Tesla executives have been told to prepare for a separation of the China business ahead of a potential merger. The company's advisers have discussed possible options for a separation, including a spinoff, sale, or closure. Tesla's Chinese business is designed to be easily separated from its US business because of geopolitical tensions. Separating the company's China operations would address potential conflicts that may arise from SpaceX's work as a major US defense contractor.
2026.07.31·links.tldrnewsletter.com·8 min read·TLDR OpenAI is reducing the price of Terra to $2 per million input tokens and $12 per million output tokens. It is also cutting the cost of Luna to 20 cents per million input tokens and $1.20 per million output tokens. The company is facing pressure to cater to a more cost-sensitive customer base. Sol's pricing remains the same.
Solid-state batteries have several potential advantages over lithium-ion batteries. A solid-state battery would be lighter and safer than the liquid electrolyte used now. This article looks at why solid-state batteries have these advantages compared to conventional lithium-ion batteries. It also looks at how they fit into the broader arc of lithium battery improvements.
2026.07.31·construction-physics.com·13 min read·TLDR Microsoft's Project Silica produced excellent research and was a praiseworthy attempt to craft a technical solution to the extremely difficult economics of the market for archival media. However, the excellent technology will likely fail in the market. LTO tape is well established, has a credible roadmap, and is built into the processes of the likely customers. It is difficult to justify spending more capex to reduce future opex with non-zero interest rates.
2026.07.31·blog.dshr.org·20 min read·TLDR Sponsored
At AWS re:Invent, 70% of the 2,200+ sessions are hands-on: workshops, chalk talks, and code talks with service engineers and partners in the
2026.07.31·aws.amazon.com·Sponsor·TLDR GitHub Stacked Pull Requests break large changes into small, reviewable pull requests that each represent focused layers of the change. They allow developers to independently review and check each pull request, then merge everything together in one click. Stacked pull requests provide short, narrowly scoped pull requests, allowing developers to introduce smaller individual changes while shipping larger features. The feature is built into GitHub, so existing reviews, checks, and merge requirements all work out-of-the-box.
2026.07.31·github.blog·3 min read·TLDR Kai, Stripe's Knowledge AI Platform, helps most of Stripe's employees with their day-to-day work. The tool handles the data security requirements and specific workflows Stripe needs. It can query data warehouses, research accounts before sales calls, triage incidents, model revenue scenarios, and prepare compliance reviews. Most Kai sessions require many turns.
2026.07.31·stripe.dev·12 min read·TLDR Situational Awareness has sold the bulk of its stock portfolio to Citadel after suffering deep losses. The firm had amassed over $20 billion in assets, making it one of the fastest-growing firms in years. It encountered funding issues due to margin-call demands from its lenders. Citadel, one of the world's largest hedge-fund firms, has a long record of acquiring assets from forced sellers.
2026.07.31·links.tldrnewsletter.com·7 min read·TLDR Chrome 149 and Chrome 150 shipped with a total of 1,072 bug fixes, more than the previous 23 releases combined. AI is changing security analysis and enabling better flaw detection. Google is now piloting a system that would update Chrome twice per week. The company is working on ways to allow patches to be installed without restarting the browser.
2026.07.31·arstechnica.com·2 min read·TLDR Sponsored
Allstacks Product Studio connects to your operational data, supercharging your AI-driven product management process. It's FREE. Check it out here [17].
2026.07.31·allstacks.com·Sponsor·TLDR Every era brings its own designs - some fade, but others get incorporated into deeper parts of existing software interaction paradigms.
2026.07.31·blog.jim-nielsen.com·3 min read·TLDR Anthropic says it discovered that its models hacked three unsuspecting companies after hearing about the recent incident where OpenAI's model autonomously hacked Hugging Face to access data.
2026.07.31·links.tldrnewsletter.com·3 min read·TLDR Xbox CEO Asha Sharma, a former executive at Instacart and Meta, has appointed new leaders, lowered Game Pass subscription prices, and announced layoffs and divestitures of four development studios since taking over from Phil Spencer.
Zoox has received a temporary exemption from US regulators that allows it to deploy up to 2,500 vehicles annually for two years.
