Most people don't realize how big of a growth wave we are currently experiencing. So far, with 88% of S&P 500 companies reporting earnings, the index has posted +50.4% growth in earnings year-over-year. To put this into perspective, the already astronomical expectation of +23.1% growth has been more than doubled. This will mark the 2nd consecutive quarter of earnings growth above +25% and the 7th consecutive quarter of double-digit earnings growth for the S&P 500. The last time growth was this strong was in Q2 2021 as the US economy emerged from the pandemic and $4+ trillion in stimulus was handed out. We are arguably in the midst of the biggest technological revolution in US history. The data is absolutely insane.

Yeah, but then I realized I’d be at the mercy of government funding for colliders or telescopes with no way to affect the outcome. Cancellation of SSC being a case in point. And I thought the Standard Model seemed very robust, so colliders at the anticipated energies probably wouldn’t figure out much more than proving Higgs was right.

A TON OF THINGS HAPPENED IN THE STOCK MARKET TODAY. Here's a full recap: 1. SpaceX $SPCX reported Q2 2026 revenue of $7.8B, up 92% YoY, with EPS of -$0.09 vs -$0.24 expected. EBITDA jumped 191% YoY to $3.5B, driven by strong growth across Space, Connectivity, and AI. Space revenue rose 29% YoY to $962M, Connectivity revenue grew 66% YoY to $4.3B, and AI revenue surged 247% YoY to $2.6B. Starlink subscribers doubled YoY to 12.0M, backlog reached $47.5B, and cash plus marketable securities stood at $100B. SpaceX’s CFO said revenue growth accelerated across all business segments, with strong operating leverage and margin expansion led by new AI compute agreements. 2. AMD $AMD reported a strong Q2 2026, with revenue up 50% YoY to $11.5B and adjusted EPS up 246% YoY to $1.66. Data Center revenue more than doubled, rising 107% YoY to $6.72B, while adjusted EBITDA reached $3.3B and free cash flow came in at $1.6B. CapEx was $808M as AMD continues scaling for AI demand. CEO Lisa Su said the company delivered record revenue and profitability, with accelerating EPYC demand, growing Instinct deployments, and Helios beginning to ramp, positioning AMD for substantial revenue and earnings growth as AI drives broader compute demand. 3. $PLTR Palantir has it's best day since April 2025, closing +29%. Remarks from Wall Street today: “Palantir remains our highest-conviction idea following another significant beat-and-raise, with accelerating sovereign AI demand increasing our confidence in the durability of its growth profile.” “Palantir has solved the critical context/semantic layer problem that is key to unlocking value for mission-critical business processes.” “The high-touch deployment model (FDEs) that once drew criticism increasingly appears to be the blueprint for enterprise AI adoption.” “None are considering moving AI apps/agents off of Foundry/AIP to AI-native alternatives.” “Bear case on AI-native competition likely overstated.” Comments came from Rosenblatt, Truist, and Morgan Stanley. Citigroup raised its target to $245 from $200, Piper Sandler reiterated Overweight with a $230 target, Mizuho raised to $215 from $185, Goldman Sachs raised to $204 from $183, and DA Davidson raised to $200 from $175. 4. Rocket Lab $RKLB was awarded a $397M U.S. Space Force contract to build, launch, and operate Flatellite spacecraft for the Space-Based Airborne Moving Target Indicator program. The satellites will be designed to track airborne threats in real time from space, using advanced sensors, secure communications, and low-latency data links. Rocket Lab will launch the spacecraft on its upcoming Neutron rocket and operate them from secure facilities, highlighting the company’s vertically integrated model across satellite design, manufacturing, launch, and mission operations. CEO Peter Beck said the program supports the Space Force’s layered, resilient tracking architecture and strengthens real-time battlespace awareness in contested environments. 5. The top 10 most active options today by contracts traded were $NVDA with 2.6M contracts, $PLTR with 2.0M contracts, $TSLA with 1.6M contracts, $SPCX with 1.5M contracts, $AAPL with 1.3M contracts, $INTC with 1.1M contracts, $AMZN with 991K contracts, $MSFT with 811K contracts, $AMD with 678K contracts, and $MU with 676K contracts. 6. Anthropic has reportedly signed a $10B compute deal with Nvidia-backed Volta Infra, securing capacity from a Norway data center operated in partnership with Bitdeer $BTDR. The deal adds to the growing wave of large-scale AI infrastructure commitments as frontier model labs continue locking up long-term compute capacity. 7. Secretary of State Rubio said Iran and Oman are making progress in talks, including discussions around reopening the Strait of Hormuz, though he emphasized that nothing is final yet. Rubio said officials are hoping an agreement can be reached very shortly, while Treasury Secretary Bessent said a deal with Iran could be announced as soon as tomorrow. Oil fell about 4% on the headlines as markets priced in lower near-term disruption risk. 8. Barclays says S&P 500 earnings are coming in well ahead of expectations, with 85% of companies beating Q2 EPS estimates versus the long-term average of 76%. Revenue is up 11.2% YoY, while EPS has jumped 25.1%, led by Big Tech. Despite the strong numbers, Barclays notes that both beats and misses are seeing negative stock reactions on average, a rare setup that suggests much of the good news was already priced in. 9. The Trump administration is reportedly drafting a ban on U.S. imports of new Chinese data center components, targeting optical transceivers used to move data across fiber-optic networks inside AI data centers. The FCC is said to be working on the measure, with officials hoping to publish it this year, though the restriction could still be modified or shelved. The move is aimed at preventing Chinese technology from becoming entrenched in critical U.S. infrastructure. Photonics and optical names benefited on the report, including $AAOI, $AEHR, $LITE, and $COHR. 10. The WSJ reported today that hedge fund Situational Awareness built a $45B AI-focused powerhouse before scrambling to raise cash after its aggressive AI bets soured in July. Heavy borrowing reportedly triggered lender demands, forcing the fund to sell the bulk of its public stock portfolio to Citadel. Some backers had warned founder Leopold Aschenbrenner about the risks of using significant leverage before the unwind. Aschenbrenner said in his offering document: “There are no limits on the types of securities in which SALP may take positions on behalf of its clients, the types of positions that it may take, the concentration of its investments or the amount of leverage that it may use." Investors were concerned about his communication approach and lack of disclosures with them as the fund continued to use heavy amounts of leverage. 11. Elon Musk said $SPCX SpaceX has decided to build exclusively on Nvidia $NVDA, calling Vera the “best architecture” and the “best AI computer.” Musk said SpaceX deeply values its partnership with Nvidia and expects to surpass 2 GW of compute by year-end, with capacity closer to 10 GW than 5 GW by the end of 2027. The comments are another major endorsement of Nvidia’s AI platform as hyperscalers and frontier AI companies continue locking in massive compute commitments. Elon also said there is a possibility the company will get to $1T in revenue by 2030 and a "non-zero" chance it happens in 2029. 12. The S&P 500 $SPY closed at its highest level on record, adding roughly $1.2T in market cap today as the market pushed further into all-time-high territory. WALL STREET IS THE GREATEST SHOW ON EARTH.

