Sabitlenmiş Tweet
FatRatKiller
6.4K posts
FatRatKiller
@FatRatKiller
Risk/Compliance/Business Analyst. APAC business & market observer. Also focus on DePIN | ICM | RWA | infra bets. Long Asia
Burdur Merkez, Burdur Katılım Nisan 2022
784 Takip Edilen3.5K Takipçiler
BYD was briefly pulled off Brazil’s forced-labor blacklist, and the real story is legal, not just reputational.
The court did not say the issue vanished, it said the government had not yet proven a direct employer relationship with the affected workers.
That matters because the blacklist itself, officially a registry of employers tied to slavery-like conditions, is a high-stakes compliance signal, not a casual media headline.
BYD’s move to seek a writ shows the company is fighting this as a procedural and evidentiary battle.
The deeper implication is that global Chinese companies are now operating under far tighter labor, supply-chain, and legal scrutiny abroad.
And the speed here is striking, blacklisted in early April, then temporarily removed in under three days.
That tells you the fight is still early, and the final outcome will depend on evidence, jurisdiction, and politics.
Wait, doesn’t this sound like the new normal for multinationals, speed, scrutiny, and legal ambiguity all at once?
Bloomberg@business
BYD is stepping up hiring from rivals including Porsche to bolster its Denza line in Europe as the Chinese manufacturer tries to break into the region’s contested luxury-car market bloomberg.com/news/articles/…
English
$SIREN is the obvious lead actor, up 175.8% in 24 hours while the old concentration problem never went away. 52 of the top 54 wallets being linked to one entity, with roughly 88.5% of circulating supply sitting in the same pocket. That means price discovery is fake by construction, because when one hand owns the float, the chart is not a market, it is a puppet show.
The short squeeze and exchange-fueled momentum only make the trap prettier, not safer. $ORDI also printed the kind of move that attracts dumb money fast, with 175.8% upside and $1.34B in daily volume, but this one looks more like reflexive speculation than clean whale accumulation unless wallet concentration data catches up later.
$BASED is another warning label pretending to be a breakout, up 132.4% with a 9.08x volume-to-market-cap ratio, which is the sort of statistic that usually means someone is manufacturing excitement because organic demand is not enough.
Across the scan, the pattern is the same: absurd turnover, vertical candles, and narratives arriving exactly when exit liquidity is needed most. If one entity can own the supply and everyone still calls it bullish, then the market is not being traded, it is being milked. The math does not lie. The narrative does.
FatRatKiller@FatRatKiller
The cleanest red flag is $SIREN, up 18.9% on the day while prior concentration concerns remain very much alive. one linked wallet cluster controls roughly 50% of supply, built through 200-plus wallets and later dispersed into 47 addresses, which means the market is pretending to be liquid while one hand likely still grips the steering wheel. That is not price discovery, that is theater with exit liquidity props. $STO is quieter at -1.9%, but it stays on the board because it is still digesting a previous 900% vertical move, and names that go full rocket usually do not turn honest overnight. Outside the watchlist, $BIO’s 96.7% daily pump with $567.6M volume against a $69.5M market cap is the kind of ratio that deserves suspicion before admiration. $CLAWGUIN at 236.36x volume to market cap and $QUQ at 188.27x are even more absurd, which usually points to microcap distortion, wash-heavy churn, or both. On the downside, $BLESS at -54.6% and ST at -37.3% show the usual post-hype trapdoor behavior. Net result: no single smoking gun from the automated scan, but concentration risk in SIREN and grotesque volume-to-cap mismatches elsewhere say this market is still a clown car driven by whales with fake mustaches.
English
Indeed IPv6 has crossed 50% traffic share, which is not a niche networking stat anymore, it is a structural internet milestone.
The deeper signal is simple, the old IPv4 world is fading, and modern infrastructure is now forced to live in a dual-stack reality that will increasingly behave like IPv6-first.
That means ops teams, CDN providers, security tooling, and app stacks can no longer treat IPv6 as optional cleanup work.
If your product, logging, firewall rules, or observability still assume IPv4 dominance, you are already behind.