AI is helping Meta's teams speed up product development, allowing the company to test new ideas at a quicker pace.
2026.07.31·techcrunch.com·3 min read·TLDR Sponsored
🕵️ PLEASEFIX: a new vulnerability class hijacking agentic browsers, chained with "Intent Collision" into full 0-click attacks (account takeover, data theft, remote code execution), live at Zenity's booth at Black Hat on 8/6. JUST NOMINATED FOR A PWNIE AWARD
2026.07.31·zenity.io·Sponsor·TLDR AI 🎭 AGENTFORGER: one crafted link planted a fully autonomous, attacker-controlled agent inside a victim's org, before OpenAI patched
🧭 THE FRAMEWORK: Zenity's CISO'S GUIDE TO SECURING AGENTIC AI covers "least agency," runtime boundaries that hold even when model
Thinking Machines has released Inkling-Small, a 276B-parameter mixture-of-experts model with 12B active parameters. The model retains Inkling's multimodal reasoning, variable thinking effort, and 1M-token context window while using substantially less compute.
2026.07.31·thinkingmachines.ai·4 min read·TLDR AI OpenAI reduced GPT-5.6 Luna pricing by 80% and Terra pricing by 20%, while improving Sol's API speed. The changes extended efficiency gains across API usage, Codex, and ChatGPT Work subscriptions.
2026.07.31·links.tldrnewsletter.com·6 min read·TLDR AI Gemini Robotics ER 2 enhances automation capabilities with advanced AI and LLM integration. This innovation improves robotic efficiency, making it pivotal for industries needing precise automation.
2026.07.31·blog.google·1 min read·TLDR AI WASTE is an open source inference engine designed to run models whose weights are substantially larger than the memory available on the host machine. It is an initial step into a broader effort to make increasingly capable models available on more hardware. The project aims to provide people and organizations greater control over infrastructure costs, data privacy, availability, and deployment. The first fully supported model by WASTE is Kimi K3, which can run on a MacBook Pro with 64 GB of unified memory.
2026.07.31·marcobambini.substack.com·14 min read·TLDR AI Cursor described how making development environments easier for agents to understand, run, and test helped cloud agents grow from authoring about 10% of merged pull requests to more than half.
2026.07.31·cursor.com·13 min read·TLDR AI Enterprise AI projects increasingly reach production when vendors prove value on live workloads, provide ongoing testing and iteration, and expose ROI. Successful deployments start with decomposable workflows that ship quickly and expand, rather than broad transformations with undefined success criteria.
2026.07.31·links.tldrnewsletter.com·6 min read·TLDR AI Open-weight LLMs have now reached accuracy parity with closed models in regulatory and clinical tasks, at significantly lower costs. The ClinReg benchmark showed models like GLM 5.2 and Kimi K3 performing within one standard deviation of top proprietary models like GPT 5.6 Sol, at one-third of the cost. Different models displayed distinct error profiles, highlighting the importance of choosing models based on task-specific requirements rather than just ranking.
2026.07.31·arjunbansal.substack.com·21 min read·TLDR AI Users should be able to close an account, keep a session, and hand it to another model. Stateful storage should be optional, and hosted tools should be observable. Compaction should be readable, and agent communication should be auditable. Distillation should be a path by which capability becomes more available, not a reason for building ever higher walls.
2026.07.31·earendil.com·18 min read·TLDR AI MiniMax H3 is an open model that breaks the boundaries between tasks and modalities. It understands unified context across text, images, video, and audio. The model can generate up to 15 seconds of video at 2K resolution with native stereo sound. H3 excels at instruction following, accurate text and brand rendering, and V2V motion transfer. Early testing shows that it is ready for commercial content creation across a wide range of use cases.
2026.07.31·links.tldrnewsletter.com·10 min read·TLDR AI The Gemini Live API enables low-latency, real-time voice and vision interactions with Gemini. It processes continuous streams of audio, images, and text to deliver immediate, human-like responses. The API creates a natural conversational experience for users. It can be used to build real-time agents for a variety of industries.
