Weekly Download #26

This week in Silicon Valley, Startups & Tech:

• Big Tech's $764M Influence Campaign: Silicon Valley went shopping on elections to fuel Congressional push preempting state AI laws

• The 50-State AI Showdown: Trump demands “Federal Standard,” warning China will win, while prominent Republicans publicly oppose him

• Google's Gemini 3 Coronation: Alphabet stock soared 6.9% to a record $299.66 high as the new model crushed OpenAI on benchmarks

• Gemini 3's Antigravity Moment: Google's terse, blazingly fast model with integrated IDE proves better pre-training and post-training wins

• Prediction Markets' Trump-Era Gold Rush: Kalshi and Polymarket race to scale amid regulatory opening as critics warn about expanding gambling

As we continue to build at Context Ventures, we've launched Startup Intros with a simple mission: to help early-stage founders find the right investors, faster and smarter.

🎪 The $764M Regulatory Heist: Big Tech Buys Fed Control

💸 Big Tech Dropped $764M to Erase State AI Laws

Big Tech spent $764.5M during the 2024 election cycle, with nearly three-quarters flowing to Republicans, and now they're cashing in:

• $1.1B total: Campaign spending, lobbying, and political influence combined

• ~$380M to GOP: Elon Musk alone contributed nearly half

• Second Only to Pharma: Tech's lobbying muscle now rivals Big Pharma's stranglehold on DC

The Prize? A provision tucked into the National Defense Authorization Act that would preempt state AI regulations and centralize control in DC, where industry lobbying reigns supreme.

Translation: States passed bipartisan laws addressing AI deepfakes, workplace surveillance, and algorithmic discrimination. Tech executives saw democracy working at the local level and decided to buy a federal override instead. OpenAI and Anthropic are already backing the push, craving the "predictability" that comes from negotiating with one government instead of fifty.

⚖️ Federal AI Power Grab Dies on Bipartisan Hill

The White House paused a draft executive order that would have steamrolled state AI laws—but not before triggering bipartisan backlash.

The Proposed Order: Ominously titled "Eliminating State Law Obstruction of National AI Policy," it would have created an "AI Litigation Task Force" to sue states into submission while withholding federal funding from non-compliant jurisdictions.

Trump’s Position: The President is still calling on Congress to block state-level AI regulation entirely, warning that a "patchwork of 50 State Regulatory Regimes" will hand China the AI crown.

The Irony: Prominent Republicans are publicly opposing federal preemption, arguing that federal seizure of jurisdiction is "not a direction, generally, you'd want to go." The GOP's commitment to states' rights—usually deployed for culture war issues—is accidentally creating the regulatory fragmentation Silicon Valley fears most.

Bottom Line: Everyone agrees AI needs rules, but nobody agrees who should write them. Big Tech is playing both sides, lobbying for "uniform standards" while quietly preferring the chaos that lets them forum-shop for the weakest regulations.

⚡ Policy Quick Hits: Chip Diplomacy and Betting on Everything

🔄 Trump Team Floats U-Turn on Nvidia China Ban: Sources report internal discussions about allowing Nvidia H200 sales to China, which a potential reversal of export controls from both administrations. The H200 is already a generation behind Blackwell, so this is Trump's team testing how much they can give away.

🏛️ White House Tells Congress: Hands Off Our China Chip Ban: Key officials are lobbying lawmakers to strip AI chip export restrictions from the defense bill, preferring executive control over legislative permanence. Translation: they want freedom to negotiate and selectively enforce without congressional guardrails.

🎲 Prediction Markets Cash In on Regulatory Green Light: Kalshi hit $1B weekly volume while Polymarket runs $464M, as both platforms race to scale during Trump-era regulatory opening. One plays by CFTC rules, the other builds around them, which is classic regulatory arbitrage.

🏆 Google's Victory Lap: From AI Laggard to Leaderboard King

👑 Google Finally Takes the Crown

Gemini 3 just surged ahead of OpenAI and every other competitor on benchmark tests, marking the first time in years that Google decisively leads the AI race:

• 1,501 Elo Rating: Highest score ever on the LMArena leaderboard, crushing GPT-5.1 on 19 out of 20 benchmarks

• 91.9% on GPQA Diamond: Crushing Ph.D.-level reasoning tasks

• 23.4% on MathArena Apex: A stunning jump from Gemini 2.5 Pro's 0.5%

• 50%+ More Coding Tasks Solved: Outperforming even OpenAI's latest GPT models

• 1M Token Context: Processing massive documents, images, and video in one session

Translation: Google's custom TPU infrastructure gives Gemini 3 the speed and scale to operationalize these benchmark wins into real-world apps. This isn't just lab performance; it's production-ready dominance from the company that invented the transformer architecture.

