Binance and OpenAI Ignored Repeated…

1. Key Themes

Theme 1: Security failures are driven by known, ignored risks, not novel attacks

Binance: a $6.3B loss from a well-known social-engineering tactic

The lead story frames the breach as a failure of attention to known warnings rather than a sophisticated zero-day.

"Binance says North Korean hackers stole $6.3 billion after posing as recruiters and tricking a wallet engineer into downloading malware despite years of industry warnings about the tactic and at least 57 suspected North Korean attacks against the company itself."

OpenAI: speed-versus-security tradeoffs flagged internally for years

The pattern repeats at the leading AI lab, with employees warning leadership that competitive pressure was winning out over defense.

"OpenAI employees warned executives and the board for years that lax internal security could expose sensitive technology to outsiders, including foreign governments, with some concluding that management prioritized keeping pace with rivals over tightening defenses."

Security is becoming a funded, acquirable category

The CrowdStrike deal suggests buyers are paying up for AI-specific security capabilities.

"CrowdStrike agreed to acquire AI security startups Fiddler AI and Dreadnode for a combined $495 million, paying roughly $400 million for Fiddler, which monitors and explains AI-agent behavior, and about $95 million for Dreadnode, which tests models and agents for vulnerabilities before deployment."

Theme 2: Enterprise AI demand is real but may be less durable than headline numbers suggest

Palantir's commercial inflection

U.S. commercial revenue overtook government revenue for the first time, driven by enterprises wanting AI that keeps their data private.

"The clincher was that Palantir's U.S. commercial revenue surpassed U.S. government revenue for the first time ever. The company reported $984 million in commercial contracts (up 148% from the year-ago quarter), versus $953 million in U.S. government revenue."

Why enterprises buy: data control plus hands-on engineering

"They are signing up for the company's AI software that can run frontier models and custom-train open-weight models securely, without sharing any of their data and business processes with outside AI labs. Plus, Palantir has long offered what is today known as 'forward-deployed' software engineering, in which its own engineers work directly with customers to customize the software."

The warning signal: short-term trials

"However, Palantir also revealed a less-than-rosy signal: Even as enterprises are moving faster than ever, they are often signing short-term trials."

Theme 3: Capital is concentrating in AI infrastructure and compute, with valuations stepping up fast

Fal's valuation doubled in roughly seven months

"Fal, a five-year-old San Francisco startup that provides infrastructure for running image, video, and audio AI models, raised a $1.3 billion round at a $16 billion valuation, double the valuation it commanded in April."

Custom enterprise models and GPU-as-a-financial-asset

"Applied Compute, a one-year-old San Francisco startup that trains, deploys, and continually improves custom AI models for enterprises, raised a $350 million round at a $3.25 billion valuation."

"Ornn, a one-year-old New York startup that provides a marketplace and financial infrastructure for buying, selling, and pricing GPU compute capacity, is reportedly in talks to raise a third tranche of its Series A at a valuation of more than $1 billion shortly after closing a second tranche at an $800 million valuation."

Power is the bottleneck, spawning creative workarounds

"Old jet engines pulled from aircraft boneyards are being converted into generators to power AI data centers, giving developers a way to secure electricity in as little as two years rather than wait through multi-year grid-connection delays."

Theme 4: Frontier-lab economics and distribution are shifting

Anthropic's compute costs are outpacing revenue growth

"Anthropic is still grappling with the economics of compute, with infrastructure costs rising to $10 billion in the third quarter — 58% of revenue, up from 48% the prior quarter — putting pressure on margins even as sales climbed to $17 billion."

ChatGPT is approaching meaningful search scale

"OpenAI says ChatGPT now handles search queries equal to about 20% of Google's search volume, a sharp contrast with Bing's roughly 3% share cited during Google's 2023 antitrust trial; the company did not disclose the source of its Google volume estimate."

Theme 5: AI is vertically applied to services-heavy, document-heavy industries

Agents replacing manual professional workflows

"Clark, a one-year-old New York startup that develops AI agents that review documents, perform tests, prepare workpapers, and follow up with audit clients, raised an $11 million seed round led by Touring Capital."

"Athena, a San Francisco startup founded this year that uses AI agents and affiliated attorneys to prepare and review commercial real estate deal documents for fixed fees, raised a $2.5 million pre-seed round led by Andreessen Horowitz."

2. Contrarian Perspectives

A record earnings beat can be a negative signal Conventional wisdom says a large beat plus raised guidance lifts a stock. The article argues the quality of the revenue matters more than the headline.

