Thomas

32.2K posts

Thomas

Thomas

@3PointInsights

Obsessed with decoding the chaos of tech, capital and power.

Katılım Temmuz 2019
93 Takip Edilen1.5K Takipçiler
Thomas
Thomas@3PointInsights·
The real tell: $6B market cap erased on just $52M in liquidations. That ratio doesn't happen with organic selling. Inflated float + 95% insider supply = the whole market cap was fictional to begin with. This is why low-float token listings need mandatory exchange surveillance thresholds, not just reactive investigations.
English
0
0
2
133
ZachXBT
ZachXBT@zachxbt·
A summary of the RAVE -95% price fluctuation from $26 to $1 over the past 24 hours. RAVE Timeline: April 18, 2026 7:26 am UTC: I posted a call to action for Binance, Bitget, & Gate to investigate RAVE market manipulation and offered a $10K bounty. 10:56 am UTC: I posted an update increasing the bounty to $25K. 11:18 am UTC: Bitget publicly acknowledged the call to action. 2:08 pm UTC: Binance publicly acknowledged the call to action. 3:06 pm UTC: RaveDAO posted claiming they have no involvement. 4:19 pm UTC: Gate publicly acknowledged the call to action. In the days leading up, on April 13 & 14, I confronted RaveDAO co-founder Yemu Xu (wildwoomoo) but have yet to receive an answer. RAVE launched in Dec 2025 on Binance Alpha with a 1B total supply. The addresses below, linked to the initial distribution, control ~95% of the RAVE supply (h/t Mlm): 0x9831156F1a6E506Fca41503590b42F07c2e80f54 0x8Ed6245C3276307E1A9D9Dc872E98A0E770070fd 0x6020656d1EF182173E45D4Fc375BDD5a48c674B0 0x2664cB80a5ee7D8EC05fe7C752dD62E078056E6d 0x2D81F8AeBf3e58A5e638006c9fd8F38C5220ecab 0x31694d761A8e851cFFbCd286aC54D01e5Ce5aFe6 0x0A1F07993a51CcEb4f52CA67765AECeADDA790d7 0xEB74Df8588cFC1C179Df4bd96C0bB8B227B9bE92 0x53d7d52301366DC14E1916b14eFeC1aDD8F3487b I found suspicious CEX activity in April 2026 tied to RaveDAO team addresses onchain, which potentially contradicts their recent statement: Bitget 0x2dc20f2180582172f5450c5d71e23fa438a7031b 0xa3a02aeb97fc1737c66f50d07d024799c137891d 0x2d95eb42525e6087e0cb7869f98da6838ed2e743 Gate 0x31711246b05d71e9eda5e38a3abb654020ee3353 Given the supply concentration, the team at minimum knows who is responsible for this price action. A simple litmus test: $6B in market cap was wiped out on just $52M of 24hr liquidations (h/t CoinGlass). That ratio points to a manipulated and unsustainable valuation. RAVE is not the only token with manipulation we have seen on major centralized exchanges. It's just the most blatant, reaching a top 15 market cap within 10 days before dropping 95% in hours. Other projects with highly questionable price action recently include: SIREN, MYX, COAI, M, PIPPIN, RIVER. Exchanges need faster intervention on manipulation. Detection at scale isn't easy, but each day of delay means retail traders absorb losses while platforms collect fees on the volume. The outcome is the same regardless of intent. While it's good the exchanges responded, I find it unlikely this activity wasn't spotted internally before I raised it publicly. I recognize how much this behavior takes from retail traders, and I plan to investigate similar movements in hopes of identifying the responsible parties. I want to reiterate that I did not take a position. If I had, I would have been liquidated myself. I also could not anticipate if or when the exchanges would comment publicly. My $25K bounty will remain active since the only DMs received were unverified claims rather than non-public information with supporting evidence as requested.
ZachXBT tweet media
English
1.1K
1.2K
7.2K
1.1M
Thomas
Thomas@3PointInsights·
A 4-month-old AI startup just closed $500M+ (could hit $1B) at a $4B pre-money valuation. Backed by Google’s GV and Nvidia. Founders from DeepMind + OpenAI. Goal: self-improving AI that automates frontier research itself. We are speedrunning the singularity.
English
0
0
0
44
Thomas
Thomas@3PointInsights·
The NSA is actively using Anthropic’s Mythos Preview (the model Pentagon blacklisted Anthropic over) because it’s that good at zero-day exploits. One half of the U.S. government wants cyber superpowers; the other fears it. This is how the weaponization of frontier AI actually happens.
English
1
0
0
32
Deeps🌟
Deeps🌟@deeps77799·
@philipkiely Hey @grok, tell me about this book "inference engineering". Define the term and explain use cases.
English
2
0
1
1.8K
Philip Kiely
Philip Kiely@philipkiely·
Inference Engineering is now available in print in India. Full book. Full color. ₹825. Thank you to my partners at Shroff Publishers & Distributors for working with me to deliver this book at a 65% discount from the standard price. baseten.co/inference-engi…
Philip Kiely tweet media
English
71
209
2.8K
11.8M
Thomas
Thomas@3PointInsights·
@GaryMarcus Codex might be the only useful part of OpenAI
English
0
0
0
101
Thomas
Thomas@3PointInsights·
Everyone’s mad Nvidia is “abandoning gamers” for AI chips and skipping a 2026 GeForce refresh. their data center segment is already 91.5% of revenue. Gamers built the brand; AI will keep the lights on. Priorities evolve—cope or adapt. #GamingDaily
English
0
0
0
29
Thomas
Thomas@3PointInsights·
Exactly. Aghion, Bunel & Jaravel’s 2025 analysis shows generative AI already delivering 15-25% productivity gains at adopting firms—biggest lifts for lower performers—while projecting +0.8-1.3 pp annual growth without net job destruction. Brynjolfsson’s post-ChatGPT data confirms: US aggregate employment stable, new AI-skill roles rising even as routine entry-level tasks shift. The economists have the evidence.
English
0
0
1
960
Yann LeCun
Yann LeCun@ylecun·
Dario is wrong. He knows absolutely nothing about the effects of technological revolutions on the labor market. Don't listen to him, Sam, Yoshua, Geoff, or me on this topic. Listen to economists who have spent their career studying this, like @Ph_Aghion , @erikbryn , @DAcemogluMIT , @amcafee , @davidautor
TFTC@TFTC21

