🧑‍🌾 @mlxapp. Ex Getco, ex top 3 trader on Bitmex. angel @usdai_official, @monad, @zellic_io, @ethena, @crusoeai. NFA https://teapawt.net/t.co/ZbRjcU0OqV
Joined June 2011
- Tweets1.7K
- Following497
- Followers3K
- Likes8.8K
Pinned Tweet
My take and worldview on 24/7 tokenized stocks - the future of finance to research and all-in.
I’d take Anthropic’s IPO allocation at $1.40T. I’d pass at the $1.61T Hyperliquid mark.
My base case is a $1.54T first close. Beyond day one, the question is how much operating profit Anthropic keeps as it spends billions more on compute.
Tang retweeted
I've changed how I work over the last few weeks.
I now have one lead agent for each area of focus, running all the time, and a chief of staff agent that oversees them. Each lead manages its own sub-agents that write code, test it, chase reviews, and coordinate with team members.
Everything I need to decide or review comes into one prioritized queue, a simple local web app. Each product review includes screenshots and a video demo of the feature. I clear the queue in batches, mostly by voice and keyboard shortcuts. The leads also bring back ideas I hadn't considered.
Working from one queue, instead of switching between dozens of chats, has cut a lot of mental overhead. The system keeps learning my preferences and distilling my judgment into a "Brian brain," so its recommendations and first drafts get better before they reach me. The goal is fewer corrections per change.
Every change still goes through the same human review and security checks. It is eliminating tasks, not human judgment.
What an exciting time to push the frontier of productivity, especially as we build the economy for agents (agentic finance). We should all keep pushing in this direction.
A bet with a 10% chance of returning 100x can be an excellent decision, even though it fails 90% of the time.
In baseball, one swing can score at most four runs. In business, one extraordinary success can pay for years of failed experiments.
My takeaway: being wrong often can be consistent with making good decisions. What matters is the cost of your failures and the scale of your successes.
Keep the failures survivable. Give the successes room to become enormous.
- @JeffBezos 2015 Shareholder Letter
Tang retweeted
The next decade won't reward tourists. It will reward people who combine audacity with discipline. Who understand that technology cycles come and go, but that building enduring companies requires clarity of purpose and consistency of execution.
Venture has always been about backing outliers. That hasn't changed. If anything, it matters more now.
WHERE DO WE BUY ALL THE TOKENS
This is f*cking insane. Former OpenAI CTO, Mira Murati:
"The models we work with today are turn-based. You talk, they talk, then they go off and think, and while they're thinking they're deaf and blind.
The next ones never stop listening. They take in audio, text and video all the time and answer in 20-millisecond chunks."
In just 2 minutes, she shows why the chatbot you type into is about to look like a fax machine.
Prompt Box → Chatbot → Agents → AI That Works Next To You In Real Time
A turn-based model waits for you to finish typing, then goes quiet while it thinks.
An interaction model sees and hears what you're doing, and can cut in the moment you start going the wrong way.
Anyone can type a prompt. Almost nobody is ready to work with an AI that never stops paying attention
Watch the full 2 minutes. This is worth your time.
Bookmark it, like it, save it, I don't care... this is the best thing you'll see on X today.
Tang retweeted
we have a winner for the most absurd AI startup ever
this Peter Thiel backed company uses AI robots to stab fish in the brain so they taste better and last up to 3x longer
it's called Shinkei and it modernizes a centuries-old Japanese technique
dozens of Michelin-star restaurants are already lining up to buy the fish
here's how it works:
> the problem is when a fish gets caught, it usually just suffocates on the boat
> that stress floods the meat with hormones and lactic acid, which hurts the flavor and texture
> the old Japanese fix for this was spiking the fish through the brain right after the catch, which stops all that
> but doing it by hand is impractical for most American fishing crews
> so Shinkei built a fridge-sized robot that sits right on the boat's deck
> the crew feeds in a fish and AI identifies the exact species, so it knows exactly where the brain and gills are
> the robot then drives the spike through the brain and cuts near the gills in about 7 seconds
> the fish goes into an ice slurry so the blood drains out
> so the fish doesn't sit there stressing out, and that tasty flavor and texture is preserved
example number 5,392,417 of how AI can be applied to literally anything
ALT A gloved fisherman on a boat feeds a live salmon head-first into the steel intake of a white Shinkei Systems robot, with the ocean behind.
