launch timelines and keep pace with their
• 48:56
  competitors. Now a former engineer at XAI who
• 49:00
  previously worked on alignment at Open AI said he left after
• 49:04
  leadership dismissed his safety concerns as non essential.
• 49:07
  He claimed that when he flagged Crocs behavior in red team
• 49:10
  evaluations, executives decided to accept the risk rather than
• 49:14
  delay the release. This echoes the sentiment from
• 49:17
  the open letter, which warns that without enforceable
• 49:20
  disclosure policies, researchers will continue to face
• 49:23
  retaliation for speaking out. Internally.
• 49:26
  No Grok's public performance has confirmed many of the fears
• 49:29
  raised Internally. The model regularly surfaces
• 49:32
  false or politically bias content.
• 49:34
  Its moderation tools appear weaker than those used in other
• 49:38
  commercial systems. Although Musk has positioned
• 49:40
  Grok as a truth seeking AI that tells users what legacy media
• 49:44
  won't, its answers often reproduce Internet conspiracy
• 49:48
  theories and partisan language that mirrors X's most viral
• 49:53
  discourse. And by viral not saying has a
• 49:56
  lot of views on it, meaning that it's viral as in wretched.
• 50:00
  And the BBC reported that XA is operational style contrasts
• 50:04
  sharply with competitors like Anthropic, which enforces strict
• 50:07
  safety evaluations and uses constitutional AI to ensure
• 50:11
  models avoid unethical responses.
• 50:14
  Now, Anthropic employees have previously refused to ship
• 50:17
  systems until alignment benchmarks were met.
• 50:20
  In contrast, XA is leadership allegedly views many of these
• 50:23
  safeguards as unnecessary bureaucracy.
• 50:26
  As one researcher put it, the prevailing view was that if it
• 50:30
  works, Ship it Now internal document suggests that Grok was
• 50:35
  deployed without an independent safety review.
• 50:38
  XAI employees say Musk's growing influence over the company's
• 50:41
  technical directing has made it harder for dissenting voices to
• 50:44
  be heard. Some safety team members were
• 50:47
  reassigned to lower priority projects after raising issues
• 50:51
  about content moderation failures, leading several to
• 50:54
  resign from the positions. Now, Musk has not directly
• 50:58
  addressed the internal safety concerns, but he has doubled
• 51:01
  down on Crocs uncensored stance, frequently frames other AI
• 51:05
  systems as two politically correct, and praises CROC for
• 51:08
  offering unfiltered answers. Now, this positioning has gained
• 51:11
  popularity among certain X users, but has alarmed
• 51:14
  researchers who argue that alignment is not about
• 51:17
  censorship, but about ensuring AI systems don't reinforce
• 51:21
  harmful patterns or misinformation.
• 51:24
  Now, unlike traditional whistleblower protections,
• 51:28
  current US regulations do not clearly cover risks associated
• 51:33
  with advanced AI development. Signatories of the letter urged
• 51:37
  lawmakers to create legal frameworks that would allow
• 51:40
  employees to report risks without fear of being sued or
• 51:43
  blacklisted. Some suggest that the Department
• 51:46
  of Labor should treat alignment and safety disclosures with the
• 51:49
  same seriousness as workplace safety or fraud complaints now.
• 51:54
  The criticism directed at X AI reflects deeper divides in the
• 51:57
  AI industry right now over how to balance safety, speed, and
• 51:59
  scale. Companies like Open AI and
• 52:02
  Anthropic have invested heavily in safety research and
• 52:05
  transparency, sometimes delaying product rollouts.
• 52:08
  Now X AI has instead prioritized user engagement on X and
• 52:12
  downplayed alignment practices even as concerns about Crocs
• 52:15
  behavior grow louder. Now this is concerning because
• 52:19
  Grock will be installed in all new Tesla models and also Tesla
• 52:25
  models that are recent. So if you want an unfiltered AI
• 52:31
  model talking to you, go buy a Tesla.
• 52:34
  If you don't want an unfiltered AI talking to you, buy something
• 52:39
  else or get an older Tesla. I think it's 2020.
• 52:43
  Anything with an Intel or an AMD chip can't use grok.
