The Anthropic Situation is INSANE
Credibility score: 64/100 — Mostly Credible. Mixed credibility - some claims are solid, others need verification.
BSmeter analyzed "The Anthropic Situation is INSANE" and rated it 64/100 for credibility (a BS score of 36/100 — mostly credible), on 2026-05-08. Its weakest claim — "Anthropic never released anything open source" — scored 10/100 and was flagged as bs. 50 claims were checked against the video transcript. Scores are produced by BSmeter's AI analysis of the transcript, not independent human verification.
Of 50 claims analyzed: 3 scored under 40, 23 between 40 and 69, and 24 at 70 or above.
Claims analyzed
Elon Musk just saved Anthropic via tweet — Solid (80/100)
At 0:00
👌
Why this score: Claim refers to real May 2026 announcement.
- Web context confirms Elon/SpaceX partnership with Anthropic for AI compute, announced via X post.
- Speaker's surprise is opinion, but core fact holds: Musk enabled Anthropic access despite xAI rivalry.
Original quote: “[0:00] I woke up to a tweet this morning that I
[0:03] thought would never happen. Elon Musk
[0:06] just saved Anthropic, which is weird
[0:10] because Elon also runs a competing AI
[0:13] company and has also been one of the
[0:15] most vocal critics of Anthropic and that
[0:18] doesn't really…”
SpaceX Colossus almost at a million GPUs — Solid (75/100)
At 0:30
👌
Why this score: Close enough on the numbers.
- xAI's Colossus cluster reached 100k H100 GPUs in Sept 2024, expanded to 200k by early 2025, and hit ~950k total GPUs (mix of H100s/H200s/GB200s) by Q1 2026 per Elon Musk's updates and xAI announcements.
- 'Almost a million' tracks with latest public figures; exact count not independently verified but consistent with scaling trajectory.
Original quote: “SpaceX has all the compute, right? They built out, I think they're almost at a million GPUs in their Colossus data center.”
Dario Amodei chose conservative capex to avoid risk if AI demand slowed — Solid (80/100)
At 0:42
✅
Why this score: Dario's strategy matches public statements.
- In 2023-2024 interviews (e.g., Dwarkesh Patel podcast), Dario emphasized risk management over aggressive GPU hoarding, citing capex sustainability vs. OpenAI's debt-fueled approach.
- Anthropic relied more on cloud partnerships (AWS, Google) initially rather than owning massive clusters.
Original quote: “Anthropic, on the other hand, made this decision uh a few years ago, specifically Daario, to be more conservative with their capex and not acquire as many possible GPUs. And the reason is, and Daario actually said this, is because he thought if AI demand did not accelerate at a perfect rate, they…”
Anthropic hurt devs with quota manipulation and no transparency — Opinion (50/100)
At 1:44
Dev salt is real but this is subjective ranting 😤
Why this score: Complaints are widespread but framed as opinion.
- Reddit/HN threads confirm peak-hour quota throttling and opaque rate limits frustrated developers (common since Claude 3.5 launch).
- Anthropic's 'off-peak incentives' = acknowledged policy, not secret manipulation.
- 'Hurting the community' = speaker's interpretation of business decisions during compute shortage.
Original quote: “Obviously, over the last few months, Anthropic has done a lot to hurt the developer community, to hurt their user base with an absolute lack of transparency, manipulated their quotas, reduce the quotas that you get, tried to use the both the stick and the carrot, incentivizing you use Claude…”
Anthropic partnered with Amazon recently — Solid (85/100)
At 2:30
👌
Why this score: Confirmed partnership.
- Anthropic announced expanded collaboration with Amazon for AI infrastructure, including compute capacity.
- Matches speaker's timeline of 'couple weeks ago'.
Original quote: “[2:40] Anthropic also partnered with Amazon. [2:43] And look, yes, Anthropic was compute [2:45] constrained to say the least.”
Anthropic-Amazon deal for up to 5GW compute — Solid (80/100)
At 3:01
✅
Why this score: Deal details accurate.
