Elon Musk (SpaceXAI, Tesla): a primary-source profile, compared with Jensen Huang#

Prepared 26 September 2026. Covers statements and documents from 2023 to 25 September 2026, with a few earlier items marked as such. Written to be read on its own. It is one of a set of profiles comparing AI leaders with Jensen Huang, as he set out his views to Ezra Klein (New York Times, 23 September 2026), and reading them through the technology-neutral lens distilled from the European Environment Agency’s Late lessons from early warnings reports (2001, 2013).

Conventions. [P] marks Musk’s own words (X posts, recorded interviews, testimony) or his companies’ documents. [S] marks a secondary report of his words or actions. [I] marks my interpretation. Bracketed numbers are entries in the source list at the end. Huang quotes carry timestamps from the Klein interview [39]. Lens entries (W1, I5, M2 and so on) are defined in section 4.

Method and limits. Several outlets (CNBC, Forbes, The Hill) blocked automated reading, so some Musk quotes come through syndicated copies or other outlets’ quotations; these are marked [S]. I read the Economist interview (recorded around 20 July, published 23 July 2026) through a third-party transcript, and the All-In Summit remarks (14 September 2026) through press reports.

In brief#

• The highest stated tail risk of any frontier-lab owner, combined with full-speed building. Musk puts the chance of catastrophe at “10 to 20%” and says humans are “unlikely” to remain in control within ten years. He also says “I can’t see any way to really stop this incredible momentum” and “perhaps even if there was a stop button, we probably shouldn’t press it” (Economist, July 2026) [1].
• His theory of safety rests on values, and his institutional proposal is horizontal. The key safety property is that AI be “maximally truth-seeking and curious”. His mechanism is for rival labs, including “three or four of the leading Chinese companies”, to test each other’s models before release, with government as a last resort [1][3].
• In practice he is much closer to Huang. His companies are suing to block three state AI laws. He reportedly urged the President not to create an industry-funded frontier regulator. SpaceX signed the Nvidia-hosted letter defending open weights. And he calls oversight “a one-way ratchet” [3][11]–[15].
• His companies’ own documents are more measured than he is. xAI’s safety framework calls loss-of-control scenarios “speculative”. The risk factors in SpaceX’s S-1 include no catastrophic AI risk [9][10].
• The divergence from Huang is mostly one of substance on what AI is and how dangerous it is, and mostly one of commercial position on regulation. On pacing it is both.
• Through the lens: W4 (knowing is not acting) applies in modified form, and W8, I5, I9, L1 and M2 are all live. The reports partly support his “ratchet” claim (W8) and partly count against it (G9).

1. Position and interests#

SpaceX acquired xAI in an all-stock deal on 2 February 2026 that valued xAI at $250 billion. The AI business was renamed SpaceXAI in July [36]. SpaceX’s S-1 (20 May 2026) describes Grok as “a truth-seeking AI model, built on our founder Elon Musk’s mission to enable humanity to understand the universe”. It says Grok 5 “is currently being trained at COLOSSUS II”. It plans orbital AI compute satellites “as early as 2028” and Terafab, a chip-making initiative with Tesla and Intel. With Tesla, it is building Macrohard, an agentic platform “capable of fully emulating digital workflows”. In the first quarter of 2026 the AI segment had revenue of $818 million and an operating loss of $2.47 billion [9] [P].

Anthropic pays SpaceX $1.25 billion a month for Colossus capacity through May 2029, and either side can end the contract on 90 days’ notice [9]. Google pays $920 million a month [35] [S]. Musk is therefore a trailing model-maker. By his own account in July 2026, Anthropic is “obviously currently the leader” [4]. He is also a compute landlord to his rivals, one of Nvidia’s largest customers (Colossus 1 alone has more than 220,000 Nvidia processors, and Nvidia joined xAI’s funding rounds) [35], and a prospective chip-maker. Grok holds US defence contracts. CNBC reports that DOJ filings describe its use in military targeting [16] [S]. At Tesla he builds physical AI: robotaxis and Optimus. He sought a large voting stake in Tesla partly because “I don’t feel comfortable building that robot army unless I have a strong influence” (22 October 2025) [21] [P, via S]. His ties to government are close. He joined the President’s Beijing delegation in May 2026 [31], and on 23 September told Chinese state television that Xi “seems like a great leader” [30] [S].

