Palihapitiya: BitTensor subnet Ridges AI hit 80% of Claude 4 in 45 days with $1M rewards
BitTensor's subnet 62 (Ridges AI) achieved 80% of Claude 4 coding capability in 45 days using $1M in token rewards, demonstrating crypto-incentivized open source training can comp…
Gerstner: Anthropic's $30B revenue run rate is fastest ramp in tech history
Anthropic hit $30B ARR by March 2025, adding $11B in a single month (Databricks + Palantir combined), driven by Claude Code launch and enterprise demand for labor augmentation; co…
Gerstner: Meta has massive compute advantage, models 'incredibly performant'
Meta's fortress balance sheet and massive compute infrastructure give it structural advantage; Llama models are 'incredibly performant' on inference speed/quality, making Meta a s…
Gerstner: Buying OpenAI shares despite vibes, Spud model on par with Mythos
OpenAI is firmly on the intelligence wave with Spud (Blackwell-trained) model matching Anthropic's Mythos; Codeex ramping fastest in history; TAM for intelligence is dramatically…
PolyMarket: 95% chance SpaceX IPOs before 2027, fortress balance sheet by June
SpaceX expected to achieve fortress balance sheet by June 2025; PolyMarket shows 95% probability of IPO before 2027, making it a key 'out of harm's way' AI bet for venture capital.
Sacks: Nvidia was first multi-trillion AI winner at chip layer
Nvidia captured value at chip layer first, becoming first multi-trillion dollar AI company; value capture now shifting to hyperscalers and model layer (Anthropic, OpenAI), with ap…
Sacks: Palantir is application layer company turbocharged by AI models
Palantir demonstrates value capture at application layer, turbocharged by frontier model capabilities; question remains whether other application companies will benefit or get eat…