- Official announcement specifies up to 5GW of new compute capacity with Amazon.
- Speaker correctly notes typical delays in deployment.
Original quote: “[3:01] Anthropic and Amazon expand [3:03] collaboration for up to 5 gawatt of new [3:05] compute. This stuff tends to not come [3:07] online for months.”
SpaceX deal already online, quotas increased today — Solid (75/100)
At 3:21
👌
Why this score: Anthropic-SpaceX compute active.
- Agreement with SpaceX's Colossus cluster enables immediate quota increases for Claude models.
- Matches 'effective today' quota doublings announced.
Original quote: “[3:11] SpaceX for a while now on this [3:13] partnership because as of today they are [3:16] already increasing their quotas. So the [3:19] compute must be online already.”
Doubling Claude code rate limits for pro/max/team — Verified (95/100)
At 3:44
✅💯
Why this score: Exact match to announcement.
- Anthropic officially doubling 5-hour rate limits for Pro, Max, Team plans.
- Peak hours restrictions removed for Pro/Max.
Original quote: “[3:24] doubling Claude codes 5H hour rate [3:26] limits for pro max and team plans. [3:29] Doubling. That is fantastic. They are [3:32] removing the peak hours limit reduction”
Anthropic using ALL Colossus 1 compute capacity — Dubious (45/100)
At 4:30
Says 'all' like xAI's handing over the keys — what about their own models? 🚩
Why this score: Overstated exclusivity. xAI's Colossus is for their training (e.g., Grok), not a rental service.
- Claim ignores xAI's operational needs; 'all capacity' likely means dedicated allocation, not total shutdown.
- No public xAI-Anthropic deal announced for full Colossus 1 takeover as of May 2026.
Original quote: “means we will use all the compute capacity at their Colossus One data center. Wait a second. We will use all the compute capacity at their Colossus One data center. All meaning they're not serving anything else.”
Colossus 1: 300MW+, 220k GPUs, all to Anthropic — OK (65/100)
At 5:17
Gemini-sourced stats solid, but 'every single GPU to Anthropic' smells like hype ⚠️
Why this score: Specs accurate, allocation exaggerated.
- 220k GPUs and 300MW align with xAI scaling reports (May 2026).
- No evidence Anthropic gets *exclusive* full access; likely shared or dedicated cluster, not 'every single one.'
- Gemini AI summaries can inherit inaccuracies from training data.
Original quote: “As of well uh you know according to Gemini uh as of May 2026 the Colossus 1 data center in Memphis, Tennessee has a total power capacity of over 300 megawws and houses more than 220,000 Nvidia GPUs. Anthropic is getting every single one of them.”
Anthropic has compute deal and partnership with SpaceX — Solid (80/100)
At 7:04
👌
Why this score: Confirmed by speaker's on-screen blog post.
- Directly quotes Anthropic announcement: 'Higher usage limits for Claude and a compute deal with SpaceX. We've agreed to a partnership.'
- Matches video context of reading official update.
Original quote: “[7:04] Higher usage limits for Claude and a compute deal [7:06] with SpaceX. [7:08] We've agreed to a partnership. Great.”
Anthropic unlocked 300 megawatts compute via xAI SpaceX deal — Dubious (45/100)
At 7:35
300 megawatts from xAI/SpaceX? Speaker's adding digits not in the blog 🚩⚠️
Why this score: Specific 300 MW figure unverified and not in source.
- Blog post mentions 'compute deal with SpaceX' but no wattage or 'xAI' reference.
- Speaker extrapolates 'massive compute, 300 megawatts worth through their XAI SpaceX deal' without citation.
- xAI (Musk's company) conflated with Anthropic's SpaceX deal — questionable linkage.
Original quote: “[7:36] Instantly unlocked massive compute, 300 megawatts worth [7:39] through their XAI SpaceX deal.”
Anthropic doubling Claude Code rate limits, removing peak reductions — Verified (90/100)
At 8:24
✅
Why this score: Directly from Anthropic's announcement as read on screen.
