Latest / Elon Musk Podcast / Daily Elon Update: Musk Sues OpenAI for 134 Billion Dollars

Transcript

Daily Elon Update: Musk Sues OpenAI for 134 Billion Dollars · Elon Musk Podcast

0:00
Elon Musk is asking a federal court to force open AI and its
0:04
CEO Sam Altman to redistribute up to $134 billion to charity.
0:09
Yeah, that dollar amount alone is just, it's difficult to even
0:12
process. And the entire premise of this
0:14
lawsuit really rests on this profound claim of deception.
0:18
Musk is arguing to the court that the creators of ChatGPT
0:22
pulled a massive bait and switch, right?
0:24
He claims they took a nonprofit research lab which was
0:27
originally built with the explicit purpose of benefiting
0:30
humanity, and systematically converted it into a highly
0:33
guarded, profit driven machine designed to basically enrich
0:36
themselves and their corporate partner, Microsoft.
0:39
Which is a wild accusation to make in a federal court.
0:41
It is. Just imagine you donate $40
0:44
million to build a beautiful public park and then you walk by
0:47
and find a toll booth, a massive corporate sponsor logo on the
0:50
gates and a keep out sign. That is exactly the feeling Musk
0:54
is trying to convey. We are looking at a fundamental
0:56
clash between early philanthropic ideals and the
0:59
incredibly expensive reality of developing artificial
1:01
intelligence. And as you listen, you're going
1:04
to see the hidden mechanics of how these organizations actually
1:07
operate because we have a huge stack of trial transcripts,
1:11
internal emails and financial records in front of us.
1:14
So we really have to ask the question that will frame this
1:17
entire conversation can an an early donor used the legal
1:20
system to force a tech giant to completely unscramble its
1:24
corporate structure and revert to a pure charity?
1:27
To understand that, we have to look at how this artificial
1:30
intelligence lab originally started.
1:32
It was a nonprofit designed specifically to counter Google's
1:35
deep mind. Because Google was acquiring top
1:38
talent left and. Right, exactly.
1:40
There was a genuine fear that one massive corporation would
1:43
just control the future of artificial intelligence.
1:47
So the stated goal from the very beginning, the whole public
1:49
mission, was to raise a billion dollars through charitable
1:52
donations to build an open, safe alternative.
1:55
But the reality of that funding is entirely different from the
1:58
public narrative that was spent at the time, despite Musk's
2:02
public claims of contributing far more.
2:05
I mean, you sometimes cited numbers around 100 million.
2:07
The trial evidence shows his actual donations total between
2:10
38 million and $44 million. Right, which is a lot of money,
2:15
but. It's a lot, but the entire
2:17
organization only managed to raise around $50 million in
2:20
total from other charitable investors, so they were nowhere
2:23
near that billion dollar goal. Not even close.
2:25
And we have to look at the mechanical reasons why that
2:27
funding gap was so fatal. You hear people talk about the
2:30
cost of computing power required to train these artificial
2:33
intelligence models, but we really need to explain why that
2:36
is different from normal software.
2:38
Yeah, because writing the code is only a tiny fraction of the
2:40
battle. It's not like building a
2:42
standard and mobile application where a small team of engineers
2:45
can just write code on their laptops in a coffee shop.
2:48
Exactly. The process requires massive
2:51
data centers. We are talking about literal
2:54
warehouses filled with thousands of specialized processing units,
2:57
specifically GPU's running constantly for months at a time
3:01
just to train a single model. And you had to purchase the
3:04
chips, which cost 10s of thousands of dollars each.
3:07
Yeah. Buy the chips, build the
3:09
physical infrastructure, and then secure enormous amounts of
3:12
electricity just to run them and cool them.
3:15
The computing infrastructure alone burns through capital at a
3:18
rate that is practically unprecedented in the technology
3:22
sector. I mean, it requires negotiating
3:24
with utility companies just to ensure you do not brown out the
3:27
local power grid. Which is wild.
3:30
And during his testimony, Altman actually tried to explain this
3:32
by comparing their financial situation to a university.
3:36
He noted they could not possibly raise $200 billion for a non
3:39
profit. He pointed out that even a
3:41
massive, established institution like Stanford faces absolute
3:45
fundraising limits. Right.
3:46
Stanford has a massive endowment, incredibly wealthy
3:49
alumni, global reach, and they usually bring in around $3
3:53
billion in a highly successful fundraising cycle.
3:56
So Altman's point was that raising hundreds of billions
3:59
through philanthropy alone is a functional impossibility.