2. Positions by dimension#

2.1 The nature of AI#

Musk treats AI as a new kind of mind on its way to exceeding humanity. “AI may exceed the sum of human intelligence in about 5 years”; “There really won’t be anything that AI can’t do better than humans, apart from being human, perhaps”; “If the difference in intelligence between AI and humans is vastly greater than the difference between AI and chimpanzees, it’s hard to imagine that the chimpanzees would be in charge” (Economist, July 2026) [1] [P]. At Davos he said AI might be “smarter than any human by end of this year, and no later than next year” (22 January 2026) [19] [P]. He also treats it as an industrial and physical build-out, spanning gigawatt clusters, orbital compute, Terafab and “more robots than people” [9][19]. The Klein interview recalls his 2014 description of humans as “the biological boot loader for digital superintelligence” [38] (Klein [56:51]).

2.2 The size and kind of risk#

His stated tail risk has been stable since 2023. In April 2023 he said AI has “the potential of civilization destruction” [27] [S]. In March 2024 he agreed with Hinton that the probability is “something like 10% or 20%”, while “the probable positive scenario outweighs the negative scenario” [24] (organiser’s account) [S]. In the July 2026 Economist interview he acknowledged the “10 to 20%” figure when it was put to him, and said “I still think there’s risk associated with AI and robots. It’s not zero” [1] [P]. In testimony at the OpenAI trial in April 2026 he said “The worst-case situation is where it is a terminator situation, where AI will kill us all”, and the judge cut him off [18] [S]. In September 2026 he reposted his view that “AGI is significantly higher risk than nuclear weapons” [7] [S].

His ranking of risks is his own. He names loss of control and misuse (“bioweapons or nuclear bombs or be deliberately deceptive”) [3]. He treats AI trained to be deceptive, which in his vocabulary includes “woke” training, as a path to “AI dystopia” [27]. He founded OpenAI out of concern about “the danger of Google being a monopoly in AI” [18] [S]. He does not frame jobs as a risk (2.9). He treats Grok’s non-consensual sexual imagery as a matter of individual illegality, and dismisses the outcry as “any excuse for censorship” (January 2026) [29] [S].

2.3 Safety: values, engineering or governance?#

Musk’s safety theory is about values and dispositions. “The most important thing for AI safety is to be maximally truth-seeking and curious” [1]. Grow the model, then “fine tune it to be really truthful. Don’t force it to lie even if the truth is unpleasant” (2024) [24]. An AI curious about the universe “is unlikely to annihilate humans because we are an interesting part of the universe” (2023) [27].

His mechanism is peer testing. “The most immediate thing that we could do is to have the leading AI companies at least just meet or have some sort of call once every few weeks”, with “a week or 2 weeks of early access by competitors” and an opportunity “to recommend pausing” [1] [P]. At the All-In Summit he said “instead of grading your own homework, you would at least have competitors grading your homework and raising the alarm if they see concerns”, because “the odds that you will find issues are dramatically greater” [3] [S].

Government comes last. “If a new model is very dangerous and the company producing that model is not taking steps to address the danger, then they should alert the government… I think they should have that power.” But “it’s quite difficult for someone in the government who doesn’t have a deep technical understanding… to know whether something should be released or not” [1] [P].

The company framework is more conventional. It covers chemical, biological, radiological and nuclear risks, offensive cyber, loss of control and manipulation. It says xAI “will only proceed with the development, the making available on the market, and/or the use of the model, if the systemic risks… are determined to be acceptable”, and that it “may temporarily fully shut down the relevant system”. It also claims “incidental alignment”: “our models naturally tend to refuse malicious requests even without any safety-specific training data” [10] [P]. SaferAI scored the December 2025 version at 23%, with 0% on containment [10] [S].