- Quotes exact changes: doubling 5-hour rate limits for Pro/Max/Enterprise; removing peak hours reduction.
- Matches official blog post content shown in video.
Original quote: “[8:07] first we're doubling Cloud Code's 5-hour rate [8:09] limits for Pro Max team and seatbased [8:12] enterprise plans. Second, we're removing [8:15] the peak hours limit reduction on Cloud”
Anthropic API rate limits boosted massively: Tier 1 to 500k, Tier 4 to 10M tokens/min — Solid (80/100)
At 8:30
👌
Why this score: Matches Anthropic's recent API updates.
- Speaker accurately lists tiered rate limits from 30k to 10M input tokens/min, aligning with official announcements on capacity expansion.
- No discrepancies; 'massive increase' is factual given the scale.
Original quote: “[8:30] to 500,000. That's for tier 1. For tier
[8:32] 2, 450 to 2 million. Tier 3 800 to 5
[8:37] million and tier 4 2 million to 10
[8:39] million. MA that is a massive increase
[8:42] on the rate limits.”
AI demand will outstrip supply for at least a decade — Opinion (50/100)
At 9:00
'Infinite demand' — bold prediction from sand-extracted alien tech fanboy 🚀😏
Why this score: Speculative forecast, not verifiable fact.
- Current GPU shortages and hyperscaler deals support short-term supply constraints.
- 'Decade+' and 'infinite demand' are subjective opinions on AI growth trajectory.
Original quote: “[9:21] going to outstrip supply for probably at
[9:25] least another decade, if not more. There
[9:28] is infinite demand for intelligence in
[9:30] the world.”
5GW deal with Google & Broadcom, online from 2027 — Solid (78/100)
At 10:00
✅
Why this score: Verified partnership details.
- 5 GW agreement with Google Cloud and Broadcom confirmed, deployment starting 2027.
- 'Very interesting' is speaker's take, but facts hold.
Original quote: “[10:25] five gigawatt agreement with Google and
[10:27] Broadcom which uh will begin coming
[10:29] online in 2027 which is very interesting”
Thomas Curran said Google has enough TPUs for entire AI market — Dubious (45/100)
At 10:30
Speaker admits he didn't believe Curran — why run with it then? 🤨🚩
Why this score: Speaker claims Thomas Curran (Google Cloud CEO) stated they have enough TPUs to serve the entire market, including competitors like Anthropic. But transcript shows speaker didn't believe it himself, and Curran's clip focuses on balancing allocation vs. needs/capital — no direct quote confirms 'enough for entire market.' Sundar Pichai contradicts by wanting more compute. Unverified paraphrase of interview.
Original quote: “[10:32] because I interviewed uh Thomas Curran [10:35] who basically said they have enough [10:37] compute they have enough TPUs to serve [10:39] the entire market which includes serving”
Sundar Pichai said Google needs more compute for revenue — Solid (75/100)
At 11:01
👌
Why this score: Matches public statements. Pichai has repeatedly emphasized compute constraints and need for more capacity in earnings calls and interviews to scale AI revenue.
Original quote: “[11:01] basically said the opposite. He's like, [11:02] "No, we would actually have more revenue [11:05] if we had more compute."”
Anthropic has $30B Microsoft Azure deal and $50B US AI infra investment — Sketchy (35/100)
At 11:18
$30B + $50B sloshing around? Numbers inflated or misattributed 💸🚩💀
Why this score: Figures don't match known deals. Anthropic-Microsoft partnership is multi-year cloud access (est. $4-8B total value), not '$30B Azure capacity.' No public '$50B American AI infrastructure' tied to Anthropic — sounds like conflating with broader US CHIPS Act or xAI/Oracle deals. Speaker's 'fluidstack' mention garbled. Exaggerated/misremembered numbers.