4:02
The math simply prohibits the original mission from
4:05
succeeding. Yes, the scale of the financial
4:07
requirement dictates the actual organizational structure.
4:11
It kind of feels like trying to fund a space program using local
4:14
school bake sales. I mean, the initial intentions
4:17
were incredibly noble, the volunteers were enthusiastic,
4:21
and the goal of reaching the stars was inspiring.
4:23
Right. And you can absolutely sell
4:25
enough cupcakes to buy some telescopes, you know?
4:27
Yeah. Or maybe find a small weather
4:29
balloon. But the financial mechanics of
4:31
selling cupcakes will never ever generate the capital needed to
4:35
build a Saturn V rocket. You need billions in heavy
4:39
industrial manufacturing. The math simply does not work,
4:43
no matter how pure the mission is or how hard those volunteers
4:46
work. That is a perfect way to look at
4:48
it, and this massive funding gap completely limits the
4:51
organization's ability to remain an open source, donation only
4:55
charity. When you realize that you need
4:57
billions of dollars just to keep the servers running and continue
5:00
the research, the leadership is forced to seek alternative,
5:04
aggressive corporate funding models simply to survive.
5:08
Because the noble idea of relying entirely on public
5:12
goodwill just collided directly with the harsh economic
5:16
realities of advanced computing. Exactly, if they stayed a pure
5:19
charity they would have simply run out of money and shut down
5:21
completely, ceding the entire field to Google.
5:24
So Musk proposed a very specific solution to this funding
5:27
problem. He saw the money running out and
5:30
he demanded majority equity and full control of the operation.
5:33
He. Proposed to either fund it
5:34
entirely himself or merge the AI lab directly with Tesla.
5:38
Basically using Tesla's resources to foot the massive
5:41
computing bill. And that demand initiated a
5:43
severe internal cultural clash, right?
5:46
Oh, absolutely. The researchers and founders
5:48
were not interested in becoming a subsidiary of an automotive
5:51
company. Furthermore, Musk utilized a
5:53
management style that involves stack ranking.
5:56
The researchers we. Should probably explain how
5:58
stack ranking actually works, because it is a very specific
6:01
and highly controversial human resources tactic.
6:04
Yeah. So stack ranking is a system
6:06
where management evaluates the staff and categorizes them by
6:10
performance on a rigid curve. It forces managers to label a
6:14
certain percentage of their team as top performers.
6:17
A large group is average and a specific bottom percentage as
6:21
underperformers, regardless of how well the team as a whole is
6:24
actually doing. And Musk would take what was
6:26
described in court as a chainsaw to those lower performing
6:30
individuals. Yeah, a chainsaw.
6:32
And Altman testified that this approach caused severe damage to
6:35
the organization's culture. He argued that it robbed
6:38
scientists that the psychological safety needed for
6:41
complex research. Which makes sense.
6:42
You have to consider the environment of a research
6:44
laboratory. Right when people are afraid for
6:47
their jobs on a constant basis, they do not take the necessary
6:51
creative risks required to solve deeply complex long term
6:54
engineering problems. Because in advanced scientific
6:57
research, failing is a core part of the process.
7:00
You try a new architecture for a neural network, it fails, you
7:04
learn from it, and you try again.
7:05
Exactly. And if a stack ranking system
7:08
punishes that failure with termination, the researchers
7:11
will only pursue safe, incremental improvements.
7:15
It completely destroys the capacity for breakthroughs.
7:17
So that tension eventually reached a boiling point.
7:20
During the trial, there was testimony recounting a highly
7:23
specific conversation where the Co founders asked Musk what
7:26
would happen to control of the technology if he passed away.
7:29
They were trying to establish a long term governance plan for an
7:32
entity they believed would eventually build highly capable
7:36
artificial general intelligence. Right.
7:38
And Musk's response to that question was that control could
7:40
simply pass to his children. Wait, back up.
7:43
He seriously suggested leaving control of artificial general
7:46
intelligence to his kids. Yes, that was the specific
7:50
testimony presented in court Underoath.
7:52
When pressed on the long term stewardship of the technology,
7:56
his default position was familial inheritance.
7:59
Wow, that uncompromising demand for a feudal style of ownership
8:03
completely changes the power dynamic among the founders.
8:07
It makes collaborative governance totally impossible
8:10
and guarantees A fractured relationship.
8:12
You cannot build a shared, open vision for the future of
8:15
humanity when one partner views the organization as a personal
8:19
family heirloom. No, you can't.