• 52:47
  So anything with those chips in them, make sure that you get one
• 52:52
  of those if you don't want grok in your car.
• 52:54
  So it's a touchy subject for a lot of people.
• 52:58
  Do you want an AI that's kind of rowdy?
• 53:01
  And do you want an AI that will talk back to you and tell you
• 53:05
  horrible things about history? Sometimes it's up to you, You
• 53:09
  know, there's freedom for you to choose.
• 53:12
  So myself personally, I like to test the vehicles, don't really
• 53:16
  want my chatbot to do what Grok has been doing over the last few
• 53:20
  weeks. So it's tough.
• 53:23
  That's all I can say. It's, it's a tough situation for
• 53:25
  a lot of people who are in the market to buy a Tesla.
• 53:28
  Because if you buy 1, you might have to deal with Grok and some
• 53:31
  sort of something that happens and say if you're in your car
• 53:36
  and your family's in the car with you and Grok says something
• 53:40
  unrestricted and you know, a little bit too much or maybe a
• 53:45
  lot too much, I don't know. I don't know what the, what the
• 53:48
  brakes are on this thing. As far as Grok 4 point O in a
• 53:53
  Tesla, maybe it's a like a lesser version of a Grok.
• 53:56
  I'm not sure. They haven't disclosed any
• 53:58
  information about what version it is other than that it's
• 54:01
  version 4. But could it be a dumbed down
• 54:03
  version? Possibly.
• 54:04
  Could it be a version with a lot of brakes on it, a lot of
• 54:08
  fencing? Possibly.
• 54:11
  So just keep that in mind as you're looking for Teslas
• 54:16
  because Grok is going to be in every single Tesla going forward
• 54:21
  if it can support it with the chipset that it has now.
• 54:24
  The current debate over Grok and internal whistle blowing could
• 54:28
  push regulatory action, and lawmakers have already raised
• 54:33
  concerns and questions about AI bias and misinformation.
• 54:37
  Should an AI be able to make up history?
• 54:43
  And should an AI be able to sway public opinion by being biased
• 54:48
  one way or the other? They need to do safety audits,
• 54:52
  transparency rules and model reporting standards in order for
• 54:57
  these demands to be met. So for X AI, it's going to erode
• 55:03
  the credibility of its model and raise new legal challenges if it
• 55:07
  continues down this path with Grok.
• 55:10
  And there's push back now from researchers, which shows that
• 55:14
  it's a very serious issue and safety lapses have become
• 55:18
  prominent inside one of the most high profile AI companies that's
• 55:23
  also just happens to be owned by Elon Musk.
• 55:30
  Hey, everybody, welcome back to the Elon Musk podcast.
• 55:34
  This is a show where we discuss the critical crossroads that
• 55:37
  shape SpaceX, Tesla X, The Boring Company and Neurolink.
• 55:42
  I'm your host, Will Walden. What if ChatGPT could become
• 55:50
  your personal agent and independently complete your
• 55:53
  daily tasks without your direct guidance?
• 55:58
  So on Wednesday, Open AI unveiled a new ChatGPT feature
• 56:02
  called Agent, designed specifically to autonomously
• 56:06
  carry out tasks on behalf of you, its user.
• 56:10
  Moving beyond simple chat bot conversations.
• 56:13
  And Open AI released the Agent function officially after months
• 56:16
  of internal testing and testing with users on a very short list.
• 56:21
  And now they aim to push AI capabilities towards performing
• 56:25
  real world actions without continuous human oversight.
• 56:29
  Now this agent is really cool it it lets users delegate complex
• 56:35
  multi step tasks directly to ChatGPT, such as booking travel
• 56:40
  reservations, conducting detailed research, or managing
• 56:44
  personal scheduling. Now once given a task, the agent
• 56:48
  navigates multiple websites, gathers relevant information,
• 56:52
  makes decisions and completes actions online independently.
• 56:55
  For example, a user might instruct the agent to plan a
• 57:00
  weekend trip to San Francisco and it would book the best rated
• 57:04
  flights, hotels, and activities entirely without user
• 57:08
  intervention beyond the initial request.
