Hey Operators,
Two seismic shifts landed this week that will reshape how you think about AI power dynamics. First, Nvidia — the company selling shovels to the gold rushers — just revealed it's now digging for gold itself, committing $26 billion to build open-weight AI models that will compete directly with OpenAI, Anthropic, and every other lab it currently supplies. Meanwhile, xAI is in free fall: nine of twelve co-founders have walked, Musk admits the company "was not built right," and the coding tools that were supposed to challenge Claude Code never showed up. The AI race is getting messier and more competitive — and this week proved that no position, however dominant, is safe.
Operation Check
Tech stocks under pressure — Nasdaq fell 1.0% and S&P 500 slipped 0.6% Friday as geopolitical tensions in the Middle East and a surprise Adobe guidance miss (stock -7.6%) pushed markets to a third straight weekly loss.
Bitcoin consolidating around $70,800 — BTC is trading just below the $73,400 resistance level, holding relatively steady as crypto traders assess macro risk from U.S. strikes on Iranian targets.
Nvidia's $26B open-weight commitment is the AI infrastructure story of the week moving from chip supplier to model builder signals a fundamental shift in how AI compute economics will evolve.
Operation Dive
Nvidia Bets $26 Billion on Open-Weight AI — and Its Own Customers Are Now in the Crosshairs
Nvidia disclosed in SEC filings — later confirmed to Wired — that it plans to invest $26 billion over five years to build and release open-weight AI models, putting the GPU giant in direct competition with OpenAI, Anthropic, and DeepSeek — three of its biggest customers. The first model out of the gate is Nemotron 3 Super, a 128B-parameter model that Nvidia claims outperforms GPT-OSS across key benchmarks. The strategy mirrors Google's Android playbook: make models freely available, but optimize them to run best on Nvidia hardware. The Insight: The company that built the rails is now running trains on them — this fundamentally changes who holds pricing power in the AI stack. Nvidia Bets $26B on Open-Weight AI Models
xAI Unravels: Musk Admits "Not Built Right" as Ninth Co-Founder Walks
Elon Musk acknowledged this week that xAI "was not built right the first time around" and is being "rebuilt from the foundations up," just six weeks after merging xAI with SpaceX at a $1.25 trillion valuation. Nine of the company's twelve co-founders have now departed — the latest being image generation lead Guodong Zhang — after Musk complained that the company's AI coding tools were failing to compete with Claude Code and OpenAI Codex. Two engineers from Cursor have been brought in to try to reverse course. The Insight: When the founder of a $1.25T company publicly admits it was built wrong, it's not a setback — it's a strategic reset the competition will exploit aggressively. 'Not Built Right the First Time' — Musk's xAI Is Starting Over
Operators in Focus
Anthropic Launches Claude Code Review — AI-Generated Code Now Gets Automated QA Anthropic rolled out Code Review inside Claude Code, a feature designed to automatically review pull requests and catch bugs before they reach production — directly addressing the review bottleneck created by the surge in AI-generated code output. The Insight: AI writing code at scale creates a new class of problem: who reviews the reviewer? Anthropic is betting it should be Claude. Claude Code Review Launch → MarketingProfs / AI Update March 13
Perplexity's Computer Agent Goes Enterprise — With Slack, Security Controls & Compliance At its inaugural Ask 2026 conference, Perplexity announced that its multi-model AI agent Computer is now available to enterprise customers with Slack integration, security controls, and compliance features — a major step beyond its consumer search roots. The Insight: Perplexity is moving up the stack from search to execution — and targeting the same enterprise buyers that Microsoft Copilot and Anthropic's Claude are chasing. Perplexity Enterprise Expansion at Ask 2026 → NeuralBuddies AI Recap
Operator’s Spotlight Read
Amazon Announces Inference Chips Deal With Cerebras
Amazon has struck a deal with Cerebras to deploy inference chips across AWS, strengthening its AI infrastructure strategy and positioning itself against rivals in cloud AI services. The partnership highlights Amazon’s intent to diversify beyond Nvidia GPUs and accelerate inference workloads at scale. The Insight: This isn’t just about one chip deal — it signals a broader reshaping of the AI hardware supply chain, where cloud providers are racing to secure alternative silicon partners to reduce dependency on Nvidia and control costs. Amazon–Cerebras Chip Alliance Shows Cloud Giants’ Push for AI Hardware Independence
Operator Industry Radar
OpenAI Release Notes → OpenAI rolled out major updates including Google/Microsoft app integrations, workspace analytics, and retirement of GPT‑5.1 models, signaling a push toward enterprise‑ready AI. Releasebot / OpenAI Updates
Meta’s $14B AI Gamble → Despite building a $14B “AI super team,” Meta faces delays with its Avocado AI model, raising speculation on whether Mark Zuckerberg might turn to Google’s Gemini for scale. Zuckerberg ask Google for Gemini
Backlash Against Always‑Watching AI Devices → Smart glasses and ambient AI agents are sparking privacy concerns, as regulators and consumers question the trade‑off between convenience and surveillance The Backlash Against AI Devices That Are Always Watching - WSJ
Subscribe to TechWithAdit for sharp, no-noise tech intelligence.
Stay sharp,
— Adit