A lot of chains have shut down recently and will continue to do so. But not Starknet. I joined the Foundation almost 6 months ago, a few reasons but 'the strong tech thesis' was a big factor. My view is that crypto still has SO much left to build but so much of what we see in crypto today feels heavily stagnant from years ago where we were really pushing the frontier. Starknet is great tech, it is actually S-tier for founders/builders with big ideas because it actually expands what can be done onchain. Firstly, Starknet is built to compute. OpenVM’s Ethereum benchmark averages 5.4 seconds per block on eight RTX 5090s and 3.9 seconds on sixteen. Using the quoted rental prices of $4/hour and $8/hour, sustaining those average rates would mean 16,000 Ethereum block proofs a day for $96, or approximately 22,154 for $192. The Starknet benchmark: one block proved in one second on $2.68/hour hardware. Sustain that rate and that’s 86,400 Starknet block proofs a day for $64 in compute. Starknet computes a lot for cheap, so you can do more on Starknet. UX was BIS back in 22 and is still BIS today. Native AA, with passkeys, recovery, session keys and sponsored fees. Baked into the network. Focus Tree showed what that UX can look like: 1m+ users brought onchain through invisible wallets. Dapps that feel like apps. The account model also enables adaptive security. Accounts can be built or upgraded to support post-quantum signatures without all the hoo-ha of a protocol fork. Starknet will be quantum-ready in 2027. Most eyes on Starknet now are for privacy via STRK20, the privacy pool on Starknet. Some neg takes as its not 'secret' but confidential. The main point is privacy on STRK20 is both compatible + composable and builders love it. Expect far more to be built. Put these capabilities together and and the design space ceiling is to the limits of your idea, not the network and the bull case is what builders make possible with it. Also read @Lyskey’s thesis https://t.co/euu4pzTxRT for a more tangible bull-mode: ethereum:0xca14007eff0db1f8135f4c25b34de49ab0d42766.