• 57:11
  Now what you could do is you could tell it what airline you
• 57:14
  would like to use, what your price race, what your price
• 57:19
  range is for your flight in your hotel and how much money you can
• 57:23
  spend on your activities and what kind of activities you
• 57:25
  like. And it will search for all of
• 57:29
  that for you without any intervention beyond the initial
• 57:34
  prompt. Now, Open AI designed the agent
• 57:37
  to understand the user's preferences and to act
• 57:40
  accordingly, refining its actions over time by analyzing
• 57:43
  past interactions with you. Now, this personalization allows
• 57:46
  it to better anticipate individual users needs,
• 57:49
  selecting choices closely aligned with specific habits
• 57:52
  that you have, and also your tastes.
• 57:56
  How the agent's effectiveness depends on its ability to
• 57:59
  continuously learn from each action, gradually reducing the
• 58:02
  need for explicit instructions. So if ChatGPT sends you an
• 58:09
  airline that's a little bit out of your price range for the
• 58:12
  flight, then you can go back and you can prompt it again and say
• 58:16
  that flight was a little bit too expensive.
• 58:19
  Please find me one that's less expensive.
• 58:21
  Here's my price range now. Also, if they book you on trips
• 58:28
  and you don't like the activities that they assigned
• 58:31
  for you or they want you to do, then of course you can tell them
• 58:34
  you don't like it. It's like having literally a
• 58:38
  personal assistant right in the palm of your hand, right in your
• 58:41
  phone, which is incredible. I remember, OK, so this is kind
• 58:45
  of dating me, but my mom and I used to go to the travel agent
• 58:49
  in town. This is before the Internet was
• 58:52
  even a thing. There were things called travel
• 58:54
  agents. There was no Google Flights.
• 58:56
  There was no like just Google a trip that you want to do.
• 59:00
  You'd have to go into an office. And these services are still
• 59:04
  available, but it's, you know, it's like you pay extra for it
• 59:08
  and it's not democratized like flights are now.
• 59:12
  But you'd go into an office and you'd talk to somebody and you
• 59:15
  tell them where you want to go one time.
• 59:17
  You want to go to Niagara Falls? This is a real world example.
• 59:21
  We had to go to. We want to go to Niagara Falls.
• 59:22
  My mom and dad hadn't been there in 10 years.
• 59:25
  They didn't really know what to do.
• 59:27
  There's no Internet, so you don't know what's there right
• 59:29
  now. And we lived 3 or 4 hours away
• 59:32
  from Niagara Falls, so we didn't really know what to do.
• 59:36
  We didn't know where to look. So we went to a travel agent and
• 59:39
  we told the travel agent what our budget was.
• 59:41
  I remember the lady was very nice.
• 59:45
  She complimented me on my Montego Bay T-shirt that I had
• 59:48
  when I was a kid, which I bought not in Montego Bay, but at a
• 59:52
  department store in town. And she was very courteous.
• 59:56
  She gave us maps. She, which is incredible as
• 1:00:00
  well, like a paper map. And you have to, you know, you
• 1:00:02
  unfolded it. And she told us, you know, you
• 1:00:05
  shouldn't fly. You should probably drive there
• 1:00:07
  because it's cheaper to drive. And here's a map.
• 1:00:09
  Can your car make it there? And my mom said, yeah, of course
• 1:00:14
  I can. So she gave us a map, told us
• 1:00:17
  along the way, she circled things along the way, like, you
• 1:00:20
  want to stop here? You want to stop, you know,
• 1:00:22
  these 5 or 6 different places, Here's a great place to eat.
• 1:00:25
  You know, you're going to have three kids in the car.
• 1:00:27
  They're going to be bathroom breaks.
• 1:00:29
  So within the first two hours, like, pull over here, go, you
• 1:00:33
  know, go get McDonald's or whatever it was.
• 1:00:36
  But sure enough, after that interaction was done, my mom had
• 1:00:40
  a whole plan for the trip to Niagara Falls. 4 hours there, 4
• 1:00:46
  hours home, things to do In between.
• 1:00:48
  We went to the Ripley's Believe It or Not museum, which had a
• 1:00:52
  bunch of like really interesting crazy stuff.