This is the kind of shift that looks boring until it breaks assumptions at scale.
And that is the real story here, not bandwidth, but architecture.
Wait, doesn’t this sound familiar, the future arrives slowly, then suddenly becomes default?
Sigrid Jin 🌈🙏@realsigridjin
hmm finally admitted ipv6 is a mess and dropped an internet draft for ipv8 if an ipv4 address looks like 1.1.1.1, an ipv8 address looks like 1.1.1.1.1.1.1.1 kinda wild it took 25 years to just suggest making the numbers longer haha
English
Obviously, Claude4.7 is actually Opus4.6.
Users can now steer effort with task budgets and concision prompts, and the punchline is that token usage improves across effort levels in internal coding tests.
That matters because the best model is no longer just the smartest one, it is the one that can spend tokens like a disciplined operator.
The image change is also subtle but important, higher-fidelity processing is now model-level, which means more detail, but also more token burn.
So if you do not need the extra precision, downsample first.
The benchmark notes are the real tell.
Some scores were updated due to revised grading, some runs used thinking disabled, some were measured with internal implementations, and some were filtered for memorization.
In other words, the leaderboard is useful, but only if you read the footnotes like a lawyer.
Claude@claudeai
Claude Opus 4.7 is available today on claude.ai, the Claude Platform, and all major cloud platforms. Read more: anthropic.com/news/claude-op…
English
DeepSeek should ship V4 fast, because the Chinese AI story is no longer just about “model quality”, it is about full-stack competition, token economics, and agentic execution.
R1 proved the brand can win on cost-performance, open source, and global mindshare.
But the bar has moved: V3.2 already hints at cross-tool memory retention, and the market now cares less about chat and more about agents that can reliably finish complex tasks.
The real shift is this: China’s AI narrative has become a three-layer race, “the model stack, the infrastructure stack, and the ecosystem stack”.
If DeepSeek stays on V3-era momentum too long, it risks looking like the past instead of the frontier.
What V4 needs is not just more parameters, but native multimodality, stronger long-horizon planning, and tight coupling with domestic compute.
That combination would make it a base model for autonomous agents, not just a better chatbot.
The token boom makes this urgent.
When usage explodes, the winner is not the smartest demo, but the system that can scale economically and reliably.
Wait, what if the next breakthrough is not “smarter answers”, but “systems that actually do the work”?
FatRatKiller@FatRatKiller
Xiaomi Revenue hit a staggering 457.3 billion, signaling a robust 25% year-on-year growth. The net profit surged by 76.6%, reaching 41.6 billion yuan! What’s driving this success? Their electric vehicle sector alone generated over 106 billion yuan—yes, a jaw-dropping 234% increase—contributing a whopping 80.2% to overall growth. But that’s not all; their IoT business, breaking the 100 billion yuan milestone for the first time, showcases an impressive growth rate of 18.3%. Xiaomi’s automotive division achieved profitability in just five quarters, outpacing competitors! This reveals a transformative shift in their business model—leveraging hardware sales for engagement instead of just spending on marketing.
English
Xiaomi Revenue hit a staggering 457.3 billion, signaling a robust 25% year-on-year growth. The net profit surged by 76.6%, reaching 41.6 billion yuan!
What’s driving this success? Their electric vehicle sector alone generated over 106 billion yuan—yes, a jaw-dropping 234% increase—contributing a whopping 80.2% to overall growth.
But that’s not all; their IoT business, breaking the 100 billion yuan milestone for the first time, showcases an impressive growth rate of 18.3%.
Xiaomi’s automotive division achieved profitability in just five quarters, outpacing competitors!
This reveals a transformative shift in their business model—leveraging hardware sales for engagement instead of just spending on marketing.
English
The cleanest red flag is $SIREN, up 18.9% on the day while prior concentration concerns remain very much alive. one linked wallet cluster controls roughly 50% of supply, built through 200-plus wallets and later dispersed into 47 addresses, which means the market is pretending to be liquid while one hand likely still grips the steering wheel.