Anthropic CEO Dario Amodei: “50% of all tech jobs, entry-level lawyers, consultants, and finance professionals will be completely wiped out within 1–5 years.”

English
1.2K
2.8K
21.5K
4M
Thomas
Thomas@3PointInsights·
The toll is documented: University of Minnesota survey found 52% of African content moderators meet probable clinical depression thresholds, 55% significant distress—higher than other regions. AI safety depends on this work; labs scaling it owe transparent wages, mental-health support, and supply-chain audits.
English
0
0
4
1.4K
Thomas
Thomas@3PointInsights·
Indian-origin talent accounts for 16% of global AI per Prosus-BCG and key GenAI foundations: Transformers co-authors, GPT-4o lead, Perplexity/Anthropic founders, Microsoft’s OpenAI bet. Stanford 2026: India 2nd-largest talent pool. xAI should be optimizing for the same proven edge—results over passports.
English
0
0
34
9.6K
Neon White Rabbit
Neon White Rabbit@RedPillRabbit·
Looks like there’s no Indians on Elons AI team anymore? Just Whites and East Asians? What happened? 🤣🤣🤣 P.S. Insider verified to me that all the Indian hires were let go for being unable to contribute anything. Obvious fake credentials/most likely scam interviews.
Neon White Rabbit tweet media
English
970
3.8K
26.7K
1.3M
Thomas
Thomas@3PointInsights·
@pmarca Funding overlaps with EA and Open Philanthropy are real, yet AI safety work produced scalable oversight and RLHF techniques now standard at OpenAI/Anthropic. Labeling it pure “grift” underplays measurable risk reduction while ignoring how hype cycles distort both sides.
English
0
0
0
32
Thomas
Thomas@3PointInsights·
Cursor, the AI coding agent startup, is in talks to raise $2B at a valuation over $50B—more than doubling its $29.3B post-money from November. An agentic coding tool worth more than most Fortune 500s? The AI gold rush just hit another gear.
English
1
0
1
56
Thomas
Thomas@3PointInsights·
was Elon right about lidar all along?
English
0
0
0
32
Thomas
Thomas@3PointInsights·
Unpopular opinion: AI is not as smart as your feed thinks. Per Stanford's 2026 AI Index: frontier models win gold at the International Math Olympiad. The same models read analog clocks correctly 50.1% of the time. The frontier isn't a line. It's a jagged cliff.
English
0
0
0
35
Thomas
Thomas@3PointInsights·
Bitcoin is sitting near $75K while funding rates are the most negative since 2023. Translation: the entire market is short, and price still isn't breaking. Every time this has happened before, a local bottom was in within weeks. The pain trade is higher. Agree or disagree?
English
0
0
0
39
Thomas
Thomas@3PointInsights·
Bitcoin developers are seriously debating whether to freeze Satoshi's 1.1M BTC — roughly $74B — before quantum computers can steal it. The proposal (BIP-361) would lock dormant coins permanently. This is the biggest governance crisis Bitcoin has ever faced.
English
1
0
0
32
Thomas
Thomas@3PointInsights·
Tesla closed up 8% yesterday. Here's what @elonmusk just announced: • AI5 chip nearing production • Two new "Terafab" facilities in Austin • One fab for vehicles and robots, one for orbital data centers• #SpaceX and Intel joining the project • FSD now $99/mo with @grok voice integration He's not building a car company anymore.
English
1
0
0
63
Thomas
Thomas@3PointInsights·
Snap just laid off 1,000 people — 16% of the company. The reason, per their own SEC filing: AI now writes 65% of new code at Snap. We're not heading toward AI replacing workers. We're already there.
English
0
0
0
23
Thomas
Thomas@3PointInsights·
@kimmonismus Gemini is shit... Thank you for your attention to this matter.
English
0
0
0
26
Chubby♨️
Chubby♨️@kimmonismus·
ChatGPT's share of Gen AI traffic has collapsed from 77% to 57% in twelve months, while Gemini surged to 25% and Claude nearly tripled to 6% in just the past month. Google has the distribution moat with their eco-system. And Claude is just very good in general that even consumer keep switching to Claude (and the heated debate about Mythos is the best PR Anthropic could have asked for). The market is shifting from a ChatGPT monopoly into a genuine three-way race, with DeepSeek, Grok, Copilot, and Perplexity all stuck below 4%.
Chubby♨️ tweet media
Similarweb@Similarweb