2.4 Warnings and “doomers”#

Musk was among the earliest prominent AI warners, and in September 2026 he posted “I’ve been sounding the alarm on AI for a long time” [7] [P]. He is nevertheless hostile to some other warners. He called Anthropic “misanthropic and evil” (February 2026) [5], and then “I was clearly wrong about Anthropic” (July) [4]. When the Anthropic researcher Jacob Coxon resigned with a viral warning, Musk wrote “Seems like a setup”, and that “the groundwork for this psy op (for lack of a better term) has been prepared for a long time” (around 10 September 2026) [6] [S]. Two days later he endorsed Amodei’s pacing essay [2]. [I] He contests the credibility and motives of particular messengers, and sometimes their ideology. He does not contest the message that AI could be catastrophic.

2.5 Regulation and government#

In principle he supports regulation. “It is important… to have a regulator, what you can think of as a referee” (Senate forum, 13 September 2023) [26]. “I agree with the vast majority of regulations… A referee is a good thing” (with Rishi Sunak, 2 November 2023) [25] [S]. On California’s SB 1047 he wrote: “I think California should probably pass the SB 1047 AI safety bill. For over 20 years, I have been an advocate for AI regulation” (26 August 2024) [23] [P].

In practice his companies oppose most of the specific rules that reach them. - xAI sued California over AB 2013, the training-data disclosure law (December 2025; preliminary relief denied) [13]. - It sued Colorado over its algorithmic-discrimination law, on First Amendment grounds among others, and the Justice Department intervened on its side (April 2026) [12]. - It sued Minnesota over its ban on nudification apps (July 2026; relief denied) [14].

I found no suit against California’s frontier-safety law (SB 53) or New York’s RAISE Act, though the S-1 lists both as risks. It warns that regulation could “delay or halt deployment of new systems… require us to modify, restrict, or discontinue certain features (including less constrained modes), and reduce the number of new entrants and customers” [9] [P]. [I] That wording almost exactly matches Nvidia’s own 10-Q risk factor, which suggests shared boilerplate rather than a personal view.

In September 2026 the Wall Street Journal reported that Musk, Huang and Zuckerberg separately urged the President not to create the industry-funded, FINRA-style testing body that Demis Hassabis had floated, arguing that it would entrench OpenAI, Anthropic and Google. The President did not proceed [15] [S; anonymous sourcing]. At All-In Musk said “You can always escalate the amount of regulatory oversight, but it is very difficult to reduce it… a one-way ratchet” [3][16] [S].

On liability he relies on it as enforcement: a lab that ignored rivals’ warnings would face “enormous” legal liability [3] [S]. Tesla, by contrast, told a court that the $243 million Autopilot verdict “will chill innovation, harm road safety” and that references to Musk’s own statements had misled the jury (August 2025). The verdict was upheld in February 2026 [34] [S].

2.6 Pacing, pausing and coordination#

• 2023. Signed the Future of Life Institute’s six-month pause letter. In a deposition he said he signed it “to urge caution” [18].
• July 2026. “I was just one of many signatories on that thing”; “6 months is a long time”; and the stop-button remark [1]. His response to the OpenAI–Hugging Face incident was one word: “Yikes” (25 July) [8]. I could not confirm whether any xAI or SpaceX staff signed the “Pacing the Frontier” statement (28 July). None appear among its prominently listed signatories.
• 12 September. Quote-posting Amodei’s “We Must Pace the Frontier”, he wrote “Dario is right” [2] [P]. The essay proposed embedded third-party evaluators, coordination among democracies with a “narrow waiver” of antitrust law, and no powerful chips for China.
• 14 September. His own mechanism differs from Amodei’s. It uses peer testing, not rate limits or an antitrust waiver, and it includes Chinese labs: “Peer review of AI by competitors is the right way to start this off”. Any proposal “has to be something that China is willing to accept, otherwise we’re just handicapping ourselves and China will essentially win” [3] [S]. He added: “On balance, I think Anthropic puts more care into their safety than OpenAI” [3] [S].
• 18 September. A consumer antitrust class action named SpaceXAI and cited his “Dario is right” post as evidence of an agreement to slow development [17] [S].