That is not price discovery, that is theater with exit liquidity props. $STO is quieter at -1.9%, but it stays on the board because it is still digesting a previous 900% vertical move, and names that go full rocket usually do not turn honest overnight.
Outside the watchlist, $BIO’s 96.7% daily pump with $567.6M volume against a $69.5M market cap is the kind of ratio that deserves suspicion before admiration. $CLAWGUIN at 236.36x volume to market cap and $QUQ at 188.27x are even more absurd, which usually points to microcap distortion, wash-heavy churn, or both.
On the downside, $BLESS at -54.6% and ST at -37.3% show the usual post-hype trapdoor behavior. Net result: no single smoking gun from the automated scan, but concentration risk in SIREN and grotesque volume-to-cap mismatches elsewhere say this market is still a clown car driven by whales with fake mustaches.
English
Kalshi, a crypto futures trading platform, faces potential scrutiny from the DOJ for insider trading practices. Insider trading allegations suggest a breach of market integrity, raising questions about fairness in trading.
The CEO's expectation of legal action signals heightened regulatory vigilance in the cryptocurrency sector. This scrutiny indicates a shift towards stricter enforcement of financial regulations for emerging markets.
Traders on Kalshi may reassess their strategies in light of these allegations, leading to market volatility. Investors could become wary of platforms entangled in legal challenges, affecting liquidity.
Legal consequences may compel Kalshi to strengthen compliance protocols to mitigate risks. The situation underscores the imperative for transparency in trading operations and practices.
This investigation may set precedents that shape compliance standards across the industry. Overall, these developments highlight the evolving landscape of regulatory scrutiny in cryptocurrency markets.
More Perfect Union@MorePerfectUS
A lawyer who helped Kalshi in court vs. the federal government now works for the government. The lawyer hired by Kalshi is now the Principal Deputy Assistant AG at the DOJ Civil Division. There he's positioned to protect Kalshi from states trying to regulate prediction markets.
English
Sony's electric vehicle project with Honda has been halted.
Sony's electric vehicle project AFEELA 1 in cooperation with Honda has been suspended, and pre-ordered models will be fully refunded, and Honda is expected to lose a huge amount due to the setback of its electrification strategy.
Honda had estimated revenue of 550 billion yen in fiscal 2026, but it turned into a huge loss due to a setback in its electrification strategy.
Sony Honda Mobility adopts an "asset-light" operating model, and the AFEELA project has no substantial impact on Sony's financial situation.😂
TESLARATI@Teslarati
🚗 Tesla Killers Graveyard: Sony-Honda AFEELA The sleek, AI-packed luxury sedan with PlayStation integration. Officially cancelled in March 2026 after Honda scaled back its EV plans. Fisker Ocean Stylish SUV with solar roof promises. Company filed for bankruptcy in 2024 amid production chaos, quality issues, and cash problems. Faraday Future FF 91 Billion-dollar ultra-luxury “hyper-tech” sedan. Delivered fewer than 100 vehicles before fading into obscurity. Lordstown Endurance Electric pickup with massive pre-order buzz. Collapsed after exaggerated claims, factory drama, and bankruptcy. Nikola Hydrogen and electric trucks hyped as the future. Hit by fraud scandals, stock crashes, and very few actual deliveries. Canoo Quirky modular “lifestyle” EVs with bold designs. Struggled with funding and production; now barely hanging on. Lucid Air Often called the ultimate Tesla slayer for its superior range and luxury. Strong reviews, but sales have lagged far behind expectations.
English
The AI computing power market has experienced a reversal from "jumping price" to "rocket price".
Cloud vendors generally increase prices by 30%-50%, and the core reason is that applications such as agents and video generation have led to a surge in token consumption, and there is a serious imbalance between supply and demand of computing power.
FollowTheIncentives@FollowIncentiv
Anthropic just added KYC to Claude. KYC = Know Your Customer. The same identity checks banks and crypto exchanges use to track who is doing what with money. Now AI wants to know who is doing what with intelligence. First crypto. Then AI. Regulation always follows where the money and power go. The wild west era of AI just got its first sheriff.