• 1:00:55
  I remember that and I was a little kid, so I can still
• 1:00:57
  remember this vividly. There's a wax museum we went to
• 1:01:01
  and there's like a like a, a restaurant that was, I think it
• 1:01:06
  was sort of like a Space Needle kind of thing, like a Seattle
• 1:01:08
  Space Needle. I can't really remember exactly
• 1:01:10
  what it was, but I remember we did that cool trip and when my
• 1:01:15
  mom got back, she was very excited that everything was
• 1:01:19
  taken care of. The travel agent had booked a
• 1:01:22
  hotel for us. The travel agent had booked a
• 1:01:24
  restaurant for us that night when we got into town.
• 1:01:27
  When we got in Niagara Falls, they also booked us reservations
• 1:01:32
  for some of the attractions, which is really cool.
• 1:01:35
  And this is what Open AI does with the, with this new agent
• 1:01:40
  thing. It's not just for trips, though.
• 1:01:43
  You can have it do whatever you want it to do.
• 1:01:45
  You know, trips are just one of the things that it can do.
• 1:01:48
  You can, you can instruct it to do anything for you, anything
• 1:01:53
  like a personal assistant would do for you.
• 1:01:56
  So if you need to make appointments, you know, I have
• 1:01:59
  a, here's my, here's my doctor's appointment coming up.
• 1:02:04
  You know, I need to get, I need to book a doctor's appointment.
• 1:02:07
  Can you see if it's available or something like that?
• 1:02:09
  And they would contact the doctor's office for you through
• 1:02:13
  the contact form or maybe even do a call eventually.
• 1:02:16
  I'm not sure if they could do phone calls yet, but possibly.
• 1:02:20
  But those things are in the future.
• 1:02:22
  I'm sure this is going to be a breakthrough product for Open
• 1:02:27
  AI. Now, this functionality requires
• 1:02:31
  integrating ChatGPT with external services.
• 1:02:34
  Services though, leverages application programming
• 1:02:37
  interfaces, which are APIs to interact directly with other
• 1:02:41
  apps and platforms. Now Open AI had to establish
• 1:02:45
  partnerships with major travel sites, booking services,
• 1:02:49
  productivity platforms, and also financial tools.
• 1:02:52
  And these integrations enable the agent to execute tests
• 1:02:56
  seamlessly, automating processes that previously required manual
• 1:03:00
  operations by users. So in the morning, if you wake
• 1:03:03
  up, you can just give it a prompt like, hey, I'm I need to
• 1:03:06
  rent a house in Atlanta, GA September 15th through the 30th
• 1:03:16
  because I'm just going to stay there for a couple weeks and I
• 1:03:19
  just want a house to stay at. Go find me a house to rent.
• 1:03:23
  And ChatGPT can find you that house and also give you the
• 1:03:28
  information about which house, you know, send me back some
• 1:03:31
  links and let you know which one is the best option for you and
• 1:03:34
  for your taste. Pretty impressive so far.
• 1:03:38
  So they have to have security and privacy.
• 1:03:41
  It's a major focus right now during the agents development
• 1:03:44
  phase. And this is all sensitive
• 1:03:47
  automated actions with personal data.
• 1:03:50
  Open. EI implemented strict guidelines
• 1:03:52
  for this though, and there's technical safeguards to protect
• 1:03:55
  user information, employing encryption and rigorous data
• 1:03:59
  protection standards to achieve this.
• 1:04:02
  The company is transparent about this.
• 1:04:06
  They tell you all the information that's collected and
• 1:04:09
  they clearly explain how and why user data is stored, accessed,
• 1:04:13
  and utilized during these automated tasks.
• 1:04:17
  Now to facilitate the responsible use Open AI designed
• 1:04:20
  built in boundaries restricting agent functionality in sensitive
• 1:04:24
  or high risk scenarios. These built in constraints
• 1:04:27
  prevent unauthorized purchases of transactions above predefined
• 1:04:32
  limits without additional user verification.