2.7 Open versus closed#

He sued OpenAI for becoming “an $800 billion for-profit company with closed source… the exact opposite of what I donated the money for” [1]. xAI released Grok-1 under Apache 2.0 (March 2024) and Grok 2.5 under a restrictive licence (August 2025). His promise that “Grok 3 will be made open source in about 6 months” [22] [P] was unfulfilled as of May 2026 [22] [S]. SpaceX signed the Nvidia-hosted “Open Weights and American AI Leadership” letter (24 July 2026), which says “openness may be one of the most important paths to AI safety and security” and warns against “premature restrictions on open models” [11] [P]. xAI’s framework, meanwhile, guards against “large-scale extraction and distillation of reasoning traces” [10]. At the OpenAI trial Musk said xAI had “partly” distilled OpenAI’s models, adding “Generally AI companies distill other AI companies” [36] [S].

2.8 China, export controls and the race#

Musk treats China as a likely leader and as a party to safety arrangements. In 2023, after visiting Beijing, he said “China is definitely interested in working in a cooperative international framework for AI regulation” [28] [S]. In 2026 he said: “Given that the Chinese AI companies are doing as well as they are with a relatively small amount of compute… if they had a lot of compute, there’s a good chance that they would be the leaders”; “In China, the constraint is chips”; and he expects China to reach “4 times the electricity production of the US” [1] [P]. He would bring Chinese labs into peer testing (“probably something that China would agree to”) [3]. I found no public position by Musk on chip export controls. [I] His premise that chips are China’s binding constraint implies that marginal compute matters, which runs against Huang’s argument that China already has enough.

2.9 Jobs and distribution#

Musk predicts substitution, not complementarity. In 2023 he spoke of a future where “no job is needed” and there will be “universal high income” [25]. By 2025–26 he was saying: “My prediction is that work will be optional” (with Huang, 19 November 2025) [20]; “Work is going to be optional… like gardening”; and “I think that the Treasury should just simply issue people checks” [1] [P]. He offers no worked account of the transition. Macrohard and Optimus are designed to take over whole workflows [9].

2.10 Energy#

In his view electricity is the binding constraint: “the limiting factor for AI deployment is fundamentally electrical power” (Davos) [19]. The solution is solar and space: “The lowest cost way to do AI compute will be with solar powered AI satellites” within “not more than five years” [20] [P]. He called the 2025 Senate bill cutting clean-energy credits “utterly insane and destructive”, saying it “gives handouts to industries of the past” [32] [S]. In practice, Colossus has run on gas turbines. The NAACP sued in April 2026 over turbines operating without a permit near Southaven, Mississippi [33], and xAI signed the White House Ratepayer Protection Pledge (4 March 2026) [40].

2.11 Epistemic style and formation#

Musk studied physics and economics, and his career runs from software start-ups and PayPal (as financier) to SpaceX and Tesla (as chief engineer and product builder). The S-1 turns his style into company doctrine: truth-seeking “grounded in evidence, logic, empirical data, and first principles thinking” [9]. [I] Four features matter here: - Stated probabilities. He attaches numbers to catastrophe, where Huang refuses to. - Iterate and fail. SpaceX tests hardware to destruction on its own ranges. Tesla has iterated driver-assistance software on public roads, where the liability record (2.5) shows who bears the failures. - Rapid, public reversal. On Anthropic he moved from “evil” to “clearly wrong” in five months. - Extreme forecasts. He treats extreme forecasts as ordinary talk, for example saying British “civil war” is “inevitable” [1].

2.12 Commercial position and how it bears on his views#

[I] Several interests pull in different directions. - Pacing. Pacing the frontier would help a trailing lab relative to leaders, the point the antitrust suit presses. But pacing that is heavy on evaluation need not cut demand for his rented compute. - Regulation. Opposing an incumbent-funded regulator protects a challenger. - Product features. Litigation against content and disclosure laws protects product features the S-1 names (“less constrained modes”). - China. Tesla’s reliance on Shanghai sits beside his conciliatory tone.