📈 Wall Street Picks Its Winner

Alphabet's stock surged 6.9% on Wednesday, its biggest single-day gain since early September, before closing at a record high of $299.66:

• 55% Year-to-date Gain: Google outpaced every major tech peer in 2025

• $50B Annual Run Rate: Google Cloud revenue, with AI driving acceleration

• 9 of 10 Top AI Labs: Running as Google Cloud customers

Incredible Timing: Warren Buffett's Berkshire Hathaway just disclosed a new stake in Alphabet right as Gemini 3 dropped.

Why This Matters: Google has something OpenAI and Anthropic can only dream about: distribution. Today, 9M devs are already building with Gemini, and the model is integrated across Search, Cloud, and consumer products, reaching billions of users instantly.

Reality Check: A 7% pop on positive reviews is Wall Street saying "prove it." The real test comes when users actually choose Gemini 3 over ChatGPT, not when tech reviewers say nice things about benchmarks.

🎯 Speed Demon With a Personality Transplant

Matt Shumer's hands-on review reveals that Gemini 3's biggest upgrade isn't raw intelligence; it's that Google finally taught its AI to shut up and move fast:

• Consistency: Less "spiky" than predecessors, quality stays high instead of randomly faceplanting

• Speed: Real-time productivity without lag, even on complex reasoning

• Personality: Concise answers that respect your time (no more AI mansplaining)

• Antigravity IDE: A genuinely mature dev environment at launch

• Catch: Higher hallucination rates versus some competitors, though the UX is solid

The Reality: The model crushes competitors in "intelligence per second" while delivering a terse, direct personality that skips the verbose preambles plaguing other models. Shumer describes it as working with a "skilled senior engineer": efficient and accurate, but still requiring verification for critical tasks.

🚀 Scaling Laws Aren't Dead, Just Misunderstood

Google's Gemini co-lead Oriol Vinyals threw cold water on the "scaling is dead" narrative, crediting Gemini 3's breakthrough to better pre-training and post-training; the thing everyone declared was hitting a wall.

The Evidence: Gemini 3 matches Gemini 2.5 at roughly 1T parameters, yet delivers what Vinyals calls the largest performance jump in Google's history. As Tomasz Tunguz notes, the virtuous cycle is confirmed: pre-training improvements, post-training advances, and inference optimization add up to compounding progress, not plateaus.

Translation: Google didn't just scale bigger; they scaled smarter. While the rest of the industry was writing eulogies for pre-training, DeepMind was quietly proving there's still juice left in the fundamentals.

🌍 Hassabis Reveals His Research Obsession: World Models

Google DeepMind's CEO Demis Hassabis pulled back the curtain on Gemini 3, describing the breakthrough as requiring "world-class research, world-class engineering, and world-class infra all working closely together:"

• 2B people use Search AI Overviews powered by Gemini monthly

• 650M Users on the Gemini app itself

• Entire Google Search index being integrated into the model

Most Revealing Detail: Hassabis shared that he personally spends his research time on world models. These AI systems build internal representations of reality for better planning and generalization.

Reality Check: On the AI bubble question, Hassabis played it cagey, saying Google is "positioned strongly whether there's no bubble or there is one and there's some retrenchment." When your CEO hedges like that while cramming the internet's knowledge into one system, he's either being cautious or building something he knows might not survive economically.

⚡ Startup Quick Hits: When Billionaires Get Bored

💾 Lambda's $1.5B "Superintelligence Cloud" Bet: The GPU cloud provider raised a massive Series E from TWG Global to build "gigawatt-scale AI factories." This came right after announcing a multi-billion-dollar deal to supply Microsoft with tens of thousands of Nvidia GPUs. 

🎰 Kalshi Doubles Valuation in Two Months to $11B: The prediction market platform raised $1B from Sequoia and CapitalG, jumping from a $5B valuation just eight weeks ago. Their annualized trading volume exploded from $300M last year to $50B this year!

🎬 Luma AI's $900M Saudi Supercomputer Deal: The video generation startup raised a massive Series C led by HUMAIN (Saudi PIF's AI company), plus AMD Ventures and existing backers a16z and Amplify Partners.