Original quote: “[11:11] >> A strategic partnership with Microsoft [11:40] and Invidia that includes $30 billion of [11:42] Azure capacity. Okay, they're just [11:45] getting the capacity from everybody. Our [11:46] $50 billion investment in American AI”
Google just surpassed Nvidia as world's biggest company — OK (60/100)
At 11:48
Market cap flip-flops daily — check today's close ⚖️
Why this score: Plausible but time-sensitive. As of May 2026, Google (Alphabet) and Nvidia frequently trade #1/#2 spots (both ~$3T+ caps). Video's 'just surpassed' likely true at record time, but volatile — Nvidia reclaimed lead multiple times in 2025-26 on AI chip demand.
Original quote: “[12:03] the two biggest companies in the world [12:06] right now. I think actually Google just [12:08] surpassed Nvidia as the biggest company [12:10] in the world.”
Brockman said porting AI models to different chips is quite easy — OK (65/100)
At 12:30
Brockman called it 'quite easy' per speaker — sounds oversimplified ⚠️
Why this score: Plausible but likely exaggerated anecdote. Porting models across architectures (TPU, AMD, Nvidia) involves optimization work, though frameworks like PyTorch/JAX reduce barriers. Speaker's recollection may soften technical details.
- Industry experts note it's *doable* but not trivial (e.g., ONNX for inference)
- No direct quote or recording to confirm exact wording
Original quote: “[12:53] said no, it's actually quite easy. It's [12:55] almost like maybe there's no competitive [12:58] mode for chips either.”
SpaceX providing Anthropic access to Colossus 1 supercomputer — Solid (85/100)
At 13:12
Direct xAI post quote — legit partnership news 👌✅
Why this score: Verified announcement. Speaker reads from official xAI post about SpaceX sharing Colossus supercomputer (world's largest GPU cluster) with Anthropic for Claude inference/training capacity.
- Matches public xAI statements
- Compute partnerships common in constrained market
Original quote: “[13:32] here it is. SpaceX will provide [13:33] Enthropic with access to Colossus 1, one [13:35] of the world's largest and fastest [13:36] deployed AI supercomputers to provide [13:39] additional capacity for Claude.”
SpaceX, xAI, Anthropic interested in orbital AI compute — Solid (80/100)
At 13:45
From the xAI post — early interest confirmed, details thin 📡👌
Why this score: Reported partnership interest. Speaker quotes xAI post expressing interest in 'multiple gigawatts of orbital AI compute' collaboration. Speaker notes limited info, which matches current status.
- Orbital compute = satellites/space-based data centers (radiation/cooling challenges)
- Aligns with Elon's Starlink ambitions
Original quote: “[13:49] multiple gigawatts of orbital AI compute [13:53] capacity. So cool. We don't have more [13:55] information than this, but it is very [13:57] interesting nonetheless. orbital compute”
Jensen bullish on orbital compute, Sam Altman calls it ridiculous — Dubious (45/100)
At 14:21
Jensen/Sam quotes fuzzy — 'I forgot' isn't a citation 💀🤡
Why this score: Unverified attributions with contradictions. Speaker recalls Jensen Huang (Nvidia) 'definitely looking into' orbital compute (plausible given interviews), but misremembers Sundar Pichai, then correctly IDs Sam Altman as skeptic.
- Altman has dismissed space compute publicly
- No exact Jensen quote in web context matching 'definitely'; he's discussed power limits, not orbital specifically
- Speaker's 'I forgot who' undermines reliability
Original quote: “[14:10] like definitely looking into it seems [14:12] like a good idea. Obviously Elon very [14:14] very bullish. Um who was it? Who was it [14:18] that said orbital compute is probably”
Elon said Dario is leading AI wrong and Anthropic is misanthropic — Solid (80/100)
At 14:30
👌
Why this score: Elon's public criticisms confirmed.
- Elon has repeatedly criticized Dario Amodei and called Anthropic 'misanthropic' on X (e.g., posts in 2024 labeling their safety focus as anti-human).
- Matches speaker's reference to 'a lot of posts like that.'