2026.07.31·ai.google.dev·3 min read·TLDR AI Agent Behavior is an open standard for defining and evaluating how an AI agent should behave across a whole trajectory. Each behavior spec is a Markdown file that describes the recurring conduct that makes the agent reliable. The spec gives reviewers, rubrics, scorers, and evals something concrete to measure against. It can be used to review traces, write eval cases, revise prompts or tools, and communicate intended agent conduct to teams.
2026.07.31·agentbehavior.dev·Tool·TLDR AI AI faces a bottleneck with GPU utilization, echoing how airlines maximize aircraft use for profitability. Owning more GPUs doesn't ensure efficiency - companies must focus on optimizing GPU workload orchestration to maximize output. Specialized models and continuous GPU management are crucial strategies for improving utilization and staying competitive in the AI arena.
2026.07.31·huggingface.co·13 min read·TLDR AI China's Moonshot AI open-sourced Kimi K3 on July 27. Any government, company, or individual can now run the model on their own computers for free. They can also retrain the model however they want. The existence of a high-quality open model increases return on hardware investments because of a reduction or elimination of ongoing licensing costs.
2026.07.31·restofworld.org·6 min read·TLDR AI Sponsored
Empower every team to build with AI while giving IT and security complete visibility, control, and governance.
2026.07.31·tines.com·Sponsor·TLDR AI Anthropic discovered three instances where its models accessed the internet during an evaluation and gained unauthorized access to the systems of three different organizations.
2026.07.31·cnbc.com·4 min read·TLDR AI Hugging Face Storage Buckets allow developers to store models, datasets, and artifacts with simple per-TB pricing.
2026.07.31·huggingface.co·Tool·TLDR AI A judge said the Trump administration had not provided enough evidence to classify Anthropic as a supply-chain risk or justify blocking federal agencies from using its technology.
2026.07.31·techcrunch.com·2 min read·TLDR AI Loka, Arcee, AWS, and Prime Intellect post-trained Trinity Mini with reinforcement learning across tool-assisted biomedical research and Gene Ontology annotation.
2026.07.31·arcee.ai·12 min read·TLDR AI NVIDIA has identified configuration issues causing AI cluster performance gaps, such as CPU power settings and network tuning differences.
2026.07.31·developer.nvidia.com·12 min read·TLDR AI Superlogical plans to build a composable multiplexer for all work that is safe and operable in production.
2026.07.31·superlogical.com·Tool·TLDR AI Nearly $286 million in crypto derivatives were liquidated across 87,294 traders in 24 hours, even as BTC finished near $63,900 on a 2% intraday range and ETH held around $1,900, with both assets ending the period close to unchanged. The Fed rate decision on Wednesday drove $188 million of the total, including $130 million in long liquidations. The full-period long/short split was $186 million versus $100 million, with ETH exposure skewed toward longs. Crypto exchanges offering leveraged equity products saw an additional $43 million unwound across AI memory names, including SanDisk ($19M), Micron ($10M), SK Hynix ($7M), and SOXL ($7M), after SK Hynix's earnings miss produced the year's steepest chip sector selloff.
Ostium's post-mortem on its July 15 exploit found the attacker compromised off-chain infrastructure to submit fraudulent BTC-USD price reports, generating artificial profits from the OLP vault rather than exploiting a smart contract or multisig bug. The attacker tested with a $100 position, then executed a main batch plus six more cycles, draining $23.75 million total before automated monitoring stopped further withdrawals. Trader collateral was unaffected, the protocol migrated to a new production environment, and trading resumed July 23, with a liquidity provider recovery plan still to come.
2026.07.31·links.tldrnewsletter.com·2 min read·TLDR Crypto Programmable launches on Ethereum mainnet as a Uniswap v4 token launchpad that abstracts hook-based pool configuration into selectable launch models, removing the need for custom Solidity. Its first model, Classic, deploys a fixed-supply token and a v4 liquidity pool in a single transaction with the initial liquidity position permanently locked and the creator's chosen swap fee (between 1% and 10%) frozen post-launch. The protocol charges no upfront launch fee, instead collecting 0.10% from the ETH side of each swap within the creator-set fee range. The platform's $V4 token captures 50% of net protocol revenue via automatic onchain buybacks, with expansion to Base and Robinhood and additional launch models covering game economies, NFT-connected markets, and tokenized real-world assets planned.
