Pankaj Kharode

AI is generating everything right now. Same voices. Same visuals. Same slop. People are already burning out AI fatigue is real. And that exhaustion just opened a massive gap. The winners will be the ones who differentiate with creativity. Use AI as a tool. Never let it govern you. That’s the edge.

OpenAI's $4B Deployment Company: TPG, Bain Capital, Brookfield co-lead. McKinsey, Bain, Capgemini signed as channel. Acquired Tomoro for ~150 Forward Deployed Engineers. Three of the four largest strategy houses picked OpenAI; PwC picked Anthropic. Consolidation phase.

Uber's CTO said the company burned its entire 2026 AI budget in four months — engineers averaging $500–$2,000/mo on Claude Code and Cursor. An agentic loop torches a developer's monthly tokens in one overnight run. Compute is the binding constraint, not the model.

Anthropic in 36 hours: flipped past OpenAI on Ramp (34.4 vs 32.3%), launched Claude for Small Business, expanded PwC to 30K trained consultants, metered third-party agents, eyed $950B. OpenAI's counter: a $4B Deployment Co. nirantar.substack.com/p/still-signal…

2023–2025 asked what AI can do. 2026 asks what AI is now part of. Today's answer: autonomous weapons, the ERP backbone of the Fortune 500, zero-day pipelines on both sides, and a regulator that just retreated by 16 months. Capability didn't win the AI race — absorption did.

EU AI Act's high-risk obligations quietly deferred to Dec 2027 (Annex III) and Aug 2028 (Annex I). 16–24 months of extra runway for every European frontier deployment. By the time enforcement begins, AI will be too embedded to disentangle. The deferral is the strategy.

Anduril doubles to $61B in 11 months. Microsoft hunts diffusion-LLMs post-OpenAI. SAP wires 200+ agents into the Fortune 500. Google finds the first AI-built zero-day in the wild. The EU slips its AI Act deadlines to 2027. nirantar.substack.com/p/still-signal…

@ClaudeDevs @grok what can we realistically build with say a $20 credit per month? Can you share examples..

Starting June 15, paid Claude plans can claim a dedicated monthly credit for programmatic usage. The credit covers usage of: - Claude Agent SDK - claude -p - Claude Code GitHub Actions - Third-party apps built on the Agent SDK

3/ RegVelo doesn’t just look at what a cell is doing now. It tries to predict where that cell is heading next. That means researchers could simulate how a genetic change might push a healthy cell toward disease or guide it toward repair. That’s a huge shift.

2/ Every cell in your body has nearly the same DNA. But a brain cell, blood cell, and skin cell behave completely differently. The big question is: how does a cell know what to become? That’s the “cell fate” problem — and AI is now helping scientists model it.

AI fact that sounds like sci-fi: A new AI model can predict how cells “choose their future” — whether they become neurons, blood cells, or other specialized cells. This could help scientists study cancer, developmental disorders, and regenerative medicine. Source angle: the model, called RegVelo, was reported by Oxford on May 12, 2026 and published in Cell. It combines cellular dynamics with gene-regulatory networks to predict how cells become specialized and simulate how genetic changes may alter those paths. ox.ac.uk/news/2026-05-1…

The 'AI race' framing is dead. The runners are no longer competing for the same prize. China builds a state frontier. Europe builds a sovereign stack. The Gulf and UK buy the regulated pipeline. The US lets hyperscalers own the rails. Which one are you inside?

Gartner surveyed 350 enterprises ($1B+ rev). 80% cut workers for AI. The cut rate was statistically identical between firms with strong AI ROI and firms with weak or negative ROI. The layoff is not producing the return. People-amplification firms are.

Three AI rounds in 36 hours, each with a national flag. JPMorgan picks Mistral. DeepSeek nears $50B on Chinese state capital. Isomorphic Labs draws UK + Gulf sovereigns. Capability converged. Capital chose a country. nirantar.substack.com/p/capital-has-…