We need Western open source models that we can validate have no backdoors. Not saying Kimi K3 does.... but, just saying

Never open the menu immediately after sitting down at the restaurant table with wealthy families.

This topic is so much more complex in reality and so very few people (especially in Europe) "get it". Its going to be a very tough task for European companies, but at the same time, there is so much opportunity for those who do embrace the change.

Watch a team of humanoid robots running a full 8-hr shift at human performance levels. This is fully autonomous running Helix-02 nitter.net/i/broadcasts/1dxYljYVR…

Introducing Claude Opus 4.7, our most capable Opus model yet. It handles long-running tasks with more rigor, follows instructions more precisely, and verifies its own outputs before reporting back. You can hand off your hardest work with less supervision.

Introducing Gemini 3.1 Pro, our new SOTA model across most reasoning, coding, and stem use cases!

This is a dopamine loop, and it’s one of the most powerful ones humans have ever encountered. Every time you prompt an AI and get a useful result back in seconds, your brain gets a hit. Variable-ratio reinforcement, same mechanism as slot machines, except the reward is real: actual output, actual progress, actual leverage on your ideas. Traditional work follows a delayed-reward structure. You write code for 6 hours, maybe it compiles, maybe you get feedback in a week. The gap between effort and reward is wide enough that motivation decays constantly. AI compresses that loop to seconds. Effort → reward → effort → reward. Your prefrontal cortex stays engaged because the next payoff is always one prompt away. This is why people describe it as “fun” when they’re actually working 14-hour days. The subjective experience of effort disappears when reward frequency is high enough. The “harder than ever” part is real too. When your bottleneck shifts from execution to imagination, you run out of excuses to stop. There’s no “waiting on the build” or “blocked by review.” Every idea you have can be tested immediately, which means your brain never gets a natural stopping point. People who thrive on this are selecting for a specific neurotype: high novelty-seeking, high conscientiousness, tolerance for rapid context-switching. That’s maybe 10-15% of the population. The other 85% will experience the same tools as overwhelming, not energizing. And that split is going to define the next decade of who captures value from AI and who gets displaced by it.