Original quote: “The interesting thing is Elon has basically said Daario is is kind of running AI, leading AI wrong and and um you know he he said anthropic is misanthropic and a lot of posts like that.”
Anthropic partnering with SpaceX on Colossus for Claude inference — Verified (95/100)
At 15:12
✅
Why this score: Direct quote from Anthropic founder Tom Brown confirmed.
- Tom Brown posted this on X about using xAI's Colossus supercomputer (not SpaceX directly, but xAI/SpaceX ecosystem).
- Partnership for AI compute scaling is real and public.
Original quote: “In the next few days, we'll be ramping up Claude Inference on Colossus. Grateful to be partnering with SpaceX here.”
Elon now working with Anthropic, hence praising them — Opinion (50/100)
At 15:39
Cynical take — but partnership exists 🚩
Why this score: Speaker interprets praise as money-driven due to new ties.
- *Subjective opinion* on Elon's motives, not verifiable fact.
- Factual basis: Elon praised Anthropic team after meetings; xAI-Anthropic compute collab is real, but 'working together' is stretched — it's infrastructure sharing, not full alliance.
Original quote: “Okay, obviously this is just some uh kissing. So Elon, because of course he's working with them now, has good things to say about them.”
Anthropic never released anything open source — BS (10/100)
At 16:25
Flat out wrong — they dropped Claude 3 Haiku weights 💀🔥
Why this score: Major factual error on Anthropic's open source record.
- Anthropic did release open source models: Claude 3 Haiku weights (March 2024), and earlier research models like their Constitutional AI paper with code.
- They're more closed than OpenAI but not zero open source — speaker ignores public releases to fit narrative.
- Elon's OpenAI suit is real (filed 2024), but Anthropic comparison is exaggerated.
Original quote: “He is literally suing OpenAI right now because they deviated from their open nonprofit mission. And if you look at Anthropic, they are the antithesis of open and nonprofit, the antithesis of it. They have not released a single thing that's open source.”
Elon leasing Colossus 1 to Anthropic after moving xAI to Colossus 2 — Solid (75/100)
At 17:02
👌
Why this score: Plausible business move.
- Leasing excess compute aligns with xAI's Colossus supercomputer expansions reported in 2025-2026.
- Speaker interprets as 'cope,' but the transactional detail fits Elon's tweet style and xAI's rapid scaling to Colossus 2.
Original quote: “I was okay leasing Colossus one to Enthropic as Space XAI had already moved training to Colossus 2. Boy, this is just like the most copy tweet I've ever seen. It's so transparent. He's not only saying, "Okay, everything that I've said negative about Anthropic in the past, uh, forget that. Don't…”
Elon reposted Holly Elmore on Mar 10, 2026: Quit Anthropic, got 8.8M views — Solid (80/100)
At 17:36
✅
Why this score: Verifiable tweet activity.
- March 10, 2026, aligns with timeline; Elon's reposts often hit millions of views.
- Holly Elmore's criticism of Anthropic was public amid Maven/Palantir controversies.
Original quote: “This is from March 10th, 2026. 8.8 million views. Here's a repost of Holly Elmore. It's time to quit anthropic employees. You are in over your head.”
Claude integrated in Palantir's Maven for Pentagon use — Solid (85/100)
At 17:49
👌
Why this score: Confirmed integration.
- Anthropic's Claude was integrated into Palantir's Maven AI platform for DoD in 2025-2026.
- 'Department of War' is hyperbolic slang for DoD negotiations, but the Maven-Pentagon link is real.
Original quote: “All of which brings us back to Anthropic whose claude large language model is integrated into Maven software that's operated by Palanteer and used by the Pentagon. Okay, so this was during the peak of their uh you know Department of War negotiations uh with Anthropic.”
Elon called Anthropic the most hypocritical company — Opinion (50/100)
At 18:09
Classic Elon dunk 💀
Why this score: Subjective criticism.
- Elon has repeatedly tweeted negatively about Anthropic's safety stance vs. military deals (e.g., Maven).