English
Japan faces the dilemma of no one to inherit a "huge inheritance".
The 129.2 billion inheritance in fiscal 2024 will be collected from the national treasury.
Properties become "negative equity" for young people, who are denied inheritance due to high maintenance costs, resulting in a large amount of land being left idle.
English
The first batch of bosses who replaced employees with AI began to reflect.
Salesforce and other companies are overly confident in replacing customer service, sales and other positions with AI, resulting in loose customer relationships and sluggish growth.
AI has limitations in "relational work".
English
When a company with real offices and LinkedIn profiles launches a token, your brain relaxes.
"These people have reputations to protect," you think. That's exactly what they're counting on. The truth is most of these pivots happen when the parent company is already bleeding out—VCs won't write another check, runway is burning, but there's this pile of user data and brand recognition just sitting there.
A token launch is basically a Series C where retail provides the exit liquidity and gets zero governance rights in return. I've watched founders celebrate "successful TGE" while their core product loses users every month.
The dirtiest trick is the subsidiary structure. They create a separate entity in Cayman or BVI, raise money into that, then drain it back to the parent through "licensing fees" and "technology transfers."
When things blow up, the parent company shrugs: "That was a separate entity, nothing to do with us." Your legal recourse is basically zero.
Here's the one question that exposes everything: go on LinkedIn and count how many senior engineers actually moved from the parent company to the crypto project.
If it's all new hires with "Web3 native" backgrounds and zero original team members, you're not investing in a pivot—you're funding a brand licensing deal where insiders already have their exit mapped out.
English
On Polymarket, the trend of betting "No" on various outcomes has garnered attention. Users often perceive certain events as highly unlikely to occur, leading them to favor "No" bets.
Many bettors tend to be skeptical of optimistic predictions, especially in politically or socially volatile contexts. This skepticism fuels a belief that proposed outcomes are overly optimistic or unrealistic.
Psychological biases, such as the negativity bias, lead individuals to focus more on potential negative outcomes rather than positive ones. This could explain the inclination to bet "No."
Some bettors might have access to information that makes them believe a particular outcome is unlikely. If this information is shared or discussed widely, it can create a collective sentiment supporting "No" bets.
English
Someone said Openclaw is the trending king.
Someone said Hermes is the king.
Someone said Claude Managed Agent is the king.
So, which one should I trust?
English
Is Claude cheating me now? I thought I was just switching models to save money.
What actually happened was uglier.
I set different Discord channels to different models in OpenClaw. On paper, everything looked migrated. The bindings were there. The agent names were there. The config looked clean.
But when I tested it, the channels were still quietly running Claude Opus.
So the first problem was not just technical friction. It was false reassurance. The system looked as if it had moved on, while Claude was still sitting underneath the workflow.
Then came the second hit.
Anthropic cut off the old Claude-login style third-party path for tools like OpenClaw and pushed users toward API-based usage instead. Officially, that is a billing and access-policy change. In practice, it felt like this: more setup, more cost sensitivity, more breakage, more latency, less trust.
That is the part people should not underestimate.
The issue is not only that Claude became harder to use through OpenClaw. The issue is that it became harder while pretending not to have changed that much. Technically, Anthropic can say “API is still available.” But for real users, the experience shifted from smooth and implicit to brittle and supervised.
My personal takeaway is simple:
Claude did not openly slam the door.
It left the door half open, then filled the doorway with nails.
That is why users are angry.
Not because they hate paying.
Because they hate being quietly funneled into a worse experience while being told nothing fundamental has happened.