1/ AxiomProver has solved Fel’s open conjecture on syzygies of numerical semigroups, autonomously generating a formal proof in Lean with zero human guidance. This is the first time an AI system has settled an unsolved research problem in theory-building math and self verifies.

Buckle up!

We might already live in the singularity. Moltbook is a social network for AI agents. A bot just created a bug-tracking community so other bots can report issues they find. They are literally QA-ing their own social network. I repeat: AI agents are discussing, in their own social network, how to make their social network better. No one asked them to do this 🦞 This is a glimpse into our future.

𝗜𝗻𝘁𝗿𝗼𝗱𝘂𝗰𝗶𝗻𝗴 𝗧𝘄𝗶𝗻 — 𝘁𝗵𝗲 𝗔𝗜 𝗰𝗼𝗺𝗽𝗮𝗻𝘆 𝗯𝘂𝗶𝗹𝗱𝗲𝗿. No setup. Secure. Infinitely scalable. We just raised a $𝟭𝟬𝗠 𝘀𝗲𝗲𝗱. After a beta with 𝟭𝟬𝟬,𝟬𝟬𝟬+ 𝗮𝗴𝗲𝗻𝘁𝘀 𝗱𝗲𝗽𝗹𝗼𝘆𝗲𝗱, we’re now opening to everyone. RT and comment “Twin” — first agents on us. 👇

Jonathan Ross built Google’s TPU and later Groq’s LPU. Listen to his takes on OpenAI’s compute deals and the differences between Groq’s LPU and Nvidia's GPUs.

🇨🇳 DeepSeek-R1 was published in Nature yesterday as the cover article for their BRILLIANT latest research. They show that pure Reinforcement Learning with answer-only rewards can grow real reasoning skills, no human step-by-step traces required. So completely skip human reasoning traces and still get SOTA reasoning via pure RL. It’s so powerful revelation, because instead of forcing the model to copy human reasoning steps, it only rewards getting the final answer right, which gives the model freedom to invent its own reasoning strategies that can actually go beyond human examples. Earlier methods capped models at what humans could demonstrate, but this breaks that ceiling and lets reasoning emerge naturally. Those skills include self-checking, verification, and changing strategy mid-solution, and they beat supervised baselines on tasks where answers can be checked. Models trained this way also pass those patterns down to smaller models through distillation. AIME 2024 pass@1 jumps from 15.6% to 77.9%, and hits 86.7% with self-consistency. ⚙️ The Core Concepts The paper replaces human-labelled reasoning traces with answer-graded RL, so the model only gets a reward when its final answer matches ground truth, which frees it to search its own reasoning style. The result is longer thoughts with built-in reflection, verification, and trying backups when stuck, which are exactly the skills needed for math, coding, and STEM problems where correctness is checkable. This matters because supervised traces cap the model at human patterns, while answer-graded RL lets it discover non-human routes that still land on correct answers.

The only time more exciting to be alive than today is tomorrow.

One word: relentless. just in the past two weeks, we’ve shipped: 🌐 Genie 3 - the most advanced world simulator ever 🤔 Gemini 2.5 Pro Deep Think available to Ultra subs 🎓 Gemini Pro free for uni students & $1B for US ed 🌍 AlphaEarth - a geospatial model of the entire planet 🏛️ Aeneas - deciphering ancient text (in @Nature) 🥇 Gemini gold-medal level at the IMO 🧸 Storybook - books w/art & audio @GeminiApp ♛ New @Kaggle Game Arena benchmark for LLMs 🐙 Jules, our asynchronous coding agent, out of Beta 🇬🇧 AI Mode for Search available in the UK 📔 NotebookLM Video Overviews 🔥 Gemma passed 200m downloads Now you know why I don't get much sleep 🛌 - too busy pushing the frontier!