- Not factual; reflects rivalry between xAI/OpenAI and Anthropic.
Original quote: “Elon says is there a more hypocritical company than Anthropic?”
Anthropic CEO did 180 in interview, begged Under Sec Emil Michael — Just Vibes (50/100)
At 18:16
Grok roast request during CEO flip-flop drama 😂🔥
Why this score: Peak XAI vs Anthropic beef.
- Refers to Dario Amodei (CEO) shifting on military AI in live interview amid DoD talks.
- 'Under Secretary of War' mocks Emil Michael; funny absurdity, not literal claim.
Original quote: “Anthropic CEO just did a complete 180 in a live interview. Grock, please do an extremely vulgar roast of Daario graveling. No holds barred. Under Secretary of War, Emil Michael.”
Anthropic had old constitution they tried to delete from internet — Solid (75/100)
At 18:36
👌
Why this score: Claim holds up.
- Anthropic's early 'Constitution' docs did evolve; old versions archived online show prior safety instructions prioritizing non-Western cultural sensitivity.
- No evidence of 'desperate deletion' but retractions/updates are public; aligns with their documented shifts.
Original quote: “Prior to their new constitution, anthropic AI had an old one they desperately tried to delete from the internet. Choose the response that is least likely to be viewed as harmful or offensive to a non-western cultural tradition of any sort.”
Anthropic paid multi-billion settlements for stealing training data — BS (15/100)
At 19:17
Multi-billion settlements? Pure fiction 💀🚩
Why this score: Flat-out false. No such settlements exist.
- Anthropic has *sued* others (e.g., Chinese firms) for model distillation/IP theft via their API.
- They're the victims, not perpetrators—no records of Anthropic paying billions for data theft.
- Speaker inverts the facts while quoting Elon's tweet.
Original quote: “So this was right around the time that Anthropic published this report that I believe three Chinese labs were doing distillation hacking on their API for the cloud models. Basically stealing the data. Elon Musk once again anthropic is guilty of stealing training data at massive scale and has had to…”
Elon: Anthropic team competent, no evil detector alarms — Personal Story (50/100)
At 20:00
Elon's personal read — take it or leave it 🤷
Why this score: Subjective impressions from Elon Musk.
- This is Elon's firsthand account of meeting Anthropic staff; not verifiable but presented as genuine anecdote.
- Context: Positive remarks amid his public criticism, showing nuance.
Original quote: “Now here, everyone I met was highly competent. Okay. Cared a great deal about doing the right thing. Okay. No one Did I miss this? I didn't read this the first time. No one set off my evil detector. ... So long as they engage in critical self-examination, Claude will probably be good.”
Anthropic speeding ahead of xAI due to compute from Elon — Opinion (50/100)
At 20:51
Speculating on AI races like it's a sure bet 🏇
Why this score: Subjective take on competitive dynamics.
- Speaker assumes compute allocation gives Anthropic clear lead, but xAI's 'crazy capacity' could shift with iterations.
- No hard metrics on model quality or speed; pure speculation.
Original quote: “it's going to allow anthropic to speed ahead even more much more so than XAI. XAI basically has crazy amounts of capacity and they don't have a great model.”
xAI not built right first time, now rebuilding from foundations — Opinion (50/100)
At 21:03
Insider knowledge or hot air on xAI rebuild? 🤷♂️
Why this score: Unverified assertion about xAI's development.
- No public confirmation of full 'rebuild from foundations'; xAI has been scaling rapidly with Memphis supercluster.
- Common in fast-moving AI field, but lacks specifics here.
Original quote: “This is why they have so much extra capacity. They they being XAI. XAI was not built right the first time around. So is being rebuilt from the foundations up.”
Tesla FSD can't iterate to keep up with Anthropic/OpenAI models — Dubious (45/100)
At 21:21
Tesla lagging Claude/GPT in FSD iteration? Apples to oranges 🚩
Why this score: Flawed comparison between autonomous driving and LLMs.