8:21
And the fallout from that fractured relationship resulted
8:24
in Musk walking away from the board entirely.
8:27
He wrote an internal e-mail stating the lab's probability of
8:29
competing with Google was absolute 0.
8:32
He essentially wrote off the entire venture.
8:34
Yeah, cutting off his financial support and leaving them to fend
8:37
for themselves. Altman actually testified that
8:39
following this departure, the organization was effectively
8:42
left for dead. They had a massive physical
8:44
infrastructure, highly paid researchers, and absolutely no
8:48
capital to keep the lights on. The environment following that
8:51
split became incredibly surreal. There was a crucial later
8:54
meeting regarding securing emergency funding for Microsoft.
8:58
This was a critical moment for survival.
9:00
Without this deal, the company would likely fold.
9:03
And Musk showed up to this meeting with good vibes, But he
9:07
spent the long conversation showing the team Internet memes
9:10
on his phone instead of addressing the existential
9:13
financial crisis they were facing.
9:15
The trial transcripts actually capture the absurdity of that
9:18
moment so perfectly. We have the specific detail from
9:22
the courtroom where the court stenographer literally struggled
9:25
to transcribe the word memes during the testimony.
9:28
We're really. Yeah, having to ask for
9:30
clarification on the spelling and meaning.
9:32
It highlights the bizarre nature of the interaction.
9:34
You have executives trying to save a multibillion dollar
9:38
research initiative and an early donor is treating it like a
9:41
casual hangout. And then you have a pretrial
9:43
text message from Musk to the company press threatening that
9:47
if they proceeded with their legal defense, they would become
9:49
the most hated men in America. It was incredibly personal.
9:53
Altman even sent a text to a former employee who happens to
9:57
be the mother of Musk's children with a crossed fingers emoji,
10:00
hoping Musk would take the news of the Microsoft investment
10:03
well. The personal dynamics are just
10:06
completely enmeshed with the corporate governance.
10:08
There is a severe irony in Musk's current lawsuit that we
10:11
have to point out here. His own rival artificial
10:14
intelligence company XAI, has been accused of using distilled
10:18
versions of the very models from the company he is actively
10:21
suing. Right.
10:23
And we should explain what distillation means in this
10:24
context. Yeah, go for it.
10:26
So distillation is a process where you take a massive, highly
10:30
capable AI model like the ones Open AI produces, and you use it
10:34
to train a smaller, more affiliate model.
10:37
You essentially have the large model generate millions of
10:40
examples and the smaller model learns to mimic the large models
10:43
behavior based on those examples.
10:45
So the accusation is that Musk is processing the
10:47
commercialization and closed nature of Open AI's technology
10:51
while simultaneously leveraging the outputs of that exact same
10:54
technology to build his own commercial enterprise.
10:57
Exactly, And Musk's formal exit opens up the legal and ethical
11:01
pathway for the remaining leaders to form a capped profit
11:04
subsidiary that restructuring results directly in the
11:08
multibillion dollar Microsoft partnership.
11:10
Because without his departure, that specific corporate
11:13
evolution likely never happens. He would have blocked any deal
11:18
that did not involve Tesla or his own total control.
11:21
Right so now, Musk's lawyers are claiming a breach of charitable
11:24
trust. They argue the original
11:26
donations were inextricably tied to a promise of eternal public
11:30
benefit. They are positioning the initial
11:32
founding vision as an unbreakable legal covenant,
11:36
basically asserting that when you take money under the guise
11:38
of philanthropy, you are bound to that mission forever,
11:41
regardless of how the technology or the economy shifts.
11:44
But there's a glaring hole in this legal theory.
11:46
There is no formal written agreement legally binding Musk's
11:50
donations to a perpetual nonprofit status.
11:52
None at all. None.
11:54
The court is looking for a contract, and the argument rests
11:56
heavily on emails, informal conversations and generalized
11:59
mission statements rather than ironclad legal documentation.
12:02
Which brings us to the legal concept of unscrambling the egg.
12:06
Yeah, it is the physical and financial impossibility of
12:09
reverting a tech giant with thousands of employees and
12:12
global investments back into a pure charity.
