Co-founder @ampli_inc. Building @mazedotapp, your agentic trading desk. Market analyst and strategist.
Joined November 2019
- Tweets329
- Following525
- Followers6.1K
- Likes17.9K
The uninitiated minds, still grappling with basic economic concepts and primitive crypto game theory, fail to grasp the limitless potential of a global, denationalized money minted by the same matmul that runs AI. They won’t comprehend a 2-for-1 kernel economy with GPUs that can serve inference and mine $PEARL in the same pass, a 2.1B cap with zero premine, post-quantum spending paths baked in, and a potential duplex miner in every data center that comes online on earth and in space, yet at first glance.
Meanwhile, we proof of noisygemm maximalists, disciples of Satoshi’s vision, students of Dwork and Naor’s 1992 paper, and AI-native money enjoyers, focused on the currency agents will hold to pay for their own compute, are here to seize the moment and mint tomorrow’s financial reality.
Grayscale now describes $PEARL the way my May report did: GPUs doing double duty, AI workloads and proof of work at once.
Six months from miner coin to a Grayscale note. The report's still on my profile.
Frontier tech and crypto are converging.
Crypto projects part of the AI buildout:
$DATA → Licensing and verifying AI training data.
$CHIP → Financing GPU hardware through onchain loans.
$PEARL → Merging proof-of-work with AI compute, letting miners earn rewards while powering compute.
Read more on The Stack by @willhamilogden: grayscale.com/the-stack/bloc…
Xavi retweeted
pearl-2:native much?
@VitalikButerin
Listening to @OurielOhayon and @cjhtech discuss AI agents trading prediction markets like degenerates was where you wanted to be today 😂
Heading to Singapore for @token2049.
@Stillm4n and I are in town Oct 5-10. Main days at Marina Bay Sands, side events all week.
If you're building agents or apps that move money onchain, let's grab a bite at the hawker nearby. DM me.
What we're showing at @ampli_inc: MPC-secured orchestration for agentic trading. Builders get the custody and policy layer from one SDK and ship onchain agents without writing that stack themselves.
Pearl is built so matrix multiplication can help secure it.
McKinsey projects that AI, which runs mostly on matrix multiplication, will take about 70% of data-center capacity by 2030.
That means every AI data center built from here adds to the hardware that could secure $PRL, and by 2030 that fleet could draw about 6 to 10x the power of Bitcoin's entire network.
Xavi retweeted
Today @joechalom laid out the required guardrails for agentic finance.
We checked them against the harness Ampli runs today. Most of the list is already live ⬇️
7 months ago I said AI agents would need a money of their own. 2 months later, Pearl launched.
My full thesis on $PRL :
- Agents will hold what they spend, and what they spend is compute
- Pearl is the first money minted by the work AI runs on
- Overhead is the number that decides adoption
- If a fifth of 2030 inference settles in PRL, working balances alone reach 27x today's market cap
Stablecoins bridge the gap today but they are not the real revolution yet.
As we scale AI agents into the trillions, it's improbable they'll rely on centralized fiat currencies tethered to nation-state legacies.
Instead, expect them to natively transact in decentralized digital units of account optimized for machine economies.
Crypto is AI money but stablecoins are human-money onchain.
Joseph Chalom was in the audience in Miami in May when our CEO @Stillm4n gave @ampli_inc's keynote at the Digital Asset Yield Summit.
This week he put numbers on the same thesis: $1T+ in financial services revenue up for grabs by 2030, and $4T a year by 2035.
AI agents are creating a new financial universe.
They will rewire $4 trillion of finance fees by 2035 according to a new estimate from @Sharplink.
The biggest winners will be consumers.
🧵
The line I keep rereading: "Whoever owns the infrastructure, owns the agent, and therefore the customer."
It's why agent control and supervision have been our north star. The capital owner sets the rules. Agents make requests. Independent security layers verify and execute.
The question behind that keynote still guides our roadmap: if agents are superhuman risk managers, who watches the watchmen?
@joechalom says his next piece covers guardrails. That's the layer we've been building at @ampli_inc.
Autonomy with supervision.
I wrote this in May when Pearl pearl-2:native was still just a miner coin before any listing
It went from pennies otc to $1.70 on safetrade → washed out and re-accumulated at $0.25 this summer → up almost 6x now back to ~$1.50
I sent this report to my friends who bought it. So they know what they own… basically the Bitcoin of AI
