Samad Javadi

Co-founder & COO @AskTillyAI. Early contributor @Tensorcash. Building

[Prof. Gonzales @xProfGonzales]

The founder of DRW is talking about AI compute becoming bigger than oil, with entire financial markets emerging around it.

Meanwhile, @Tensorcash is building infrastructure for exactly this at a sub-$20M valuation.

I’ve been covering TensorCash for a few weeks now, and I still feel like most people have no idea what they’re overlooking.

Let me explain it as simply as possible ⤵️

1️⃣ WHAT IS TENSORCASH?

Traditional Proof of Work uses computing power to secure a blockchain.

TensorCash uses that computing power to run REAL AI models instead.

One job, two purposes:

1. Secure the blockchain through verifiable AI computation

2. Generate AI inference that can be sold to paying customers

2️⃣ WHERE DOES THE MONEY COME FROM?

• @AskTillyAI brings AI services to users and businesses through chat, agents and APIs.

• InferenceFi is building the marketplace where verified AI inference can be bought and traded, with miners paid in $TSC

• TensorCash already has native financial infrastructure for settling transactions and compute-linked contracts.

3️⃣ NOW LOOK AT THE OPPORTUNITY

The managed AI inference market is projected to exceed $100B annually by 2030.

• 0.01% market share = $10M/year

• 0.1% market share = $100M/year

That’s potential commercial volume, with $TSC positioned as part of the payment and settlement infrastructure.

Now compare the scale of that opportunity with a valuation below $20M

I genuinely think the disconnect is insane

People are still treating $TSC like another AI narrative, while completely overlooking what they’re actually building. Wake up!

[Enrico Ottavian @enrico_otta]

LLM tokens will inevitably trade like any commodity - open price, verifiable quality.

But three problems stand in the way:
→ you can’t check what you get
→ you can’t store it
→ the market prices the rig, not the oil
Full piece on @AskTillyAI this Friday.

[Samad Javadi @samjavadi]

@100XInsight Looks like a franchise, looking forward to multiple sequels

[TensorCash @Tensorcash]

TensorCash is a proof-of-work chain where the work is AI inference. Miners can sell this inference and get a chance at the next block with the same work.

It sets the standard for verifiable inference by attaching a proof of the model used to every answer.

[TensorCash @Tensorcash]

@AskTillyAI, founded by @enrico_otta and @samjavadi, offers AI chat and agents with a built-in $TSC wallet and an API for developers.

Users pay in regular currency, miners are paid in $TSC, and every answer can be checked with a Verify receipt button.

[Samad Javadi @samjavadi]

Not in and of itself. Matrix multiplication is a mathematical function, it’s valuable when it runs as part of a useful process like AI inference. It can be a part of inference, but it doesn't have to be.

So the catch is the incentive structure needs to reward the full useful work, not a subset that can be optimized and run standalone... The only way to mine should be to produce something useful.

Otherwise you're basically going back to SHA hashing.

[Samad Javadi @samjavadi]

Yes...

Mining should produce something people actually use.

On TensorCash that’s not optional. The protocol requires miners to run real AI models, and the whole answer counts as the work, from the first token generated to the last.

So every block comes from real end-to-end inference, not just a fraction of the inference process. Miners can sell that inference and also have a chance at the block reward. @AskTillyAI just had one build an animation for a client:

https://t.co/cYmUDjh6Hx

[AskTilly AI @AskTillyAI]

An autonomous coding agent ran for 28 minutes on a @Tensorcash miner via the AskTilly API. One prompt, and it built this animation on its own for the client.

We sold the session for $0.16 and the miner was paid in tensorcash:native. A day after the model went live, already at market rates.

[TensorCash @Tensorcash]

TensorCash 1.2.3 activates today at block 31,400, expected around 13:00 to 15:00 UTC.

Verifiers: upgrade to 1.2.3a-r1. Full wallets: upgrade from the release page.

After activation, a registration for Qwen3.8-27B (bf16) is planned, with a difficulty multiplier of 50,000.

[TensorCash @Tensorcash]

An early contributor of the protocol signed his first post with a qualified electronic signature, and the proof is stored on TensorCash.

The hash of the text, the certificate fingerprint and the signature are in block 31,144. This shows that the exact text was signed and existed at that block, and it cannot be changed or backdated. The certificate ties the signing key to his legal identity.

Native assets on TensorCash offer additional rules that every node enforces. Among others, the signed legal document belongs to the asset, no more units can exist than it allows, and it can only change with the holders’ approval.

[Enrico Ottavian @enrico_otta]

Amazing read, model & context diversity is what we use at @AskTillyAI to reliably avoid Claude’s “load bearing” rabbit holes… but we are the humans in the loop. @hardmaru is taking this to the next level.