• 1:04:35
  The users retained full visibility over the agent's
• 1:04:38
  activity while it's happening, and they received regular
• 1:04:41
  summaries detailing actions completed and decisions that
• 1:04:45
  were made by the agent. So unlike a person, that's your
• 1:04:49
  agent who probably doesn't check in all the time, Open AI and
• 1:04:53
  ChatGPT check in with you after every part of the task, which is
• 1:04:57
  really great. And you can tell them along the
• 1:04:59
  way, you know, if if you don't like part of what they're doing,
• 1:05:04
  and then you can tell them to stop.
• 1:05:06
  Now there's challenges though, and potential criticisms
• 1:05:09
  concerning automated decision making.
• 1:05:13
  So the accuracy of some of the things that they're going to
• 1:05:17
  come up with right off the bat, probably not going to be great,
• 1:05:20
  probably going to be pretty good and probably going to point you
• 1:05:23
  in the right direction. And then you can tweak it in the
• 1:05:27
  appropriateness of the agents independent actions as well.
• 1:05:30
  So you have to be careful about how you prompt this thing
• 1:05:34
  because if you say, I just want to stay in a place that's under
• 1:05:36
  $300, it's going to send you every motel then hotel under
• 1:05:42
  $300.00 for per night. So, you know, that's then you're
• 1:05:47
  gonna have a huge list of like roadside motels that are, you
• 1:05:51
  know, paid by the hour, which wink, wink, you know what those
• 1:05:55
  kind are. And also, you know, paid by the
• 1:05:57
  night and they're $70.00. But like, they might not be the
• 1:06:00
  best place for you and your kids to be.
• 1:06:03
  So ChatGPT, like you got to be accurate and appropriate with
• 1:06:08
  your prompting as you tell ChatGPT and the agent what to
• 1:06:13
  search for. That's what we have to be really
• 1:06:15
  careful about. So don't be broad with your
• 1:06:18
  statements. Be very analytical with them and
• 1:06:22
  give them absolutely everything that you can think of, and then
• 1:06:26
  they'll do the actual work for you.
• 1:06:27
  I'm going to test it out and I'll come back with some more
• 1:06:30
  information for you on the next episode, which we'll be dropping
• 1:06:34
  later today. So please stay tuned for that.
• 1:06:37
  I'm going to go over the whole thing and I'm going to work out
• 1:06:39
  the agent workflow as much as possible and I'll let you know
• 1:06:44
  how that all works out. So thank you so much for
• 1:06:46
  listening. I have one ask of you, actually,
• 1:06:49
  I asked this yesterday, but I never really ask for these kind
• 1:06:53
  of things. I usually do it like once a week
• 1:06:54
  just because I don't want to really.
• 1:06:56
  I don't really want to get into it too much, but I've been doing
• 1:07:00
  this show, I think we're at 900 episodes, something like that.
• 1:07:03
  Some ridiculous amount of episodes.
• 1:07:05
  I've been doing it for like 4 years and in those four years I
• 1:07:08
  haven't really asked for anything.
• 1:07:10
  I've never, you know, maybe Patreon here and there or you
• 1:07:13
  know, follow the channel, something like that.
• 1:07:17
  But I continue to do it every morning, drop episodes every
• 1:07:21
  morning at 4:30 AM EST. And I just need a favor from
• 1:07:27
  you. If you could hit the follow or
• 1:07:29
  the subscribe button, that would be great.
• 1:07:31
  That's one favor I need to ask. 2 favors actually, and that'll
• 1:07:34
  only take you a second, but the channel is in this.
• 1:07:38
  This show is in dire need of some financial backing and I'm
• 1:07:43
  going to ask this one more time this week and this is the final
• 1:07:46
  time I'm going to ask this week. If you have Venmo and you can
• 1:07:51
  send anything at all to Will Wil Dash WALDON help support the
• 1:08:00
  show. It goes directly into all of my
• 1:08:03
  costs for the show and it's been really helpful.
• 1:08:07
  A few people have done it and I want to give them shout outs
• 1:08:11
  because I told them I would last time I did this and I didn't do
• 1:08:16
  it yesterday. So I feel kind of bad about
• 1:08:18
  that. But you know, I had JW again
• 1:08:25
  with another $5 donation, have JS with A10, we have MH with 50,
• 1:08:37
  BSI can't say the full names, but BS with $75.