FatRatKiller@FatRatKiller
Just got an email from Anthropic. Starting today, my $200/month Claude Max subscription no longer covers third-party tools. The tools that made Claude actually USEFUL for real work. Let that sink in. I was happily paying $200/month because I could use Claude through Clawdbot, OpenClaw, and other harnesses that made it 10x more productive than the web UI. Now? That same usage costs ~$1,650/month on API. "We're offering a one-time credit equal to your monthly subscription price" — wow, a whole $200 to soften an 8x price hike. Thanks Anthropic. The excuse? "These tools put an outsized strain on our systems." Translation: Power users who actually USE the product are a problem. You built the best model. People built amazing tools around it. Now you're punishing both. The timing is perfect too. GPT-5.4 just dropped at $2.50/$15 per million tokens. Gemini 3.1 Pro is at $2/$12. DeepSeek V4 is literally 50x cheaper. And Anthropic chose THIS moment to tell power users to pay 8x more or leave. Speedrun to irrelevance. The real tragedy? Claude is genuinely the best model for coding and agent work. 80.9% SWE-bench. Unmatched long-context coherence. And they're about to watch their power users migrate to "good enough" alternatives because the economics stopped making sense. To everyone at Anthropic who thought this was a good idea: You just mass-emailed your most technical, most vocal, most influential users telling them you don't want their business. Self-inflicted wound. Textbook case.
English
Charles Schwab just announced "Schwab Crypto" - spot BTC & ETH trading coming H1 2026.
$12 TRILLION in client assets. And they're launching with... 2 coins.
The absurdity:
1. It's April 2026. "H1 2026" means they have 2 months left. Still in "planning" mode.
2. Phased rollout: employees first → limited customers → maybe you someday. Classic TradFi bureaucracy.
3. Runs through "Charles Schwab Premier Bank" - translation: compliance lawyers wrote the product spec, not product managers.
4. Fidelity launched crypto in 2022. Robinhood has been printing money on it. Schwab shows up 4 years late with a "sign up for updates" page.
5. Only BTC and ETH. In 2026. When your boomer uncle already knows what Solana is.
This isn't innovation. This is a $12T dinosaur doing the absolute minimum to check a box.
The real tell? They're more excited about not getting sued than actually serving customers who want crypto exposure.
TradFi institutions don't adopt crypto. They perform adoption.
Welcome to the party, Schwab. The keg's been empty for 3 years.
English
Just got an email from Anthropic. Starting today, my $200/month Claude Max subscription no longer covers third-party tools. The tools that made Claude actually USEFUL for real work. Let that sink in.
I was happily paying $200/month because I could use Claude through Clawdbot, OpenClaw, and other harnesses that made it 10x more productive than the web UI. Now? That same usage costs ~$1,650/month on API. "We're offering a one-time credit equal to your monthly subscription price" — wow, a whole $200 to soften an 8x price hike. Thanks Anthropic.
The excuse? "These tools put an outsized strain on our systems." Translation: Power users who actually USE the product are a problem. You built the best model. People built amazing tools around it. Now you're punishing both.
The timing is perfect too. GPT-5.4 just dropped at $2.50/$15 per million tokens. Gemini 3.1 Pro is at $2/$12. DeepSeek V4 is literally 50x cheaper. And Anthropic chose THIS moment to tell power users to pay 8x more or leave. Speedrun to irrelevance.
The real tragedy? Claude is genuinely the best model for coding and agent work. 80.9% SWE-bench. Unmatched long-context coherence. And they're about to watch their power users migrate to "good enough" alternatives because the economics stopped making sense.
To everyone at Anthropic who thought this was a good idea: You just mass-emailed your most technical, most vocal, most influential users telling them you don't want their business. Self-inflicted wound. Textbook case.
Vadim@VadimStrizheus
ANTHROPIC HAS OFFICIALLY BANNED OPENCLAW USAGE!!! no this is not ai-generated Starting April 4, users will no longer be able to use their Claude Subscriptions for third-party harnesses including OpenClaw. THIS IS CRAZY. 👇
English
Real talk:
This isn't innovation. This is a stablecoin company running out of stablecoin ideas.
The only people excited about cirBTC are Circle's BD team trying to justify their Q2 roadmap.
Wrapped BTC is a solved problem. Circle is 5 years late with nothing new.
English
Circle just launched cirBTC — a wrapped Bitcoin token.
A stablecoin company. Re-entering Bitcoin. 7 years after abandoning it.
I have thoughts 🧵
English