The latest on TECHi.

Supernatural plans a fall 2026 Quest relaunch with returning coaches, a rebuilt library and higher prices, but music licensing limits what can return.

U.S. sanctions target Shelbit and Aban Tether over separate Iran-linked crypto risks after Dubai VARA’s earlier action against Shelbit General Trading.

A noRecognition pattern reportedly fooled one surveillance camera, but its public benchmarks remain simulated and the field-test evidence is not yet available.

Anthropic is making Claude Code auto mode the default August 14 after tests showed it beat manual review while still missing some dangerous commands.

X is replacing creator revenue sharing with Original Content Rewards, excluding reposts while keeping Premium-user impressions at the center of payouts.

OpenAI says Astra may reach its Critical cyber threshold, but key test results remain private. Here is what the disclosure proves—and what it does not.

Firebird opened its Armenia AI factory, but the 70,000-GPU, 300 MW vision still depends on power, financing and customers. Here is what is actually live today.

Meta AI can now use email and calendar context for recurring tasks. Here is what to test before granting primary-account access and trusting its research.

Kalanick's Atoms raised $1.7B led by a16z, with no valuation disclosed — a bet that physical-AI profit belongs to whoever owns the kitchens, mines and trucks.

Microsoft's expanded multibillion-dollar Mistral pact reverses their 2024 roles — Microsoft now buys European GPU capacity and ships Mistral's models to regulated firms.

NAVER, NVIDIA and Brookfield are tripling Korea's AI factory to 200MW. The real story isn't the megawatts — it's the ~$10 billion capital stack behind sovereign AI.

Alphabet posted record Q2 2026 net income and its first negative free cash flow together. Why $44.9B of AI capex — not the profit — is the real story.

Black Forest Labs' FLUX 3 puts image, video, audio and robot-action prediction in one model — and holds the open weights back. Here's the market read.

The FTC's proposed statement calls secretly steering AI outputs deception under Section 5, even to obey a state law like Colorado's. Comments close July 31.

Upper90 will lend General Compute up to $400 million secured by SambaNova inference chips, not Nvidia GPUs. Why inference silicon as loan collateral is an untested bet.

Fireworks AI raised $1.505B at a $17.5B valuation. The number that matters is 95%: nearly all its tokens come from customer-specialized models, not raw open weights.

Akash set a $5M compute-spend record and routes 1.7B AI tokens a day, yet AKT sits ~94% below its high. Inside crypto's AI-compute value-capture problem.

Tesla fell 14.5% to an 11-month low after Q2 margins hit 1.4%. Post-earnings analyst price targets now split from $130 to $600 — here's what it means for TSLA.

Google's Ironwood TPU has been generally available since April, yet there's still no public per-hour price. What that silence reveals about AI inference costs.

Bloomberg says Z.ai switched on a ~1GW data center running only Chinese chips to train GLM. What it means for export controls, open models, and Nvidia's China thesis.

France's competition authority found OpenAI, Google, and Anthropic hold 84% of the AI-agent market and warned of lock-in. What Opinion 26-A-05 means for you.

DeepSeek retires the deepseek-chat and deepseek-reasoner model names at 15:59 UTC on July 24. Migrate to deepseek-v4-flash and mind the thinking-mode cost catch.

Tesla's Q2 2026 revenue rose 26% to a record $28.24B, but operating income fell 57% and margin sank to 1.4% as robotaxi and AI capex hit $5.79B. What it means for TSLA.

Moonshot AI's Kimi K3, the largest open-weight model ever, opens July 27. It's cheap per token but needs ~1.4TB resident — here's who can really run it.

Grayscale's amended TAO filing bars the trust from staking. Under dTAO, emissions follow staked TAO into subnet pools — so a spot wrapper sits outside the loop.

AMD and Cerebras will split AI inference across Helios racks and the Wafer-Scale Engine. The 5x per-watt claim is modelled against Cerebras-only hardware.

Coinbase Business can now accept USDC from AI agents over x402. The on-chain record shows a crypto-native payer cohort, not commercial demand.

OpenAI, Meta, NVIDIA/ServiceNow, and Google shipped enterprise agent platforms in a month — and disagree on what the product is. The market-structure read for buyers.

Google shipped Gemini 3.6 Flash, 3.5 Flash-Lite, and a restricted 3.5 Flash Cyber model. Why the gated cybersecurity variant is the real signal in the release.

The digital omnibus pushed high-risk AI deadlines to 2027-2028 — but GPAI and Article 50 duties still hit August 2, 2026, and the omnibus is not yet law. The map.

OpenAI's Presence platform productizes agent governance — policies, guardrails, simulations. What its FDE-led model and 75% claim mean before you sign a pilot.

Amazon's custom silicon runs at a $20B pace with $225B in Trainium commitments. What sold-out chips through 2027 mean for anyone buying AI training capacity.

Palantir stock fell about 6% on July 22 as NHS scrutiny, an open-source tool, and valuation nerves converged. Why the real test is the August 3 earnings print.

From August 2, 2026 the EU's AI Office can audit GPAI models, demand fixes and fine up to 3% of turnover. Who is exposed — and what to verify now.

CoreWeave measured 10× more DeepSeek-R1 tokens per megawatt on Vera Rubin NVL72, but missing methodology limits what NVDA investors can conclude today.

LangChain now standardizes reasoning effort across OpenAI, xAI and Fireworks, but identical labels still hide different cost, latency and model behavior.