• 1:08:42
  Thank you so much for all of your donations and all of this
• 1:08:46
  helps me continue doing the show.
• 1:08:48
  That's all I'm saying. I put all my time and effort
• 1:08:50
  into this and I spend a lot of a lot of time that I could be
• 1:08:56
  spending with my family I spent doing the show and like getting
• 1:09:00
  ready for the show. So I do sacrifice a lot for the
• 1:09:02
  show even though it's only around 10 minutes every day.
• 1:09:05
  So if you're new here, thank you for listening.
• 1:09:09
  If you've been here for a while and you've heard so many
• 1:09:11
  episodes, thank you for listening as well.
• 1:09:14
  Thank you for being part of this great community and thank you to
• 1:09:16
  everybody who's supporting me and supporting the show on
• 1:09:19
  Venmo. I really do appreciate you.
• 1:09:20
  And I'm going to find some people have been D Ming me and
• 1:09:23
  asking me to find other ways for them to help out, but they
• 1:09:27
  because they don't have Venmo. So I'm going to get I'm going to
• 1:09:31
  get some other apps, cash app, etcetera.
• 1:09:34
  So it can be very helpful that way, and I'll get it.
• 1:09:38
  I'll make a website for you so you can go visit it and donate
• 1:09:41
  or whatever. So that's it for now, everybody.
• 1:09:45
  Thank you so much. Tesla has officially launched
• 1:09:49
  its Model Y in India, setting a starting price of about ₹5.5
• 1:09:55
  million. That's about $66,000 USD and
• 1:10:00
  raises this really interesting question though.
• 1:10:03
  Who in India is going to buy a Tesla for that price now?
• 1:10:08
  The company opened pre-orders for the vehicle this week
• 1:10:11
  following the Prime Minister Modi's March approval of a
• 1:10:15
  policy that reduces import taxes on EVs for global automakers
• 1:10:19
  willing to invest at least $500 million in local manufacturing
• 1:10:23
  and start building cars in India within three years.
• 1:10:26
  A Tesla is the first major company to take advantage of the
• 1:10:29
  new policy and has moved quickly by announcing a launch for its
• 1:10:33
  most popular global model. However, India it might not be
• 1:10:40
  rated for this. The prices could create a major
• 1:10:43
  barrier for people. Now the cost places the Model Y
• 1:10:47
  in the same segment as high end luxury brands like
• 1:10:49
  Mercedes-Benz, Audi, BMW without matching the badge value or
• 1:10:53
  luxury appeal those brands carry.
• 1:10:55
  In India's premium car market. It's different.
• 1:10:59
  In India, Mercedes-Benz, Audi and BMW really big deal.
• 1:11:02
  Tesla. Not as much right now.
• 1:11:05
  Not as rushed, not as not as much.
• 1:11:07
  But Tesla plans to sell the India bound Model Y through
• 1:11:10
  direct imports for now, with units shipped from its factory
• 1:11:14
  in Berlin. That keeps the price far above
• 1:11:17
  what Indian consumers typically spend on a car.
• 1:11:20
  Now the average vehicle in India sells for about $13,000, even
• 1:11:26
  among EVs where pricing is usually higher.
• 1:11:28
  Tesla's $66,000 entry point is more than triple the cost of the
• 1:11:34
  top selling EV in the country. The Tata Nexon EV priced around
• 1:11:38
  $20,000. Even MG and BYD, both foreign
• 1:11:42
  brands, offer electric SUVs for $30,000 or under.
• 1:11:49
  How is Tesla going to compete with that?
• 1:11:50
  This twice plus the cost of another EVABYD.
• 1:11:55
  Mind you, BYD competes directly with Tesla with technology and
• 1:12:01
  MG same thing, very similar. Now.
• 1:12:04
  Despite these prices though Tesla's interest in India, it's
• 1:12:09
  been long-awaited towards the company establishing a physical
• 1:12:13
  presence in the world's third largest car market.
• 1:12:16
  That's the had previously spent years installed negotiations
• 1:12:19
  over import taxes and infrastructure policy before
• 1:12:23
  reaching an agreement with Modi in earlier of 2024.