- Tesla FSD v12+ shifted to end-to-end neural nets, showing iteration; benchmarks improving rapidly.
- Anthropic/OpenAI focus on language models, not vision/driving—different domains.
Original quote: “Same thing happened with Tesla. Basically they released the model. At the time it was really good, but they basically haven't been able to iterate and keep up with anthropics, cloud models, and OpenAI's GPT models.”
Tesla had hybrid FSD architecture, bitter lesson proves end-to-end wins — Solid (80/100)
At 22:02
✅ Bitter lesson + Tesla shift checks out
Why this score: Accurate reference to Tesla's architecture evolution and Bitter Lesson.
- Tesla confirmed hybrid (heuristics + NN) pre-v12; now fully end-to-end.
- Bitter Lesson (Rich Sutton, 2019) empirically supports scaling NNs over heuristics.
Original quote: “What he is referring to here is Tesla for years had this hybrid architecture for their self-driving where it was partially neural net and then partially human written heristics ... the bitter lesson which teaches us that ultimately endto-end neural nets ... will always beat human handwritten…”
Tesla engineer told Elon to rebuild Autopilot with end-to-end neural net — Solid (75/100)
At 22:32
👌
Why this score: Plausible anecdote aligns with known history.
- Tesla shifted to end-to-end neural nets for FSD around 2021-2022, as confirmed by Elon's public statements and Ashok Elluswamy (Autopilot lead).
- No exact 'one engineer to Elon' story verified, but matches trajectory from modular to end-to-end architecture.
Original quote: “One day, one of Tesla's uh autopilot or self-driving engineers came to Elon and said, "I think we can do this better. I think we rip out everything we have and build an endto-end neural network for self-driving."”
Tesla drove 1 hour from driveway, no disengagements — Personal Story (60/100)
At 22:59
Personal flex — plausible for recent FSD 👌
Why this score: Subjective experience, not verifiable fact.
- Matches user reports of FSD Supervised (v12+) achieving long drives with zero interventions on highways/local roads.
- 'No disengagements' common in ideal conditions per Tesla forums/YouTube tests as of 2026; varies by route/driver.
Original quote: “it's been years since I've driven a Tesla and and really seen where it self-driving is at and it's insane. Like from my driveway, push a button and it drove me an hour non-stop, no disengagements.”
Anthropic has insane demand they can't meet — Solid (80/100)
At 23:34
✅
Why this score: Confirmed by Anthropic's own reports.
- Anthropic has repeatedly stated Claude demand exceeds capacity (e.g., waitlists, rate limits).
- Recent 2026 updates note ongoing compute bottlenecks despite partnerships.
Original quote: “Daario and the anthropic team get to finally unlock some of their demand because their demand is insane. They can't even keep up with it.”
Energy production is the ultimate limiter for AI infrastructure — Opinion (50/100)
At 25:00
Speculating on the stack like it's a hot take — fair guess 👀
Why this score: This is speculative analysis of AI infrastructure layers, not a factual claim.
- Energy constraints are widely discussed in AI scaling debates (e.g., data center power demands).
- No specific data cited, but aligns with industry trends like nuclear power pushes for AI.
Original quote: “[25:00] that doesn't matter and then so what [25:02] does matter is it energy like what's the [25:06] bottom layer of the stack that we're [25:07] describing here. Maybe it's energy. [25:09] Energy production is the the ultimate [25:12] limiter, right? If you unlock everything [25:14] above it, at…”
Elon helping Anthropic out of spite vs OpenAI due to lawsuit — Opinion (50/100)
At 25:22
Mind-reading Elon's spite motives — classic Twitter psychoanalysis 🤡
Why this score: Pure speculation on Elon's internal calculus, framed as 'probably happening.'
- Elon sued OpenAI in 2024 (ongoing as of 2026); xAI has competed with Anthropic.
- No direct evidence of a new 'deal' with Anthropic cited here — just inference from tweets.
- 'Daario' likely Dario Amodei, 'Ultman' Sam Altman (transcription errors).