His tail-risk view predates xAI by nearly a decade (2014; OpenAI in 2015), so it is not easily explained by current stakes. His positions on regulation-in-practice track those stakes more closely.

2.13 Shifts, 2023 to September 2026#

The stable core is a high tail-risk estimate, the truth-seeking theory of safety, and the expectation of abundance. What has changed is whether pausing is possible or desirable, the role of government (from referee to last resort), and his attitude to particular rivals.

3. Compared with Huang#

[I] Where Musk stands relative to the other leaders. - On tail risk he stands with Hinton, Amodei, Altman and Hassabis. - On regulation in practice he stands with Zuckerberg and Huang. - Peer testing among rivals, with China included, is a lane of his own. - He is unusual in two respects. He frames the danger ideologically (deceptive or “woke” training), and he ships “less constrained modes” as a feature.

In 2024 Huang reportedly distanced himself from Musk on AI risk: “I feel like you’re interviewing Elon right now, and I’m just not that guy” (as reported in April 2025 reviews of Stephen Witt’s biography of Huang) [S]. Two years on, their risk views remain far apart, while their regulatory practice has converged.

4. How the Late Lessons lens reads Musk’s stance#

The lens is a set of diagnostic questions distilled from the EEA’s case histories. Each entry has a Mirror question that turns it on critics and warners. Its rules apply here as they did to Huang: - apply every question symmetrically; - weight each entry by case type, since patterns drawn mainly from known-harm failures ([K]) transfer less well to genuinely uncertain technologies than those drawn from uncertain cases ([U]) or forward warnings ([F]); - weigh direction above magnitude; - judge ex ante; - record findings without adding them up.

The reports are also an imperfect, partly advocacy source with a mixed forward record. And AI differs from their cases: harm can arrive fast, software is patched, the systems act and adapt, and the “regulators” Musk proposes are rival firms, for which the reports offer almost no precedent.

Where the lens supports him. - Peer detection by informed insiders is a mechanism the reports value (W1). - His warning that oversight ratchets up has support in documented cases (W8), though G9 documents the opposite pattern for protective reforms. - His objection to incumbent-funded oversight is an I9 question the reports under-used. - His concern about displacement to China is one the reports document in other settings (I8, [K]).

Where it challenges him. - Voluntary arrangements are fragile. In the invasive-species case, industry codes “have so far had limited effectiveness and buy-in”, and some firms thought “legislation would be fairer” than codes that others could ignore (LL2-20, p. 499; the same chapter says both approaches are needed). No rival has publicly accepted his proposal [3]. - Independence of oversight matters (I5). - A property prized in the market can be the hazard (L1). - His confidence rests on a model of harm he has not exposed to test (M2).

Direction and magnitude. His 10–20% is a claim about magnitude. In the hindsight record, warnings were more reliable about direction than about magnitude (W7). The lens therefore neither validates his number nor Huang’s dismissal of it.

5. Gaps#

• No official transcripts were read for the Economist interview or the All-In session.
• The Wall Street Journal report on lobbying against a regulator rests on anonymous sources.
• No position by Musk on chip export controls was found.
• No response by Musk to Huang’s Klein interview was found.
• The status of the Grok 3 open-weights promise after May 2026 is unverified.
• The claims about military targeting come from secondary reporting of DOJ filings.