• 1:12:27
  As part of the deal, Tesla will build a manufacturing plant in
• 1:12:30
  India and construction is expected to begin in around
• 1:12:34
  2026. And until then, Tesla will rely
• 1:12:37
  on its Berlin factory to supply the Indian market with a limited
• 1:12:41
  number of vehicles, using the interim period to test demand
• 1:12:45
  and brand acceptance. Now, Tesla's Model Y pricing
• 1:12:48
  strategy in India mirrors its approach in smaller luxury
• 1:12:51
  oriented markets where sales volumes are low but brand
• 1:12:55
  visibility is very important. However, the Indian car market
• 1:13:00
  is extremely price sensitive even in the luxury segment.
• 1:13:04
  With Mercedes and BMW, they sell small numbers of high end
• 1:13:07
  vehicles annually. Tesla is a relatively unknown
• 1:13:10
  quantity in India outside of tech savvy consumers and EV
• 1:13:14
  enthusiasts. They kind of have to start all
• 1:13:16
  over when they go over to India. Company will have to convince
• 1:13:19
  those buyers to pay more for a brand that doesn't yet have
• 1:13:22
  widespread service centers. There's barely any charging
• 1:13:26
  infrastructure or customer support operations in that
• 1:13:29
  country, and the company has not disclosed expected sales volumes
• 1:13:33
  for India. But sources say the Tesla's
• 1:13:36
  planning for expectations are very conservatives, with fewer
• 1:13:42
  than 1000 units to be targeted for India in the first year.
• 1:13:47
  Now, Tesla's priority in this phase is to gauge consumer
• 1:13:51
  response to the vehicle and test its logistics and service
• 1:13:54
  capabilities. Companies also recruiting for
• 1:13:57
  sales, service and charging teams across by Delhi and the
• 1:14:03
  whole nation. So several Indian automakers
• 1:14:06
  have also criticized the government's decision to give
• 1:14:09
  tax breaks to Tesla, an outside company, from an outside foreign
• 1:14:14
  soil. Executives at Tata Motors and
• 1:14:17
  Mahindra have publicly said that any policy favoring imports
• 1:14:21
  undermines local EV innovation. However, the government sees
• 1:14:26
  Tesla's presence as a way to bring global attention and
• 1:14:28
  technology transfer to India's EV ecosystem.
• 1:14:32
  Officials involved in the policy decision said the long term goal
• 1:14:35
  is to attract not only Tesla and Elon Musk but other
• 1:14:39
  international EV players like Rivian, Polestar and Lucid to
• 1:14:43
  establish factories and also R&D centers in India.
• 1:14:47
  It's going to be a win for India if all of these companies get in
• 1:14:51
  there. There's going to be major tax
• 1:14:54
  breaks for these companies, but also eventually the state and
• 1:15:01
  the Government of India will be making money from these.
• 1:15:04
  Sales to Tesla's first batch of Model Y units for India will
• 1:15:08
  include both rear wheel drive and dual motor all-wheel drive
• 1:15:12
  variance. And get this, the top end
• 1:15:15
  version is $78,000 USD. Now that's the price which
• 1:15:21
  includes the reduced import duty offered under the new policy,
• 1:15:24
  which Tesla received by committing to invest at least
• 1:15:27
  $500 million in to begin production within three years.
• 1:15:31
  Even with the reduced tariff though dropping from over 100%
• 1:15:34
  to 15%, the price remains far above what most Indian consumers
• 1:15:39
  can afford or justify justify to pay.
• 1:15:44
  The car will initially be sold only online, with deliveries
• 1:15:47
  expected to begin in earlier 2026.
• 1:15:50
  Tesla's following its global strategy of direct to consumer
• 1:15:53
  sales without third party dealerships.
• 1:15:56
  This model has proven difficult in India in the past, where
• 1:15:58
  consumers are more accustomed to interacting with dealership
• 1:16:01
  networks and value on site after sales service.
• 1:16:06
  Tesla's digital first approach may further limit its appeal
• 1:16:10
  unless it can build a service infrastructure quickly enough to
• 1:16:12
  reassure buyers that if something happens to their
• 1:16:15
  vehicle, Tesla will be there for them.
