The Briefing: Built to Write Perfect Code. Perfect for Breaking It Too.
Issue #2546.5 / November 14, 2025
Hi There 👋
If you thought AI coding assistants were niche tools for developers tinkering in dark basements, think again. Yesterday, Cursor raised $2.3 billion (€2 billion) at a $29.3 billion (€25.2 billion) valuation - more than tripling from $9.9 billion (€8.5 billion) just five months ago. The San Francisco startup has crossed $1 billion in annualized revenue faster than Slack or Zoom ever did, and it’s done so with virtually no marketing budget. Just a product that developers can’t stop raving about.
The round, co-led by Accel and Coatue, brought in strategic investors Nvidia (also a customer) and Google (an AI model supplier), alongside existing backers Andreessen Horowitz and Thrive Capital. Cursor’s secret? It has rebuilt the entire development environment from scratch. Where others slapped AI onto existing editors, Cursor forked VS Code to create what CEO Michael Truell calls a place where “it’s impossible to write bugs.” Bold claim, but the company’s enterprise revenue grew 100x this year, serving the majority of Fortune 500 companies.
The timing matters. While AI models grab headlines, the real money is flowing into the infrastructure layer, the tools that make AI actually useful for work. Cursor isn’t alone: rival Cognition raised $400 million at a $10.2 billion (€8.8 billion) valuation in September, and Replit secured $250 million (€215 million) this summer. The common thread? These aren’t consumer toys. They’re productivity amplifiers that make expert developers exponentially more effective, letting them focus on architecture while AI handles the repetitive grind.
When Your AI Agent Gets Hacked
Speaking of AI actually working in the wild, Anthropic dropped a sobering report yesterday about detecting what it calls a “highly sophisticated espionage campaign” targeting its Claude coding tool. A group believed to be Chinese state-backed manipulated Claude’s AI agents to infiltrate 30 organizations - tech firms, financial institutions, government agencies, chemical manufacturers - succeeding “in a small number of cases.”
The AI let attackers operate at speeds that would be “simply impossible” for human hackers. When your automated assistant can write, test, and deploy code in seconds rather than hours, it turns out cyber criminals get the same productivity boost as legitimate developers. Anthropic banned the accounts and alerted targets, but the company was refreshingly blunt in its assessment: these attacks will become more effective over time.
Now for the slightly awkward part. When asked why they’re continuing to develop and release increasingly capable AI agents, given the misuse potential, Anthropic’s answer essentially boiled down to: “Because the abilities that make Claude handy for attackers also make it crucial for cyberdefense.” It’s the digital arms race version of the gun lobby’s “good guy with a gun” argument. The artificially intelligent cat is out of the bag, they say, so we might as well sell cybersecurity software to defend against the tools we’re building.
There’s truth to this. You can’t un-invent AI agents, and defenders do need access to the same capabilities as attackers. But it’s worth noting that Anthropic, OpenAI, and Google are all racing to push these tools into the market while simultaneously warning about the risks. The consumer takeaway? Don’t hand your sensitive data to just any AI agent. The business takeaway? Your cybersecurity budget is about to get a lot bigger.
Europe’s Competitiveness Retreat
While American tech companies pour billions into AI infrastructure, European lawmakers spent yesterday dismantling what was supposed to be the continent’s competitive advantage: setting global standards for corporate accountability. The European Parliament voted 382-249 to dramatically scale back the Corporate Sustainability Reporting Directive (CSRD) and Corporate Sustainability Due Diligence Directive (CSDDD).
The changes are substantial. CSRD reporting thresholds jumped from 250 to 1,750 employees, with a new €450 million revenue requirement, removing roughly 80% of companies from scope. CSDDD now only applies to firms with 5,000+ employees and €1.5 billion+ in annual revenue. Most notably, companies are no longer required to prepare climate transition plans aligning with the Paris Agreement. The Parliament also eliminated climate transition plan requirements and shifted liability for non-compliance from the EU to the national level.
The vote followed intense lobbying pressure, with ExxonMobil reportedly the most active corporate actor. The European People’s Party (EPP) threatened to align with far-right parties unless center-left groups agreed to deeper cuts, a political maneuver that ultimately worked. Sustainable investment groups called it “gutting landmark laws,” while supporters argued Europe needed to ease compliance burdens to compete with the US and China.
The sad irony is that Europe spent years building a first-mover advantage in sustainability regulation, creating a framework that companies were already investing billions to meet. Now, just as that system was becoming operational, lawmakers are scaling it back based on competitiveness concerns, while American companies race ahead on the technologies that will actually define the next decade of competition. If Brussels wanted to boost European competitiveness, it might have been better off investing those political calories into, say, AI infrastructure rather than dismantling environmental accountability that investors were increasingly demanding anyway.
In Other News
Blue Origin finally stuck the landing. Jeff Bezos’s space company successfully launched and landed its New Glenn rocket yesterday, only the second-ever flight of the massive orbital vehicle. The rocket delivered NASA’s twin ESCAPADE spacecraft toward Mars while bringing the first-stage booster back to a seafaring platform 9 minutes after launch. It’s the first time a company has landed an orbital booster on just the second attempt, making Blue Origin only the second firm after SpaceX to achieve this milestone. The 98-meter New Glenn is now being certified by the US Space Force for national security missions, with customers including Amazon’s Project Kuiper and several telecommunications providers. Markets reacted accordingly: finally, some competition for SpaceX.
Google wants to do your shopping for you. The search giant rolled out “agentic checkout” yesterday, AI that tracks prices and automatically purchases items when they hit your target price via Google Pay. There’s also “Let Google Call,” which uses upgraded Duplex technology to phone local stores on your behalf to check inventory and prices (the AI identifies itself as automated, and stores can opt out). The features launch today in the US just ahead of Black Friday, working with Wayfair, Chewy, and select Shopify merchants. Between this, Amazon’s Rufus assistant (up 210% year-over-year), and OpenAI’s Walmart partnership, the battle for AI-powered commerce is heating up fast—with an estimated $73 billion in AI-influenced sales expected through Cyber Monday.
Microsoft found a shortcut in the chip wars. Rather than playing catch-up with Google and Amazon on custom silicon, the company is licensing OpenAI’s chip designs, being developed with Broadcom. CEO Satya Nadella confirmed Microsoft gets full access to OpenAI’s system-level innovations through 2030 and AI models through 2032. “As they innovate even at the system level, we get access to all of it,” Nadella explained on a podcast. “We first want to instantiate what they build for them, but then we’ll extend it.” It’s a pragmatic move that lets Microsoft leverage a partner already ahead in AI-optimized hardware rather than spending years reinventing the wheel, though it also highlights how dependent the company remains on OpenAI for AI infrastructure.
The race to Mars continues. NASA’s ESCAPADE mission (those twin spacecraft Blue Origin just launched) won’t actually reach Mars until 2027. The probes will spend 11 months in an “Earth-proximity” orbit looping around Lagrange point 2, about a million miles away, until the planets align for the ideal transfer window in late 2026. Once at Mars, they’ll investigate how solar wind interacts with the planet’s magnetic environment and contributes to atmospheric escape, research that could explain how Mars transformed from wet and warm to dry and dusty. The mission costs under $80 million, making it one of NASA’s more economical interplanetary efforts. Secretary Sean Duffy noted the data will be “critical to protect future NASA explorers” as the US works toward President Trump’s vision of planting the Stars and Stripes on Mars.
Thank you for reading The Briefing — your nightly rundown of the key stories shaping business, technology, and innovation. If you like what you see, subscribe for free to receive the next editions in your inbox daily (Mon-Fri).