Saffercoin @saffercoin

Assume everyone you meet & interact with will act in their own self-interest

PeerDAS in Fusaka is significant because it literally is sharding. Ethereum is coming to consensus on blocks without requiring any single node to see more than a tiny fraction of the data. And this is robust to 51% attacks - it's client-side probabilistic verification, not validator voting. Sharding has been a dream for Ethereum since 2015 , and data availability sampling since 2017 ( github.com/ethereum/resea… ), and now we have it. That said, there are three ways that the sharding in Fusaka is incomplete: * We can process O(c^2) transactions (where c is the per-node compute) on L2s, but not on the ethereum L1. If we want to scaling to benefit the ethereum L1 as well, beyond what we can get by constant-factor upgrades like BAL and ePBS, we need mature ZK-EVMs. * The proposer/builder bottleneck. Today, the builder needs to have the whole data and build the whole block. It would be amazing to have distributed block building. * We don't have a sharded mempool. We still need that. But even still, this is a fundamental step forward in blockchain design. The next two years will give us time to refine the PeerDAS mechanism, carefully increase its scale while we continue to ensure its stability, use it to scale L2s, and then when ZK-EVMs are mature, turn it inwards to scale ethereum L1 gas as well. Big congrats to the Ethereum researchers and core devs who worked hard for years to make this happen.

keyboard_arrow_left Previous keyboard_arrow_right Next

I hear all the complaints: 'Why isn't alt season happening?' The answer isn't the Bitcoin dominance chart only. It's also the ISM Manufacturing PMI. Historically, we don't get an altseason until the ISM prints and holds above 55. It’s not only a crypto event; it’s also global liquidity event. We're getting close to that signal. Once ISM crosses this threshold, the patience for altseason is likely to be over. The minute that number prints, the euphoric, 'everyone makes money' phase of the alt market begins. ================== Register & complete KYC2 for $50! partner.pionex.com/p/0yANcbWQZEy

BIG BREAKING 🚨 16 YEAR OLD BITCOIN WHALE OWEN GUNDEN HAS NOW SOLD ALL OF HIS $1.3 BILLION BITCOIN .HE BOUGHT BITCOIN UNDER $1 .

November 14, 2025 @ 10:20 AM (UTC) Current Price of Bitcoin (BTC-USD): $96,946.90 (BTC-EUR): €83,457.14

GN to those who GN back 😴

CeGwgbokv2dURmjSd2Lg7mUFqWQnHLAzBJvyZB14gbcj

$SFFR

@0xRarest GN

Even the best ideas are ruined by bad execution.

The top 10 Solana products by revenue. Not financial advice.

There are two metrics I focus on more than any other when producing episodes: 1. Time to value. The first lesson has to be sub 3 mins. 2. Word to value ratio. We don’t like fluff. Attention is currency. Protect it. 🚀

AI can now design perfect custom chips for other AIs 9.5x faster. Researchers presenting at the International Conference on Computer-Aided Design developed a framework that automates ASIC chip optimization for LLMs. And they open sourced it on Github. Here's how it works: The system, called Coflex, gives designers a menu of optimal choices based on maximizing accuracy, speed or power consumption. It navigates a search space with over 10¹⁸ hardware and software configurations. Instead of slow, exhaustive testing, it uses Sparse Gaussian Processes (SGP). SGP creates an intelligent, probabilistic "map" of the entire space using only a small set of representative "landmarks" called inducing points. This allows Coflex to accurately predict the performance of untested designs. It then simultaneously optimizes for conflicting goals like minimizing error rate and power consumption (Energy-Delay-Product) to identify the Pareto front of ideal trade-offs. Why this matters: Business Leaders: This dramatically cuts R&D time and cost for creating custom ASICs. It democratizes access to high-performance edge AI, enabling next-gen products in robotics, wearables, and IoT with longer battery life and superior performance. Practitioners: Coflex is open-source. You can implement it now by going to Github. It finds superior solutions for Transformers (LLMs) and CNNs in as few as 2 iterations, compared to 30+ for other methods. Researchers: The key innovation is using SGP to reduce the computational complexity of multi-objective Bayesian optimization from O(n³) to near-linear O(nm²), solving the scalability bottleneck in Hardware-Aware Neural Architecture Search (HW-NAS). This isn't just about speed; it's about achieving better results faster, making custom AI hardware a practical.

I was the first check in a healthcare co 3 years ago that I’d written off for dead Anyways, just got an investor update - on track to end the year at $7m ARR up from $550k last yr You can debate an investment ad nauseum, but founder quality is the only thing matters in startups

This is not a commercial. It's an AI generation. Higgsfield Product-to-Video makes ads so perfect and easy, you'll never trust a real shoot again. Let me explain.

Three truths to remember when making hard decisions: 1. You are never wrong to do the right thing. 2. Often the harder choice is the right choice. 3. Very few things are truly one way door decisions. Very few.