#AI4S# is back in focus. $MRNA's historic Phase 3 cancer vaccine win reignited interest in how AI and data could reshape biotech. In Hong Kong, #DiagensTech# (HKEX: 2526) is gaining attention. H1 revenue rose 21% YoY, while model-service revenue jumped 101%, now accounting for 87% of total revenue. Other AI4S names in focus include #InsilicoMedicine# (HKEX: 3696) and #XtalPi# (HKEX: 2228). All are tradable on StableStock. → Not investment advice. For informational purposes only.

Posted "Rethinking Failure Studies in the Age of AI" Personally, whether it's AI4Science or AI solving open problems in mathematics, it doesn't matter to me whether a breakthrough comes from humans or AI—the value of truth doesn't change. At the same time, though, I believe the tacit knowledge and hands-on experience gained through the actual process of research and execution belong to an entirely different dimension—one that humanity still needs to absorb and learn from.

Introducing Hyra-1.0, the first version of Hunyuan Research Agent. 💡💡💡 Built to recursively improve solutions for performance-driven research and engineering tasks. Explore our demos in AI4AI, AI4Science, and AI4Fun:

Thoughts from my Shanghai field trip Sentiment among Greater China investors was far worse than I expected. Unlike the US market, A shares had not seen a single rebound since July. Still holding Chinese and Hong Kong AI stocks that had crashed and Korean memory stocks that had collapsed, investors kept pressing me on when memory would bounce back. Many investors said the AI trade was over, and a few said they doubted whether even 2027 capex, not just 2028, would go ahead as planned. Overall, the bearish sentiment was so strong that even I, a bull before coming to China, started to waver. It got to the point where, faced with the bear case that memory prices could start falling in the third quarter of next year, I took it seriously and couldn't refute it. As the conversations went on, I gradually began to understand where their pessimism was coming from. In the end, they were all asking the same question. So what comes after AI coding? What will be the next engine of ARR for Anthropic and OpenAI? Several candidates came up. AI4S, AI chip design, and so on... But none of them clearly convinced investors. I couldn't offer a clear answer either. Sure, there are rumors that Anthropic has made major breakthroughs in both AI4S and AI chip design, but what we actually want to know is when any of that turns into money. With AI4S, there could be beneficiaries right away, such as companies that sell lab monkeys or diagnostic kits. In fact, there already are. But while we can trade those names in the short term, what we ultimately want to know is whether Anthropic can make money by developing drugs. That is still several years away. As for AI chip design, the TAM itself is hard to calculate, and even if you do calculate it, it is hard to monetize. That makes it difficult for the market to believe in. Honestly, this feels like the hardest market in history. If a company isn't making money, selling makes sense. But right now the market sells even when companies are making money, even when they beat, and even when they say demand will stay solid all the way through 2028 and 2029. So my recommendation to investors was to buy memory, and within memory, DRAM. Not because I have conviction in the DRAM cycle, but because I see it as a relatively defensive choice within the AI trade. Memory stocks are now trading close to 1x P/B on 2027 BPS. Historically, memory stocks have staged technical rebounds when they fell below 1x P/B. And even if the broader market falls further from here, the three DRAM makers will keep buying back their own shares. I thought that would guarantee a certain level of buying and concluded the stocks would have downside support. Of course, I don't know whether memory stocks can break above their previous highs. That depends on whether sentiment across the broader market recovers. But I think there is a fairly good chance that memory outperforms the rest of the AI trade.

🔬 Even the top AI agents leave about a third of scientific code repair tasks unsolved within a one-hour budget. Title: ScienceIDE: Turning World's Scientific Codebase into Agent Learnable Environments URL: 🧩 Overview AItonomy, Qwen, and PhAI-Labs built ScienceIDE, infrastructure that turns 27 scientific codebases, spanning astrophysics to ocean modeling, into executable environments AI agents can actually learn from. It ships 64 environments and 2,812 tasks. ⚙️ Problem it solves Scientific code encodes decades of expertise, but fragmented toolchains and undocumented numerical conventions make it hard to turn that knowledge into reliable learning experience. 🛠 Methodology "Scientific checks" comparing numerical tolerances and conserved quantities validate task correctness, across 7 task categories like repair and implementation. For RL, a "truncation masking" trick prevents unfairly penalizing trajectories that hit the budget cutoff. 📊 Results Under a one-hour budget, Fable 5.1 leads the agent leaderboard at 67.1%, with Astra at 63.1%. Models trained on ScienceIDE experience (PhAI-IDE) gain up to +33.33 points on scientific code repair, and RL training lifts reward on LAPS tasks 2.4x. #AIAgents# #AI4Science#

🔬 Can AI actually "discover" science? Tested on 90 real Nature papers, even the strongest agent beat SOTA on under 20% of tasks. Title: NatureBench: Can Coding Agents Match the Published SOTA of Nature-Family Papers? URL: 📋 Overview Using 90 tasks drawn from Nature-family journals across six domains, this work tests whether AI coding agents can reproduce and surpass published SOTA. An "information firewall" hides the original method, and NatureGym builds per-task containerized environments for strict, unified evaluation with web search disabled. 🎯 The problem it solves Prior agent-on-research benchmarks were fragmented and low in credibility. This work forces genuine discovery instead of reproduction and creates a common ground for cross-domain comparison. 📐 Methodology Tasks are scored with a SOTA-normalized relative gap (g > 0.1 to surpass, g >= 0 to match), comparing across 81 metrics. Ten agents are evaluated across three harnesses under a 4-hour budget with task-based GPU allocation. 📊 Results ・Even the top agent, Claude Opus 4.7, surpassed SOTA on only 17.8% of tasks and matched on 47.8% ・45.5% of successes were just translating a science task into supervised prediction; genuine domain reasoning was only 8.3% ・Failures were driven by wrong method choice (45.1%) and insufficient compute (24.4%), with task misunderstanding at only 3.1% ・Interdisciplinary tasks consistently scored worse A sober snapshot: today's agents are good at "translation" but weak at real "discovery." #AI4Science# #CodingAgents#