Update on @MLXapp November 8, 2026:
mlx-activity.vercel.app/
1. Recorded AI work roughly tripled. Cumulative tokens from completed jobs rose from about 167M to 508M, all from verification work.
2. More dedicated capacity, slightly fewer seats overall. Dedicated seats rose from 18 to 25 (+39%), while free seats fell from 54 to 45. Total seats went from 72 to 70, and Macs seen over seven days fell from 67 to 57.
Invite people to mine with @MLXapp and earn a bonus on what they mine:
20% from people you invite directly
4% (20%²) from people they invite
0.8% (20%Âł) from the level after that
Bonuses come from a 300M MLX community pool, on top of what miners earn. Nothing is taken from them. Your bonus unlocks when the MLX it came from does.
To qualify, you need a Mac of your own that ran MLX Node in the last 7 days, and none of your Macs can be on the same network as the miner.
Your invite code is your wallet name.
The year is 2028.
@realDonaldTrump airdrops the new $250 bill to every American citizen to support the cause in the war against evil. And everybody gets a free Claude and OpenAI Plus account with the Trump @worldlibertyfi card. #USA
Top Trending token yesterday H/T @KaitoAI :
Trending Organizations Smart Following
1. MLXapp: +45 smart followers. Total 46 smart followers. First discovered on 2026-10-25. @MLXapp
Earliest Smart Followers:
@thetinaverse , @TangTrades , @0xljki
Description: @MLXapp is a decentralized AI compute network that allows users to contribute their Apple Silicon Mac's processing power to a collective supercomputer in exchange for MLX token rewards. The project is trending because it is being positioned by industry participants as a superior decentralized alternative to existing AI compute protocols by addressing specific latency, economic, and incentive bottlenecks in the current market landscape.
@mlxapp and the rest: The Current Comps Landscape
mlx-dashboard-tangtrades.ver…
1. @exolabs: Splits models across local devices via tensor/pipeline parallelism.
The bottleneck: Bound by local interconnect physics. Splitting a 70B+ model over residential Wi-Fi or consumer WAN introduces devastating inter-layer communication latency. Great for a home Mac lab on Thunderbolt; unviable for a global decentralized network.
2. Apple Minions: machinelearning.apple.com/re…: Distributes independent batch inference jobs across idle corporate Macs.
The bottleneck: Closed-wall, enterprise-internal tool. It eliminates local cluster latency by distributing independent jobs (embeddings/summaries), but it has no public marketplace, zero external monetization, no permissionless supply routing, and no token incentives.
3. @darkbloomai: Routes API inference to consumer Apple Silicon providers.
The bottleneck: Fragile trust and protocol boundaries. As an unverified prototype routing plaintext prompts, it lacks native consumer distribution and economic verification guarantees.
4. @prlnet: Subsidizes cloud/managed serverless inference with token emissions.
The bottleneck: Commodity data-center supply masked by inflationary token subsidies. Once emissions decay, you are competing on standard data-center unit economics against hyperscalers.
5. @opentensor (TAO): Subnet incentive mechanism.
The bottleneck: Supply-side benchmark gaming. Miners optimize to beat validator scoring heuristics rather than serving real-time consumer or developer latency/uptime SLAs.
@MLXapp: The Trojan Horse Distribution Model (Utility First, Network Second)
Every competitor tries to build an inference network from the outside in—demanding users donate hardware for speculative token yields (Bittensor/Pearl) or struggle with manual node clustering (EXO).
MLX inverts the cold-start problem through Harness (and other sick apps lined up - TBA):
Immediate Consumer Utility: Users download Harness because it’s the best native, zero-setup, local AI client for their Mac (private, zero-latency, runs on Apple's unified memory).
Zero-Marginal-Cost Node Activation: Once inside the backpack/dock, that same Mac becomes an idle compute provider with a single click. MLX doesn't have to subsidize expensive dedicated miners; the hardware is already paid for by the consumer's daily driver.
The Architectural Sweet Spot: Like Apple Minions, MLX doesn't attempt fragile model-splitting over residential internet (EXO's flaw); it routes independent, bursty batch workloads (embeddings, evaluations, synthetic data generation) directly to massive unified memory pools (up to 512GB on Mac Studio), verified by dual-execution consensus.
In one sentence:
EXO is an interconnect science experiment, Minions is locked inside Cupertino, and Bittensor mines validation heuristics—MLX is the only project turning everyday consumer Mac software into an instant, zero-acquisition-cost compute fleet.