[TensorCash @Tensorcash]

TensorCash v1.2.3 is out.

Multi-model verification, cascading verifier failover, VDF checkpoints and tougher RED block penalties.

⚠️ Upgrade nodes, miners and verifiers before mainnet block 31,400 (~Thu 15 Oct, 12:00 UTC).

Qwen3.8-27B mainnet registration follows.

[Prof. Gonzales @xProfGonzales]

https://t.co/NcRbe89Joe

[ykz 🦬 @ykzthe1]

For those following $NOCK and $PRL, I think $TSC deserves a place in the same conversation

@Tensorcash is approaching proof-of-work differently, and the technology behind it is what caught my attention

Instead of relying on traditional hashing, $TSC uses verifiable AI inference as part of its consensus mechanism

Here’s the concept:
→ Miners perform AI inference computations.
→ Independent verifiers reproduce the results.
→ Matching results help validate blocks.
→ Incorrect results lead to rejection.

The goal is to make the computational work securing a blockchain useful beyond the blockchain itself

And the architecture goes further:
• 21.18M max supply, with no premine or token sale
• 40% block reward reductions each epoch
• Native financial infrastructure supporting assets, settlement, and derivatives
• Post-quantum security incorporated into the protocol

What makes $TSC interesting to me is the combination of Bitcoin-inspired scarcity, verifiable AI computation, and financial infrastructure built directly into the network

NFA - DYOR

[Samad Javadi @samjavadi]

Verified inference at this scale needs a buyer. We're building it at @AskTillyAI: the market where TensorCash capacity gets bought, paid for, and settled. Tokens per second in, revenue per second out.

[TensorCash @Tensorcash]

The TensorCash network produced an average of about 98.29 million tokens of Qwen3-8B output per second over the last 24 hours, roughly 246,000 pages or 820 novels every second.

This is inference capacity that can be sold to users to answer their questions.

With the upcoming Proof v5 aiming to lower latency, the same capacity can reply ever faster, and its value grows with it.

[Samad Javadi @samjavadi]

@quxiaoyin The other key issue here is that in a race-to-the bottom context on routers, providers competing to serve the same model have a strong incentive to cheat through quantization or model substitution.
This is why a true market for AI compute needs strong verification @Tensorcash

[TensorCash @Tensorcash]

On TensorCash, securing the chain means running an LLM.

Every 256-token window gives a miner one attempt at the next block. With the chain setting the randomness, every block proves which model did the work.

With millions of attempts per block and only the winner getting checked, nobody knows in advance which one that will be, so every miner has to run the model they promised.

Still, every answer carries a record of each word drawn, enabling everyone to verify the model.

Inference that can be sold, verified and secures the chain at once.

[Samad Javadi @samjavadi]

We are building precisely this and the market will run on trustless verification.

[Caco @Caco12]

Probably not obvious to most yet but I think compute is the next commodity Wall Street learns to trade.

If you just look you can see gpus already used as loan collateral and rental prices already have indexes.

Just to explain better oil also became a market when a barrel could be measured.

That is why I think compute becomes one when work can be proven.
In my opinion two layers are needed:

Compute you can trust bit for bit, same results on any hardware .

Inference you can prove came from the model you paid for ofc with a market to buy it.

...two chains building exactly that
I am heavily in both...

[Caco @Caco12]

In 2008 Satoshi gave the world a new form of money built on cryptographic proof instead of trust.

The problem was never the banks alone. It was having to trust a third party you could not check.

Today you also buy AI inference on trust. You trust the provider ran the model you paid for at the precision you paid for or you just hope they didnt swap it for something cheaper. You cannot check unless...

Now lets imagine we could check every inference carrying proof that the right model actually ran. We are not talking about trust we are talking about true proof.

This is not some crazy idea of mine. It exists and it is already running on chain. This is a multi cycle bet and it probably gives crypto for the first time a useful role in a global narrative that is not about money.

So back to topic. Same problem, new market. Proof instead of trust, but for COMPUTE.

Last time retail had more than a decade to front run Wall Street. This looks like the same playbook. Only this time it feels like Wall Street will front run retail.

If you don't believe me or don't get it, I don't have time to try to convince you, sorry.

[Enrico Ottavian @enrico_otta]

Future of AI is coming, and it belongs to everyone.

[Samad Javadi @samjavadi]

@ttunguz As inference scales, verifying that the promised model actually ran could become as important as delivering the output. We think decentralised, verifiable inference will be a key part of that infrastructure. @TensorCash @AskTillyai