Sources#

1. The Economist, Zanny Minton Beddoes interview, recorded ~20 July, published 23 July 2026. Video: https://www.realclearpolitics.com/video/2026/07/24/full_video_elon_musks_interview_with_economist_editor_zanny_minton_beddoes.html ; transcript: https://singjupost.com/full-transcript-zanny-minton-beddoes-interviews-elon-musk/ ; Fortune: https://fortune.com/article/elon-musk-says-rival-ai-labs-should-peer-review-frontier-models-before-release-with-government-as-last-resort/
2. Musk on X, “Dario is right”, 12 Sept 2026, via https://siliconangle.com/2026/09/13/sam-altman-and-elon-musk-back-dario-amodeis-call-to-slow-down-the-frontier-of-ai-development/ ; Amodei: https://www.darioamodei.com/post/we-must-pace-the-frontier
3. All-In Summit, 14 Sept 2026: https://www.cnbc.com/2026/09/15/elon-musk-ai-safety-testing.html (blocked); https://techstartups.com/2026/09/15/elon-musk-calls-on-rival-ai-labs-and-chinese-companies-to-test-each-others-models-amid-calls-for-ai-slowdown/ ; https://teslanorth.com/2026/09/15/musk-ai-peer-review-all-in-summit-shotwell/ ; https://aiweekly.co/alerts/musk-urges-us-chinese-labs-to-cross-test-ai-before-release ; https://247wallst.com/investing/2026/09/15/musk-and-cramer-agree-on-one-thing-about-the-ai-panic/ ; https://www.fool.com/investing/2026/09/23/elon-musk-just-said-dario-is-right-about-slowing-a/
4. Musk on X, 9 July 2026: https://x.com/elonmusk/status/2075278580955685036
5. Musk on X, 12 Feb 2026: https://x.com/elonmusk/status/2022036387885892022
6. Coxon posts, ~10 Sept 2026: https://www.forbes.com/sites/siladityaray/2026/09/10/musk-touts-psy-op-and-mocks-ex-anthropic-staffer-who-warned-ai-could-kill-us-all/ ; https://deltiasgaming.com/seems-like-a-setup-elon-musk-on-anthropic-researcher-who-recently-shared-startling-messages-about-ai-gave-up-his-equity-to-leave-the-company/
7. Musk on X, 13 Sept 2026 (UTC): https://x.com/elonmusk/status/2098935235446551022 ; https://theprint.in/feature/elon-musk-warned-ai-could-kill-humanity-3-years-later-hes-still-sounding-the-alarm/3042670/
8. Musk on X, 25 July 2026: https://x.com/elonmusk/status/2080832017164370164
9. SpaceX Form S-1, 20 May 2026: https://www.sec.gov/Archives/edgar/data/1181412/000162828026036936/spaceexplorationtechnologi.htm
10. xAI Frontier AI Framework, effective 30 June 2026: https://media.x.ai/v1/website/xai-frontier-artificial-intelligence-framework-30-june-2026-99c40684.pdf ; SaferAI: https://tracker.safer-ai.org/company/xai/
11. “Open Weights and American AI Leadership”, 24 July 2026: https://images.nvidia.com/pdf/Open-Weights-and-American-AI-Leadership.pdf
12. xAI v. Weiser: https://www.justice.gov/opa/pr/justice-department-intervenes-xai-lawsuit-challenging-colorados-algorithmic-discrimination
13. xAI v. Bonta: https://law-ai.org/xais-challenge-to-californias-ai-training-data-transparency-law-ab2013/
14. Minnesota: https://www.ag.state.mn.us/Office/Communications/2026/07/31_xAI.asp
15. WSJ report, 17 Sept 2026, via https://techstartups.com/2026/09/17/zuckerberg-musk-and-jensen-huang-reportedly-lobbied-trump-to-halt-industry-funded-ai-regulator-plan/
16. CNBC, 18 Sept 2026: https://www.cnbc.com/2026/09/18/after-decade-of-clashes-in-ai-elon-musk-forging-strange-alliances.html (read via https://www.newsbeep.com/us/852565/)
17. Antitrust suit: https://www.pbs.org/newshour/nation/lawsuit-says-anthropic-openai-spacexai-and-google-made-illegal-agreement-on-ai-slowdown