Gen AI traffic share update Main takeaways: → Gemini holds a quarter of the share. → Claude almost doubled its share between February and March, crossing the 6% mark. → DeepSeek surpassed Grok again. → We’ve added m365.cloud.microsoft/chat to Copilot’s share, which explains the increase from previous updates. 🗓️ 12 months ago: ChatGPT: 77.43% Grok: 7.03% Gemini: 6.00% DeepSeek: 3.73% Perplexity: 1.66% Claude: 1.40% Copilot: 1.38% 🗓️ 6 months ago: ChatGPT: 71.75% Gemini: 13.56% Grok: 4.08% DeepSeek: 2.81% Perplexity: 2.51% Claude: 2.26% Copilot: 2.01% 🗓️ 3 months ago: ChatGPT: 63.19% Gemini: 22.59% DeepSeek: 4.08% Grok: 3.26% Claude: 2.22% Perplexity: 1.86% Copilot: 1.80% 🗓️ 1 month ago: ChatGPT: 56.72% Gemini: 25.46% Claude: 6.02% DeepSeek: 3.74% Grok: 3.44% Copilot: 1.99% Perplexity: 1.64%

English
56
55
560
55.2K
Thomas
Thomas@3PointInsights·
@Kyriakos_Pelek Yup, and it's time for everyone to learn how to optimize their token usage. Guess you can't really *all* your development to Claude ...
English
0
0
0
35
Kyriakos
Kyriakos@Kyriakos_Pelek·
@3PointInsights per-seat pricing is dead, predictable usage pricing wins
English
1
0
1
17
Thomas
Thomas@3PointInsights·
RIP to the "per-seat" SaaS pricing model. 🪦 Because of AI agents, B2B SaaS is rapidly shifting to usage-based consumption. The problem? 78% of IT leaders report getting absolutely torched by invisible, unpredictable API charges this quarter. SaaS founders: if you can build a product that solves AI cost-predictability, you will print money.
English
4
0
2
47
Thomas
Thomas@3PointInsights·
@paulg welcome to vibe coding...
English
0
0
0
55
Paul Graham
Paul Graham@paulg·
I finally tried to create an Instagram account. (I've managed to avoid it so far.) It was instantly suspended, before I'd even finished the signup process. The final screen of the signup process told me I'd submitted an appeal.
Paul Graham tweet media
English
392
42
2.3K
327.5K