🐙 Kraken's $800M Wall Street Embrace at $20B: The crypto exchange that previously raised just $27M in 14 years suddenly pulled in $800M from the most TradFi investor group imaginable: Jane Street, Citadel Securities ($200M), DRW, and Oppenheimer.

🤖 Physical Intelligence Hits $5.6B for Robot Brains: The year-old robotics AI startup raised $600M led by CapitalG, with Jeff Bezos, Thrive Capital, Lux Capital, Index Ventures, and T. Rowe Price piling in. Their π0 model is a "universal robot brain" that can control any robot.

💰 Investor Quick Hits: From $100B AI Plays to Lawyers Getting In

🏗️ Brookfield's $100B AI Infrastructure Moonshot: The Canadian asset manager launched a $100B global AI infrastructure program with Nvidia and the Kuwait Investment Authority. The anchor fund targets $10B in equity, which they'll leverage to buy everything from raw land to data centers to compute.

⚖️ The LegalTech Fund's 4x Fund II: The Ft. Lauderdale firm closed $110M for its second fund, nearly quadrupling its $28.5M debut from 2021. The fund backs 80+ legaltech startups going after the $1T legal services market.

🇮🇱 Meron Capital's Quiet 80x Returns: The Israeli pre-seed & seed investor closed $70M for Fund III, bringing total capital to $170M. Their first two funds deployed just $100M but built a portfolio now valued at $8B+, with unicorns EON, Immunai, and Solugen plus seven exits.

💳 Foxe Capital Emerges in DC: Ruth Foxe Blader (ex-Anthemis, backed Lemonade) and Dan Dall'Asta (ex-Fifth Third Bank) are raising $50M for their debut fintech fund. The thesis: as AI levels the playing field on "good tech," deep industry expertise becomes the differentiator. 

💸 In Other Bay Area Funding News

Here's a roundup of notable recent funding rounds across the Bay Area:

🤖 AI & Tech

• Genspark: Raised $275M Series B for an AI search platform

• Peek: Raised $70M Series D for travel booking & management software

• Gridware: Raised $55M Series B for grid monitoring & wildfire risk prevention

• Archetype AI: Raised $35M Series A for AI that interprets real-world sensor data

• Maxima AI: Raised $30M Series A for AI that automates enterprise accounting

• Bedrock Data: Raised $25M Round for unified data governance & management tools

• Coverbase: Raised $16.5M Series A for vendor intake, risk & contract workflows

• Automat AI: Raised $15.5M Series A for AI Chrome automation

• Parallax Worlds: Raised $4M Seed for AI deployment simulations that test failures

🏥 Healthcare & Biotech

• Solve Therapeutics: Raised $120M Round for biopharma drug discovery

• VahatiCor: Raised $23M Series B for cardiovascular medical hardware

• Voio: Raised $8.6M Seed for AI medical imaging diagnostics

• Erg Bio: Raised $6.5M Seed for sustainable biofuel & chemical production

💸 Fintech & Financial Services

• Federato: Raised $100M Series D for AI insurance underwriting

• Numeric: Raised $51M Series B for automated financial workflows & reporting

• Obex: Raised $37M Round for fintech infra backed by compute & energy credit

• Pibit: Raised $7M Series A for underwriting insights & loss data tools

• Coreworks AI: Raised $4M Seed for natural-language finance analytics

• Kaaj AI: Raised $3.8M Seed for AI credit intelligence & underwriting

🏭 Hardware & Industry

• PicoJool: Raised $12M Round for scalable optical communications hardware

• Cavela: Raised $6.6M Seed for global manufacturing network tools

• alphaXiv: Raised $7M Seed for accelerating AI deployment from research to production

🌟 Editor's Note

At Startup Intros, our mission is to bring the latest founder-investor news straight to your inbox, keeping you ahead in the fast-paced world of Silicon Valley.

💭 Parting Thoughts

So there you have it: a week where Michael Burry and Peter Thiel are shorting AI's golden children while the industry burns $371B annually and pitches lunar data centers.

Remember: when your bull case requires ignoring investors, breaking climate pledges, and launching GPUs into orbit, you're not disrupting the future, you're in the final act of a very expensive bubble.

Forward to a friend or hit reply to let me know what you're seeing in your world.

Till next time!

P.S. Raised this week, and we missed you? Want to be featured? Have tips or funding questions? Reply or DM us as we’re here to help.

Startup Intros Weekly Download: Your trusted source for founder-investor insights, delivered with clarity and focus.