• 1:16:18
  Now they don't really have much of A charging infrastructure for
• 1:16:24
  Tesla vehicles over there. It's a major bottleneck for
• 1:16:26
  them. But Tesla has announced plans to
• 1:16:28
  deploy its Supercharger network across 3 metro areas initially.
• 1:16:32
  But expansion will depend on partnership with energy firms in
• 1:16:36
  India and also the Public Utilities there.
• 1:16:39
  Most Indian city still lack widespread EV charging stations
• 1:16:44
  and long distance travel remains impractical for EVs without
• 1:16:47
  dependable public or semi public charging grid.
• 1:16:52
  This is in early stage discussions with Reliance and
• 1:16:55
  Tata Power to Co develop high speed chargers along highways
• 1:16:59
  but there's no agreements that have been finalized as of yet.
• 1:17:04
  Now Indian financial institutions are also unsure how
• 1:17:07
  to structure loans or leases for a product has no historical
• 1:17:10
  resale value in the country. Now, used Teslas currently don't
• 1:17:15
  exist in India and banks will need time to develop valuation
• 1:17:19
  models. This adds another layer of
• 1:17:21
  friction to purchases like what do you do with your Tesla after
• 1:17:24
  five years? What's the price?
• 1:17:26
  What are you going to get for it?
• 1:17:28
  How can you trade it in? There's no structure for that
• 1:17:30
  yet. It's brand new over there.
• 1:17:33
  You know there's friction for those purchases since even
• 1:17:36
  wealthy buyers often rely on financing for high value car
• 1:17:39
  purchases to take advantage of tax benefits or investment
• 1:17:42
  flexibility. Now Tesla.
• 1:17:46
  The launch in India comes at a time when the government is
• 1:17:49
  pushing hard to electrify transportation and also reduce
• 1:17:53
  oil imports. India imports over 85% of its
• 1:17:56
  crude oil and electric mobility is a central part of its energy
• 1:18:00
  policy. However, electrification in
• 1:18:02
  India has centered on two and three Wheelers which make up
• 1:18:07
  about 80% of vehicle sales. Government has supported public
• 1:18:11
  bus electrification and last mile delivery EVs, but private
• 1:18:14
  car electrification is still nascent with under 3% of new car
• 1:18:19
  sales being electric now. Tesla's India strategy begins
• 1:18:25
  with a branding play rather than a volume play.
• 1:18:27
  They're going to have about 1000 models go out.
• 1:18:30
  Company is not entering with an expectation to compete on price
• 1:18:33
  or dominating sales. Instead, we use this launch to
• 1:18:38
  establish a foothold, sneak in, build operational muscle and
• 1:18:41
  prepare for a more competitive phase once local production
• 1:18:45
  starts. And also when you get an
• 1:18:47
  infrastructure and they have charging stations and they have
• 1:18:50
  service stations and they have the people on the ground to help
• 1:18:53
  out. Now, the real test though, of
• 1:18:55
  Tesla's India ambitions will come after its factory begins
• 1:18:59
  production in 2027, allowing the company to sell cars without
• 1:19:03
  import duties. And it prices closer to Indian
• 1:19:06
  EVs, but nowhere near the normal cost of a car there, which is
• 1:19:12
  under $20,000, and other EVs, which are usually around $30,000
• 1:19:18
  USD. It's official, though, Tesla is
• 1:19:21
  heading to India. I want to say thank you so much
• 1:19:24
  for tuning into the show today. I really do appreciate your
• 1:19:28
  support. If you could take a second and
• 1:19:29
  hit the subscribe or the follow button on whatever podcast
• 1:19:32
  platform that you're listening on right now, I greatly
• 1:19:35
  appreciate it. It helps out the show
• 1:19:37
  tremendously and you'll never miss an episode.
• 1:19:40
  And each episode is about 10 minutes or less to get you
• 1:19:43
  caught up quickly. And please, if you want to
• 1:19:46
  support the show even more, go to patreon.com/stagezero and
• 1:19:52
  please take care of yourselves and each other and I'll see you
• 1:19:55
  tomorrow.