12:16
The company has evolved into a highly complex multinational
12:20
entity. They have established intricate
12:22
partnerships, distributed equity to early employees, and embedded
12:27
their software into the infrastructure of countless
12:29
other businesses. You cannot simply reverse a
12:31
decade of corporate evolution, nullify billions of dollars in
12:35
active contracts, and pretend the commercial entity never
12:38
existed. It is exactly like trying to
12:40
unbake a cake and expecting the original flour and sugar to
12:44
remain perfectly intact. You have already mixed the
12:47
ingredients applied the heat and fundamentally change the
12:50
chemical structure. That's right, you can complain
12:52
that you wanted a pie instead of a cake, but you cannot separate
12:55
the corporate reality back into its raw philanthropic
12:58
components. The raw materials do not exist
13:00
in their original form anymore. So let's look at how a capped
13:03
profit subsidiary actually functions mechanically, because
13:06
it is central to their defense. Right.
13:08
In a standard corporation, profits are limitless.
13:11
If the company makes a trillion dollars, the shareholders reap
13:14
the benefits indefinitely. But in a cap profit structure,
13:18
the investors and employees agree to a strict limit on their
13:22
financial returns. So, for example, they might
13:24
agree that they can only earn a return of 100 times their
13:28
initial investment. Exactly once that specific cap
13:31
is reached, any additional profit generated by the company
13:34
flows directly back to the original nonprofit organization.
13:38
Furthermore, the attorneys general of two separate states
13:40
already reviewed and approved the transition to a public
13:43
benefit corporation. State regulators examined the
13:46
shift from nonprofit to a capped profit model, evaluated the
13:49
corporate charter and permitted it to proceed.
13:52
And that absence of a written contract really severely limits
13:55
Musk's legal leverage. But simultaneously, this highly
13:59
publicized dispute changes how future tech startups will be
14:02
forced to document their public benefit promises and donor
14:05
agreements. The casual agreements of the
14:07
past, you know, those Silicon Valley handshake deals based on
14:10
shared idealism, will be replaced by intense legal
14:13
scrutiny from day one. Future founders will have to
14:16
explicitly define what happens to a donor's money if the
14:19
company needs to pivot to a commercial model 10 years down
14:22
the line. The trial also took a deeply
14:25
personal turn. During the cross examination,
14:27
Musk's lawyer, Steven Molo, looked at Altman directly and
14:31
asked a simple cutting question. He said.
14:33
Are you completely trustworthy? And the trial immediately became
14:36
a referendum on Altman's credibility as a leader.
14:40
Molo presented damning statements from former
14:42
executives. This included the former chief
14:44
scientist and the former chief technical officer, who both
14:47
accused Altman of a consistent pattern of lying.
14:50
They testified that he pitted executives against each other
14:53
and created a culture of chaos and manipulation behind the
14:56
scenes, which is. Huge.
14:57
We have to remember that the board previously fired Altman
15:00
temporarily. Right and Altman brushed this
15:02
off in court as a mere miscommunication and a breakdown
15:05
in trust. He framed a highly public,
15:08
chaotic board coup as a simple misunderstanding among
15:11
colleagues who were stressed and overworked.
15:13
But Microsoft CEO Satya Nadella testified about that specific
15:17
event, and he called the previous board's actions amateur
15:21
city and expressed intense panic that the workforce would leave
15:24
en masse. Microsoft had committed massive
15:27
resources essentially supplying the essential computing power
15:30
and infrastructure, and they were genuinely terrified that
15:34
the entire engineering team would walk out the door and
15:36
protest, leaving Microsoft with a multibillion dollar
15:39
partnership tied to an empty building.
15:42
The trial also brought heavy scrutiny over Altman's personal
15:45
investments in outside companies that do business with the AI lab
15:49
he holds stakes in a nuclear energy startup and Reddit.
15:52
Which raises serious questions about the intersection of his
15:54
personal financial interests and his role as a steward of the
15:57
supposedly Humanity Focus organization.
16:00
If the AI lab signs a contract with a nuclear company that
16:03
Altman personally invested in, the lines between philanthropic
16:06
stewardship and personal enrichment become incredibly
16:09
blurry. Yeah, and when your own chief
16:11
scientist and closest allies consistently accused you of
16:14
deception, it reveals a severe flaw in the organization.
16:18
Score integrity. If the internal leadership team
16:21
cannot trust the chief executive to be honest about strategic
16:24
decisions, the entire foundation of the company's ethical mandate
16:28
is compromised. A mission to build safe
16:30
technology for humanity requires absolute internal transparency.
16:35
See, I look at the reality of their situation and I actually
16:38
see it differently. I think Altman's aggressive
16:40
political maneuvering and his reality distortion were entirely
16:44
necessary to keep the underfunded company alive
16:47
against massive competitors. Really.
16:49
You think the deception was necessary?
16:50
Yeah, because when you are fighting a corporate giant like
16:53
Google with a fraction of the resources, you cannot survive
16:56
strictly on pure idealism and complete transparency.