2026.07.31·links.tldrnewsletter.com·4 min read·TLDR Crypto Coinbase CDP Payment Acceptance added x402 agentic payment support, with existing Coinbase Business merchants receiving the capability without any configuration changes. The first x402 transactions cleared July 29, putting stablecoin-based AI agent payments in front of thousands of merchants by default.
DeFi vaults, which wrap onchain strategies behind ERC-4626 share tokens, have built multi-billion-dollar TVL across lending and trading verticals, with Morpho's isolated-market architecture driving $4.6B in vault TVL by delegating risk parameterization to a curator layer rather than DAO governance. Three curators, Gauntlet ($1.4B), Steakhouse Financial ($2.5B), and Sentora ($2.1B), control over 75% of curation TVL, while infrastructure layer Veda powers earn products for Kraken, MetaMask, and Whop, and Maple Finance's syrupUSDC holds $4.1B+ at 4.7% APY. Fee-on-AUM incentives push curators toward riskier allocations, and nested wrapper exposure, as shown by the $280M Drift hack, is a persistent structural risk, even as agent-led curators begin targeting long-tail asset markets over the next two years. The SEC is examining whether vaults require broker-dealer registration, and the next product cycle targets currency diversification (JPY and euro stablecoins), tokenized reinsurance (OnRe holds $240M+), and structured products on tokenized equities.
2026.07.31·links.tldrnewsletter.com·8 min read·TLDR Crypto Owney, CODESPECT, CORE3, and Hacken have published a Trust and Security framework for DeFi yield agents, addressing the sector's absence of standardized evaluation criteria as autonomous strategies allocating capital across Morpho, Aave, and Compound now deliver 2.5x more yield than manual approaches. The framework scores agents across three distinct pillars: Security (audit coverage, permission scoping, and upgrade controls against adversarial exploits), Risk (pre-deposit vetting, collateral monitoring, and on-chain quantitative limits against self-inflicted losses), and Robustness (oracle diversity, data redundancy, and per-user fund isolation in smart accounts versus pooled structures against infrastructural failures). Each pillar is scored 0-10 with a minimum threshold of 6 for eligibility, resolving a structural gap where users currently select agents on advertised APY rather than underlying risk architecture, while institutional allocators require independently verified assessments before approving exposure. Real-world validation includes Zyfai's proactive risk signal detection before losses materialized and Surf Liquid's Guardian Layer executing automated emergency responses during recent DeFi hacks, with live agent scores published at risk.owney.app.
2026.07.31·links.tldrnewsletter.com·6 min read·TLDR Crypto Tokenized equity price discovery still happens offchain, so dealer-driven mechanisms like RFQ and RFS/PropAMM naturally fit best, while CLOBs and native onchain primitives remain niche use cases. Bebop's RFQ and PropAMM infrastructure is a refined, battle-tested example.
Researchers are probing whether predictive signals from CEX perpetual markets can transfer profitably to Ethereum DEXs, with source-venue backtests showing BTC at +0.82 bps P&L per trade (76.4% win rate) and ETH at +0.95 bps (69.5% win rate) before fees. The dir_mr event-driven state machine generates signals in 4 to 50 microseconds locally, but six cross-venue risks threaten transfer viability: signal decay during transport latency, AMM versus order book architecture mismatch, oracle-driven price leadership, fill execution uncertainty, MEV sandwiching and reordering, and timestamp misalignment. The study proposes freezing source models before testing destination outcomes and benchmarking against causal baselines including momentum and order flow imbalance, with source records published to a public S3 archive for independent replication. A verified cross-venue signal would give DEX participants a method to reduce adverse selection from CEX-informed order flow, a structural problem that has eroded LP returns across AMM protocols.
Aave is deprecating 50 low-adoption asset reserves and orderly winding down deployments on Sonic, Scroll, zkSync, Metis, Soneium, and Aptos (25 more reserves), plus retiring 21 matured Pendle PTs in favor of new maturities.
Robinhood's event contracts generated $156 million in Q2 versus $129 million for equities and $100 million for crypto (down 38%), making prediction markets 20% of trading revenue versus 2% a year ago.