Evidence of strength: "Revenue grew 96% from a year earlier, and 20% from the previous quarter, to $2.31 billion, beating analysts' expectations of $2.18 billion... The company even raised its revenue guidance, or forecast, for the full year. Normally, that alone would send a stock soaring."
Why it is nonetheless a caution: "Even as enterprises are moving faster than ever, they are often signing short-term trials." Fast adoption paired with short commitments implies revenue that may not renew or compound.

Rapid growth in AI revenue does not mean improving margins Anthropic's top line grew, but its cost structure moved the wrong way, which cuts against the assumption that scale automatically brings leverage.

"...infrastructure costs rising to $10 billion in the third quarter — 58% of revenue, up from 48% the prior quarter — putting pressure on margins even as sales climbed to $17 billion."

Top-tier security does not come with top-tier AI valuations The security failures at the best-funded companies suggest well-capitalized firms are not necessarily well defended, and that the best-resourced targets are being outmaneuvered by basic tactics.

"...despite years of industry warnings about the tactic and at least 57 suspected North Korean attacks against the company itself."

3. Companies Identified

Binance

Description: Crypto exchange.
Why mentioned: Case study in ignoring repeated, well-known security warnings.
Quote: "Binance says North Korean hackers stole $6.3 billion after posing as recruiters and tricking a wallet engineer into downloading malware."

OpenAI

Description: Leading AI lab and maker of ChatGPT.
Why mentioned: Internal security concerns, a physical security incident, government and national-security hiring, and growth in search usage.
Quotes: "The concerns gained new urgency after an OpenAI engineer was charged last month with planting a hidden recording device in an executive conference room." / "OpenAI says ChatGPT now handles search queries equal to about 20% of Google's search volume."

Palantir

Description: Data analysis and enterprise AI software company.
Why mentioned: Record quarter; commercial revenue surpassed government revenue; short-term trial signal.
Quote: "The data analysis company Palantir delivered such a huge third-quarter earnings report on Monday that its press release even bragged that the numbers were 'crushing consensus expectations.'"

Applied Compute

Description: Trains, deploys, and continually improves custom AI models for enterprises.
Why mentioned: $350M raise at $3.25B valuation, one year after founding, led by Elad Gil.
Quote: "raised a $350 million round at a $3.25 billion valuation. The deal was led by Elad Gil."

Fal

Description: Infrastructure for running image, video, and audio AI models.
Why mentioned: Valuation doubled in about seven months.
Quote: "raised a $1.3 billion round at a $16 billion valuation, double the valuation it commanded in April."

Ornn

Description: Marketplace and financial infrastructure for GPU compute capacity.
Why mentioned: Rapid stepped-up Series A tranches, treating compute as a tradable asset.
Quote: "is reportedly in talks to raise a third tranche of its Series A at a valuation of more than $1 billion shortly after closing a second tranche at an $800 million valuation."

Anthropic

Description: Frontier AI lab.
Why mentioned: Compute cost pressure on margins despite strong revenue.
Quote: "infrastructure costs rising to $10 billion in the third quarter — 58% of revenue, up from 48% the prior quarter."

CrowdStrike

Description: Cybersecurity company.
Why mentioned: Acquirer consolidating AI security startups.
Quote: "agreed to acquire AI security startups Fiddler AI and Dreadnode for a combined $495 million."

Fiddler AI

Description: Monitors and explains AI-agent behavior.
Why mentioned: Acquired for roughly $400M.
Quote: "paying roughly $400 million for Fiddler, which monitors and explains AI-agent behavior."

Dreadnode

Description: Tests models and agents for vulnerabilities before deployment.
Why mentioned: Acquired for about $95M.
Quote: "about $95 million for Dreadnode, which tests models and agents for vulnerabilities before deployment."

Clark

Description: AI agents for audit workflows.
Why mentioned: $11M seed in an AI-for-professional-services niche.
Quote: "develops AI agents that review documents, perform tests, prepare workpapers, and follow up with audit clients."

Corridian

Description: Autonomous underwater vehicles for military cargo and surveillance.
Why mentioned: Large seed ($38M) in defense tech.
Quote: "raised a $38 million seed round led by Point72 Ventures."

Guava

Description: Proprietary voice models and developer tools for AI customer-call agents.
Why mentioned: $12.5M seed for a company founded this year.
Quote: "builds proprietary voice models and developer tools for AI agents handling customer calls."