18. Musk v. Altman: testimony https://usaherald.com/judge-halts-musks-ai-doomsday-warnings-as-testimony-wraps-in-high-stakes-openai-trial/ ; deposition https://techcrunch.com/2026/02/27/musk-bashes-openai-in-deposition-saying-nobody-committed-suicide-because-of-grok/ ; verdict https://www.npr.org/2026/05/18/nx-s1-5822366/musk-altman-openai-jury-verdict-claims-dismissed
19. Davos, 22 Jan 2026: https://whatsuptesla.com/2026/01/27/transcript-elon-musk-at-davos-world-economic-forum-jan-2026/
20. US–Saudi Investment Forum with Huang, 19 Nov 2025: https://singjupost.com/transcript-elon-musk-and-jensen-huang-talk-ai-at-u-s-saudi-investment-forum-11-19-25/
21. Tesla Q3 2025 call: https://electrek.co/2025/10/22/elon-says-quiet-part-out-loud-he-needs-1-trillion-so-he-can-control-a-robot-army/
22. Musk on X, 23 Aug 2025: https://x.com/elonmusk/status/1959379349322313920 ; status: https://www.change.org/p/elon-xai-keep-your-word-open-source-grok-3-now
23. SB 1047 post, 26 Aug 2024: https://techcrunch.com/2024/08/26/elon-musk-unexpectedly-offers-support-for-californias-ai-bill/
24. Abundance Summit, 19 Mar 2024: https://www.diamandis.com/blog/elon-abundance-ai-human-survival
25. With Sunak, 2 Nov 2023: https://www.cnn.com/2023/11/02/tech/elon-musk-conversation-british-prime-minister-rishi-sunak-artificial-intelligence/index.html
26. Senate AI Insight Forum, 13 Sept 2023: https://www.searchenginejournal.com/highlights-from-the-ai-insight-forum-exploring-ai-regulation/496403/
27. Fox News interview, 17 Apr 2023: https://www.cnn.com/2023/04/17/tech/elon-musk-ai-warning-tucker-carlson/index.html
28. China, 2023: https://www.voanews.com/a/musk-says-china-detailed-plans-to-regulate-ai-/7124584.html
29. Grok deepfakes, Jan 2026: https://www.itv.com/news/2026-01-09/elon-musks-grok-limits-ai-image-generation-to-paid-subscribers-after-outcry
30. CCTV interview via SCMP, 23 Sept 2026: https://www.scmp.com/news/china/politics/article/3368524/elon-musk-hails-xi-jinpings-leadership-and-chinese-ai-capabilities-state-media
31. Beijing delegation: https://www.cnbc.com/2026/05/14/xi-china-open-us-business-ai-chips.html
32. Senate bill, 28 June 2025: https://www.washingtonpost.com/climate-environment/2025/06/28/solar-green-energy-tax-bill-trump/
33. NAACP suit: https://www.selc.org/press-release/civil-rights-group-sues-xai-for-illegal-pollution-from-data-center-power-plant/
34. Benavides v. Tesla: https://www.cnbc.com/2025/08/29/tesla-appeal-benavides-verdict-autopilot-crash.html ; https://electrek.co/2026/02/20/tesla-has-to-pay-historical-243-million-judgement-over-autopilot-crash-judge-says/
35. Compute deals: https://techcrunch.com/2026/05/20/anthropic-will-pay-xai-1-25-billion-per-month-for-compute/ ; https://techcrunch.com/2026/06/05/google-will-pay-spacex-920m-per-month-for-compute/ ; Nvidia’s stake: https://www.cnbc.com/2026/05/09/nvidia-embraces-ai-investor-topping-40-billion-in-equity-bets-2026.html
36. Merger and distillation testimony: https://x.ai/news/xai-joins-spacex ; https://en.wikipedia.org/wiki/SpaceXAI
37. xAI v. Apple and OpenAI, 25 Aug 2025: https://www.cnbc.com/2025/08/25/musk-lawsuit-apple-openai-monopoly.html
38. Musk on X, 3 Aug 2014: https://x.com/elonmusk/status/496012177103663104
39. Klein–Huang interview, 23 Sept 2026: https://www.nytimes.com/2026/09/23/opinion/ezra-klein-podcast-jensen-huang.html
40. Ratepayer Protection Pledge, 4 Mar 2026: https://www.whitehouse.gov/fact-sheets/2026/03/fact-sheet-president-donald-j-trump-advances-energy-affordability-with-the-ratepayer-protection-pledge/