The most bullish thing you’ll observe in markets maybe in our lifetime is people posting one fractal but ignoring the expansive fractal that led to it and the fact that we never saw that same move in this market. Mind boggling really
There will be no Q4 low #BTC
STRC down to $82.6 today. Here's my read:
1. Strategy is fine. If everything stays as is, they can pay STRC dividends for 32 years. If BTC appreciates at ~2% CAGR, they can pay dividends indefinitely.
2. Why the sell-off? This appears to be a liquidation cascade.
Over the last 6 months, the narrative became that STRC volatility was reducing, and price began to spend all its time in $99-100 range.
This invites leverage. If you expect the price to always be north of $95, you can take on 20x leverage with your portfolio to buy more STRC and dramatically increase the yield on your portfolio.
This works great, until it doesn't.
STRC is designed as a free-market asset. When attention shifted to SATA and STRC price flagged, it may have raised the attention of opportunistic short-selling hedge funds.
By shorting aggressively, they could push the price down and start triggering margin calls and liquidations from folks who aggressively levered up their STRC positions.
The price action today is a clear liquidation cascade, rapidly pushing prices lower, in turn triggering additional liquidations.
3.
What happens now? The market will heal itself.
Opportunistic hedge funds will recognize that this is a firesale and the fundamentals are unchanged for STRC and step in as buyers. Shorts will close, becoming buyers. Individuals are getting a tremendous entry price for long-term holding STRC shares.
Buyers at this level will get ~13.7% effective yield. If STRC trades back to $100 and they sell, they get an easy +18% return.
4.
What will Strategy do?
Strategy will likely increase the dividend rate on June 30 - maybe to 11.75% but possibly to 12%. Buyers at the current price level then would get 14.2% effective yield from that point forward.
Strategy may also step in to buy STRC shares back. They could do this by issuing new shares of MSTR (currently at 1.14 mNAV) or by taking on traditional debt and deploying those funds to buy discounted STRC shares on the market.
If/when STRC trades back to $100, Strategy could then re-issue those STRC shares. The ~$15 delta per share could be used to buy BTC as pure accretion to MSTR holders, with no net change to amplification.
No doubt that Saylor has already at least considered this, and it wouldn't surprise me if they're currently doing this.
5.
In summary...
The market is freaked out that this depeg is like Terra/Luna... but this is not an asset like that. Strategy's balance sheet determines whether STRC continues to receive dividend payments... and Strategy's balance sheet is completely unchanged.
This is a leverage wipeout.
From this, the market will learn that Digital Credit is mostly very low volatility. But because it is a free market asset, the longer that a Digital Credit instrument trades within a tight range to par... the more leverage will inevitably pile up as people get greedy.
And that creates the conditions for a leverage wipeout depeg. Following that, the instrument will make its way back to par value as the market heals itself and recognizes that the dividend payments will continue uninterrupted because the issuer's balance sheet is unaffected.
Very proud to be the first ever Subnet with an official government partnership to grow commercially in its region and beyond 🏴
Lots of work to be done! Onwards! 🧬
A NIOME future is now. Join us!
https://t.co/nL8J33gATQ
Bear flags are bullish just like the headline conflict last month was supposed to be bearish for risk assets
Coiling for the biggest move of the cycle shouldn’t look obvious #BTC
Are you sending the same system prompt on every API call? You're already paying half price on the tokens that repeat.
Input caching is on by default across the Chutes catalog. When the same prompt prefix repeats, you pay 50% of base input on those tokens. No flag, no setup, no code change.
Each GPU instance holds a local prefix cache in VRAM, encrypted on TEE models. The Chutes router hashes your prompt and sends you back to the instance that already has it, so the model reuses what it computed before.
Quick math on DeepSeek V3.1-TEE at $0.27 per million input tokens. A 4,000-token system prompt sent across 10,000 requests a day is 40M repeated tokens. Uncached: $10.80 a day. Cached: $5.40. That's $162 a month off one prompt, on a model already 91% cheaper than Claude Sonnet 4.6.
Want to confirm a cache hit? Check usage.prompt_token_details.cached_tokens in your response. The number tells you how many tokens got the discount.
Pull that count off your next response. Reply with what you see.
Once $tao does this it wont only make headlines of pioneering a new political system which will be widely used within a decade but it will also plug most of the leakage it has now
A novel design like this could show the power of blockchain and governance, its necessary
It doesn't matter what you say about TAO, there's always a queue of people in the comments ready to FUD it. Good sign in my opinion, just another way I know I am in the right project. Keep it coming.
Bitcoin is holding up really nicely here, but still trading in a channel, so no confirmation is available either way. At the upper channel boundary of a bear flag, we would expect Bitcoin to quickly sell off, but this hasn't been the case. As long as Bitcoin continues to range in the top half of the channel, above circa 74k, I am inclined to remain bullish, and believe Bitcoin will continue higher. Equities are showing you the bull market has not ended, the ISM is showing you the economy is in expansion, and this is where I believe Bitcoin and large cap altcoins more broadly will stage a recovery that the majority is not expecting.
podcast appearances usually aren't worth a tweet. this one is.
@jon_durbin is heading to @twistartups next week.
bittensor, decentralized inference at 100B+ tokens a day, and a small tease on what's next for training.
episode drops first week of may. what should we cover?
$TAO
there’s one team on earth turning vision foundation models into deployable specialists at industrial scale.
it isn’t OpenAI.
it isn’t Meta.
it’s us.
and it’s happening on sn44 on bittensor $TAO
Progress! < $10/hr to pre-train a 20b MoE. Not too bad.
8 VMs, each with a single l40s GPU, spread across Poland, France, and the USA, with a few extra 4090s and a 5090, training a 20B MoE @ 6s/step (I also had it fully functional with only 4090s, just slower)
Should have results soon, then we can try this out on a much, much larger model!
Underwriting has been a broken process for as long as property loans have existed.
That ends now.
RESI has been accepted to the @Plaid partner program, bringing KYC and income verification directly into the platform.
The full-stack real estate hub is loading...
Auto Top-Up is now in billing.
Head to: https://t.co/tyh8p0c6TO
Pick a floor and a refill amount, save a card, and your balance refills automatically.
People asked for it and we delivered.
https://t.co/TrASBRm9cF
I'm positively surprised with the resilience of $TAO.
After the debacle of the subnet, you'd assume the price to continue to fall further.
As a matter of fact, it stalls and it's actually looking to be breaking back upwards.
Strong signal for the entire ecosystem, and I think it's reasonable to expect that it just continues to rally further as the AI <> #Crypto convergence becomes stronger.