Marc Andreessen joins the Pentagon’s board, and the old line blurs

Cargo Culture

A sharp critique of Silicon Valley's AI boom as a 'cargo cult' — where tech giants, VCs, and AI labs mimic past success patterns without genuine innovation. The piece argues that the industry has run out of ideas, with companies like OpenAI and Anthropic burning tens of billions while lacking viable paths to profitability. It covers the collapse of VC returns, the 'SaaSpocalypse,' the hollow hiring of AI luminaries, the push for token-burning 'loops,' and how executives cargo-cult management philosophies like 'founder mode' — all symptoms of an industry doing impressions of success rather than creating real value.

Mirendil raises $200M to build AI that improves AI

Mirendil, a startup founded by two ex-Anthropic researchers, has raised $200M at a $1B valuation in one of the largest AI seed rounds on record. Co-led by Andreessen Horowitz and Kleiner Perkins with Nvidia participating, the company aims to build a platform that automates AI research itself — designing experiments, tuning models, and running training loops. The pitch targets organizations without large ML teams, letting them build domain-specific models in days rather than months. The core thesis exploits a structural gap: major labs like Anthropic use AI to improve their own AI internally but prohibit customers from doing the same via terms of service. Mirendil wants to sell that capability openly. The concept touches on recursive self-improvement, which Anthropic has flagged as a safety concern, though the founders argue it can be supervised. The team of ~20 draws from Anthropic, xAI, Google DeepMind, and OpenAI, and a product is expected in the coming months.

"I literally built an investment firm that wouldn't have funded my own pitch" – Why the money isn't reaching the games that need it

A workshop at Madeira Games Summit revealed a fundamental disconnect between how game developers pitch and how investors evaluate opportunities. When participants were asked to design investment funds with $1 million, almost none allocated capital to traditional game content — favoring tools, infrastructure, and platforms like Roblox instead. Key insights include a language gap (developers lead with creative vision; investors demand return structures), the importance of sequencing pitches strategically over 9-12 months, and the need to explicitly address return expectations and timelines upfront. Recommendations include researching fund remits before pitching, talking directly with fund managers, and advocating for game-specific fund structures like revenue-share models and three-game deal structures.

How to invest when everything is moving too fast

Hai nhà đầu tư mạo hiểm tập trung vào AI là Carter Reum (M13) và Chang Xu (Basis Set Ventures) chia sẻ cách đánh giá và đầu tư trong giai đoạn bùng nổ AI hiện nay. Họ nhấn mạnh thách thức định giá khi tăng trưởng vượt ngưỡng lịch sử, ưu tiên các startup trong ngành được quản lý chặt hoặc thị trường ngách (depth markets) thay vì thị trường cạnh tranh khốc liệt (velocity markets), đồng thời khẳng định lợi nhuận thường đến từ những khoản đầu tư muộn hơn trong chu kỳ công nghệ. Họ cũng đề cập đến tác động tiềm tàng của đợt IPO SpaceX đối với hệ sinh thái công nghệ Los Angeles (LA) và vai trò quan trọng của "gu văn hóa" địa phương khi AI tự động hóa công việc kỹ thuật.

Lập trình viên nên đọc bài này để hiểu cách đầu tư vào AI không chỉ là kỹ thuật viết code mà còn là chiến lược tìm kiếm cơ hội giá trị lâu dài trong một thị trường biến động nhanh như hiện nay.

Top Quantum Computing Investors in 2026

A comprehensive overview of the leading quantum computing investors in 2026, covering dedicated quantum VC funds (Quantonation, 55 North, Firgun Ventures), corporate venture arms (Google Ventures, IBM Ventures, Microsoft M12, Amazon), sovereign wealth funds (Temasek, Mubadala), and government-backed programs (In-Q-Tel, SGInnovate, HTGF). Private VC in quantum reached $4.9 billion in 2025, more than doubling the prior year. The piece profiles each investor's focus, confirmed portfolio companies, preferred funding stage, and geographic orientation. It also covers how to pitch quantum investors, key metrics they evaluate (qubit count, gate fidelity, coherence times, commercialization roadmap), typical funding amounts by stage, and the complementary roles of government grants versus venture capital for quantum startups.

Premium: Notes From The Bubble, Volume 1

Ed Zitron argues that major hyperscalers (Google, Meta, Microsoft, Amazon) have run out of genuine innovation and are propping up valuations through massive AI capex spending that will never generate sufficient returns. He contends that LLMs are a flawed interface bolted onto existing products, that none of these companies are building truly new platforms, and that Meta is the most vulnerable hyperscaler due to its near-total dependence on advertising revenue. The AI bubble, he predicts, will burst not with a bang but through a slow capex pullback — first signaled by a semiconductor firm filing or canceled order — triggering brutal stock drops and leaving the tech industry with nothing meaningful to look forward to afterward, unlike the post-Dot-Com era which had Google, AWS, and PayPal in the pipeline.

Helsing lands Germany's €580M combat-cloud contract

Germany is set to award a €580M contract to Munich-based AI defense startup Helsing to build the Combat Fighter System Nucleus (CFSN), a combat-cloud software platform salvaged from the collapsed Franco-German FCAS fighter jet project. The contract covers two experimental uncrewed combat aircraft, ground control stations, and autonomy software, with a government-owned reference architecture for future suppliers. Helsing beat out Airbus Defence, MBDA Germany, and Diehl Defence. The deal is politically sensitive: Germany is bypassing EU tender rules via a national-security exemption and may skip a Bundestag budget vote for a second phase. Founded in 2021, Helsing is now valued at ~€12B and has products deployed in Ukraine.