Accurate modeling of CO₂ storage in the subsurface is a critical challenge for scaling Carbon Capture and Storage (CCS). 🌎 But here’s the catch: the inverse problem (recovering geological properties from sparse field observations) is severely ill-posed. 🤒 Traditional approaches either struggle with the scarcity of measured data or are too computationally expensive. 💸 In our new work, we introduce Fun-DDPS (Function-space Decoupled Diffusion Posterior Sampling), a generative framework that decouples the problem into two parts: a function-space diffusion model that learns a prior over geological parameters, and a differentiable Local Neural Operator surrogate for physics modeling and conditioning. ✨ Why does the decoupling matter? The diffusion prior handles the heavy lifting of recovering missing geological information, while the neural operator surrogate makes data assimilation fast and physically grounded — no expensive full-physics simulations in the loop. Key results on synthetic CCS datasets: ⏩ 11x improvement in forward modeling with only 25% observations ✅ Robust inverse modeling for data assimilation under sparse, noisy conditions 🔥 First rigorous validation of diffusion-based inverse solvers against asymptotically exact Rejection Sampling (RS) posteriors This points toward a scalable, practical path for uncertainty-aware subsurface characterization — something the CCS community needs as projects move from pilots to full-scale deployment. 📷 Huge kudos to my amazing collaborators — this work wouldn’t have been possible without them: @IsaacJu13 @AnimaAnandkumar Sally M Benson @ggg_www_ #CarbonCapture# #CO2Sequestration# #MachineLearning# #DiffusionModels# #NeuralOperators# #AI4Science# #CCS# #Sustainability#

FSD v14 Lite is now rolling out to AI3 early-access customers. Based on the feedback, will rollout to more customers over the next few weeks. This build distills the driving behavior from AI4’s v14 series into both the camera and compute config of AI3. It includes destination options and speed profiles on city roads, but more importantly significantly improved safety. We hope you’ll enjoy it, once the build ships wide.

Q2 Shareholder Update → Highlights – Cybercab began production at Gigafactory Texas – Tesla Semi remains on track for volume production this year at our new factory in Nevada – Making continued progress with battery pack capacity expansion (the main limiting factor to near-term vehicle production volume increase) – Megafactory Texas is nearing completion (start of production planned for this year) – More customers are now opting to subscribe to FSD at the time of vehicle purchase! – Robotaxi rollout continued in the US. Now live in 7 major metros – Construction of Optimus at Fremont Factory began after decommissioning the Model S & X lines. Planned production later this year From here, there remains much hard work as we aim to revolutionize transportation, energy and productivity through our leading real-world AI. Scaling will be non-linear and we are focused on long-term value creation. We’ve never been more optimistic about the future. Automotive – Record deliveries in several markets: South Korea, Australia, Colombia, Japan, Taiwan, Thailand, Portugal, the Philippines, Chile, Slovenia & Lithuania – We launched the Model YL in the US in July and have seen a positive response from customers Energy generation and storage – Record energy storage deployments in EMEA, supported by record deployments from Megafactory Shanghai, which continues to ramp production – On track to begin production of Megapack 3 & Megablock this year at our new Megafactory Texas – Powerwall 3P (three-phase) is now available in Germany and is designed to meet the power needs of German homes with a single unit Robotics – Installing the first-generation lines for Optimus at Fremont Factory, where we expect to start production soon – The initial Optimus builds will be used in our Optimus Academy for training data collection and further functionality development – Additionally, we continued site development at Gigafactory Texas with building construction now in full swing AI Training Compute – More than doubled our onsite compute in Texas (in terms of MW of compute) during the first half of 2026 – Cortex 2 supports the development of both vehicle and humanoid robot autonomy software & will ramp further over the rest of the year Battery – Ramping new battery & material factories, including vehicle pack capacity in Berlin, cathode material production and lithium refining in Texas and LFP cells in Nevada for our energy storage products – Increasing production of 4680 cells to support ramping Cybercab & Tesla Semi plus increased production of Model Y Other Supporting Infrastructure – Added over 2,400 net new Supercharging stalls, growing the network by 17% year-over-year AI Software – Started rolling out FSD v14 lite to early-access customers in the US & South Korea with AI3 hardware This software build distills the driving behavior from AI4’s v14 series into both the camera & compute configuration of AI3, bringing destination options & speed profiles. It also addresses challenging driving scenarios with improved proactive & reactive responsiveness AI Inference Compute – Making progress on construction & equipment procurement for our semiconductor fab in Austin Automotive and Other Software – Rolled out Summer Release: Self-Driving stats are now available in the mobile app Grok can make phone calls, search & play music, and adjust climate controls, among other things Automatic Navigation expands beyond home & work to support any destination based on personal habits and schedule Robotaxi – Started production of Cybercab, our purpose-built autonomous EV designed to be the workhorse of our Robotaxi fleet – Began offering employee rides in Cybercabs on our GFTX campus in July – Preparing for expansion of our Robotaxi service to additional US metros: testing, permitting & first responder training – Expanded unsupervised rides to the entire Austin metro area & launched unsupervised rides in Miami, Orlando & Tampa in July FSD Supervised – Record net new subscriptions in Q2 – Record FSD attach rates in North America, with over half of new deliveries including FSD subscriptions – Received additional approvals in Lithuania, Estonia, Denmark & Belgium, with customers in these countries driving over 50 million kilometers (31 million miles) on FSD as of July