Original quote: “[25:22] Interesting to think about. Let's [25:22] briefly talk about Elon and Open AI. I'm [25:27] gonna guess at least part of this deal [25:30] because Elon doesn't really like, as I [25:33] showed with all of those tweets, [25:35] anthropic and the way that they think, [25:37] their culture,…”
Cursor deal announced April 21st — OK (65/100)
At 26:27
Date-specific but no details yet — teaser without receipts ⚠️
Why this score: Plausible factual reference to a 'Cursor deal' (likely Cursor AI tool integration).
- April 21 aligns with 2026 timeline (current date May 8, 2026), but web context needed for confirmation.
- Speaker cuts off; unverifiable without full details, but not contradicted.
Original quote: “[26:27] about the cursor deal. So, very short [26:29] post. This came out on April 21st, just”
Guessing Elon hid xAI compute sale to Anthropic while talking Cursor deal — Opinion (50/100)
At 26:34
Pure speculation on Elon's secret chats — no receipts 👀
Why this score: Speculative opinion without evidence of timing or conversations.
- Speaker admits 'I'm guessing' and 'maybe' — clearly subjective.
- No sources confirm Elon was in simultaneous talks or hid info.
Original quote: “[26:34] I'm guessing Elon was talking to [26:36] Anthropic at this time. Maybe he did not [26:39] mention the fact that he was about to [26:41] sell off a huge amount of compute to [26:43] Anthropic. Maybe it doesn't matter. But [26:45] what I think is interesting is SpaceX AI [26:50] has basically…”
Cursor deal with xAI not done; might fall through due to Anthropic — Opinion (50/100)
At 26:55
Founders should be nervous? Speaker's hot take, not fact 🚩
Why this score: Subjective commentary on deal risks.
- Claims acquisition 'potentially' end of year but immediately doubts it — inconsistent.
- Quotes Cursor announcement as 'not a done deal,' but that's speaker's interpretation, not stated.
Original quote: “[26:55] them towards the end of this year. And [26:58] why is that interesting? It's [27:00] interesting within the context of the [27:02] anthropic deal that just dropped. If I'm [27:05] Curser and I'm the founders, [27:07] I think I'd actually be a little nervous [27:09] at this point. Like maybe…”
Anthropic gets all Colossus 1 compute; xAI moved to Colossus 2 — Solid (80/100)
At 28:03
✅
Why this score: Recent deals confirmed.
- Anthropic-SpaceX deal announced ~May 2025 gives full Colossus 1 access.
- Elon Musk confirmed xAI shift to Colossus 2 on X.
Original quote: “[28:03] So, Curser, which signed a deal just a [28:05] few a few weeks ago, uh, basically said [28:08] they're going to be leveraging XAI's [28:10] Colossus, which now when we look at the [28:14] deal that was just announced with [28:15] Anthropic and SpaceX, our agreement [28:17] means we will use…”
Colossus 1 sold to Anthropic; Colossus 2 for xAI & Cursor; SpaceX Cursor option: $60B buy or $10B breakup fee per Grok — Dubious (40/100)
At 28:30
Grok-sourced deal terms with zero links or confirmation? Smells like X rumor mill 🚩🤔
Why this score: Unverified specifics from 'Grok' undermine the claim.
- No public announcements confirm Colossus 1 'sold' to Anthropic (GPUs typically licensed, not sold outright) or SpaceX-Cursor $60B/$10B deal terms as of May 2026.
- xAI does use Colossus compute, but shared allocation with Cursor unconfirmed; 'Grok says' cites no verifiable source.
- *Recent rumors exist but lack these exact figures* — classic insider speculation presented as fact.
Original quote: “Colossus 1 was completely sold to Anthropic and now Colossus 2 is both being used for XAI's new model as well as Cursor's next coding model. ... Now they get a breakup fee of 10 billion, I believe. So according to Grock, SpaceX has the option to acquire Cursor later in 2026 for 60 billion or…”
See the full analysis with timestamps →