16:59
If you act like a naive academic club, you get crushed.
17:02
You have to make ruthless, calculating decisions to secure
17:05
funding, appease powerful partners and retain top talent.
17:08
I understand the survival argument, but the public airing
17:12
of these internal grievances completely limits Altman's
17:15
carefully curated public image as a selfless steward of
17:18
humanity. It exposes the raw, ruthless
17:21
corporate tactics operating behind the scenes.
17:24
I mean sure it damages the image.
17:26
The facade of the benevolent philosopher king guiding
17:29
humanity toward a bright future is completely fractured by
17:32
testimonies of manipulation and strategic deception.
17:36
It shows that beneath the lofty mission statements, it operates
17:39
like any other cutthroat tech monopoly.
17:42
That's a fairpoint, but the original nonprofit still exists,
17:46
and it now acts as a foundation, holding an equity stake in the
17:49
for profit arm. Because of the incredible
17:52
success of the software, that stake is currently valued at
17:55
over $130 billion. And Altman's defense relies
17:58
heavily on this structure. He argues that this setup makes
18:01
someone of the wealthiest charities in the world.
18:04
He points out that this immense wealth allows them to fund
18:06
external initiatives like massive grants for Alzheimer's
18:10
Research and technology resilience programs.
18:12
Right. He positions the for profit
18:14
engine not as an abandonment of the mission, but as the
18:17
necessary mechanical engine to generate the capital required to
18:21
fulfill the charitable mission on a global scale.
18:23
Yet the organization has plummeted to the second to
18:25
lowest ranking on the Stanford Foundation Model Transparency
18:29
Index. Which is an index that measures
18:31
how open AI companies are about their training data, their code,
18:35
and their safety testing. They are becoming significantly
18:39
more secretive and less open with their research, locking it
18:42
behind commercial APIs. And European regulators and
18:46
state attorneys general are actively using the trial
18:49
testimonies to formulate strict new rules.
18:52
They are closely watching how the board operates and
18:54
considering concepts like mandatory internal safety boards
18:57
or the implementation of golden shares for public authorities to
19:00
ensure oversight. Yeah, the golden shares concept
19:03
is particularly interesting. What exactly is a golden share
19:05
and how would that work here? A golden share is a specialized
19:08
type of corporate stock that gives the holder veto power over
19:12
major company decisions regardless of how much actual
19:15
equity they own. Historically it was used when
19:18
European governments privatized state owned utilities.
19:21
They would sell the company but keep 1 golden share to ensure a
19:24
foreign entity could not buy the national power grid.
19:27
Oh I see. Applying this to artificial
19:29
intelligence means a government authority could hold a share
19:32
that allows them to legally veto the release of a new AI model if
19:36
they deem it unsafe, basically bypassing the corporate board
19:40
entirely. That would be a huge shift.
19:42
But you have to ask, is a charity truly serving its
19:45
purpose if its wealth is entirely dependent on the
19:48
aggressive commercialization of the exact technology you're
19:51
supposed to keep open and safe? That is the core conflict.
19:54
If your philanthropic power comes directly from locking down
19:57
the software you explicitly promised to share with the
20:00
world, the mission itself feels completely compromised.
20:04
It feels like a rationalization for abandoning the original
20:07
goal. This highly public trial opens
20:09
up the entire artificial intelligence sector to
20:12
unprecedented regulatory scrutiny.
20:15
It shifts the global focus away from just Product Safety, like
20:19
worrying about whether the software will generate bad
20:21
information, and directly on to the internal corporate
20:24
governance of these labs. Regulators are no longer just
20:27
looking at what the software does.
20:29
They are scrutinizing how the boards operate, how decisions
20:32
are made, and how profit motives influence long term research
20:35
directions. This whole trial just exposes
20:38
the harsh reality that developing advanced technology
20:42
requires astronomical capital, a pressure that forces even the
20:45
most idealistic charities to adopt aggressive corporate
20:49
tactics simply to survive. And it leaves you wondering, if
20:52
the brilliant minds building artificial intelligence cannot
20:55
manage their own organizations without descending into bitter
20:58
legal warfare and accusations of deception, how can they be
21:01
trusted to govern the technology itself?
21:03
If you're not subscribed yet, take a second and hit follow on
21:07
whatever a you're using. It hels us Kee making this.
21:09
We appreciate you being here. Also check out our YouTube
21:12
channel for more business and tech updates.
21:14
There's a link in the descripcion.