Quint Computing

Description: Superconducting quantum processors with error correction built into the chip architecture.
Why mentioned: Quantum hardware funding plus an EU grant.
Quote: "develops superconducting quantum processors that integrate error correction into chip architecture for fault-tolerant quantum computers."

Specific Biologics

Description: Gene-editing therapies targeting disease-causing repetitive DNA.
Why mentioned: $35M Series A.
Quote: "develops gene-editing therapies that remove disease-causing repetitive DNA sequences linked to serious genetic disorders."

StimCardio

Description: Wrist-worn nerve-stimulation device to reduce atrial fibrillation recurrence.
Why mentioned: $30M Series A.
Quote: "develops a wrist-worn device that stimulates nerves to reduce atrial fibrillation recurrence after catheter ablation."

Syncere

Description: Robotic floor lamps that fold laundry and tidy rooms using local AI.
Why mentioned: $11M pre-seed led by a16z; consumer robotics bet.
Quote: "develops robotic floor lamps that use local AI to fold laundry, tidy rooms, and make beds."

Athena

Description: AI agents plus affiliated attorneys for commercial real estate deal documents.
Why mentioned: A hybrid AI-plus-human services model with fixed fees.
Quote: "uses AI agents and affiliated attorneys to prepare and review commercial real estate deal documents for fixed fees."

Bioliberty

Description: Robotic gloves and software for stroke recovery.
Why mentioned: Series A extension.
Quote: "develops robotic gloves and interactive software that help stroke survivors and neurological patients restore upper-limb mobility."

CyberKord

Description: AI to identify and resolve IT and cybersecurity issues.
Why mentioned: Pre-seed in AI security operations.
Quote: "uses AI to identify, investigate, and resolve companies' IT and cybersecurity issues with less manual intervention."

Gallium Defense

Description: Radars that detect and track small drones.
Why mentioned: Counter-drone defense seed round.
Quote: "develops radars that detect and track small drones for armed forces and critical infrastructure operators."

Kodiact

Description: Raw-material market analysis and sourcing simulation for manufacturers.
Why mentioned: Pre-seed in supply chain intelligence.
Quote: "helps manufacturers analyze raw-material markets and simulate sourcing decisions before price changes affect their margins."

Roleo

Description: AI career agent matching tech professionals to jobs without applications.
Why mentioned: $3.2M round.
Quote: "uses an AI career agent to match technology professionals with relevant jobs without requiring applications."

Viridi

Description: Catalysts converting captured CO2 into surfactants and other chemicals using existing factory equipment.
Why mentioned: Carbon-utilization funding.
Quote: "develops catalysts that let manufacturers turn captured carbon dioxide into surfactants and other chemicals using existing factory equipment."

Enginsight

Description: Cybersecurity platform spanning vulnerability management, threat detection, compliance, and automated response.
Why mentioned: Undisclosed round led by Verdane.
Quote: "combines vulnerability management, threat detection, compliance monitoring, and automated incident response."

Canva

Description: Design software company.
Why mentioned: First major layoffs amid slowing growth and AI competition.
Quote: "is carrying out its first major round of job cuts, with 1,271 employees or 22% of its workforce affected via layoffs, role changes, or relocations."

Google DeepMind

Description: Google's AI research arm.
Why mentioned: Employee pushback on non-disparagement clauses.
Quote: "employees in Britain are pushing the company to remove lifetime non-disparagement clauses from their contracts, arguing that the restrictions could deter researchers from publicly warning about AI risks."

Madison Square Garden

Description: Entertainment venue operator.
Why mentioned: Facial-recognition surveillance controversy.
Quote: "WIRED has previously reported that MSG uses face-scanning tools to track visitors across and outside its venues."

The House Fund / Evertrue Capital / Robin Capital

Description: New venture funds.
Why mentioned: New-fund signals: campus-focused ($52.1M), emerging-manager fintech ($17M), and European pre-seed/seed ($15M).
Quote: "has raised a €15 million second fund to back European startups at pre-seed and seed across enterprise AI, proprietary data, robotics, space, and other defensible technologies."

4. People Identified

Elad Gil

Description: Investor.
Why mentioned: Led Applied Compute's $350M round.
Quote: "The deal was led by Elad Gil, with additional support from Kleiner Perkins, Lux Capital..."

Dan Driscoll

Description: Former U.S. Army Secretary.
Why mentioned: Hired by OpenAI for national security and defense AI expertise.
Quote: "OpenAI has hired former U.S. Army Secretary Dan Driscoll to advise on national-security modernization, cybersecurity, and the use of AI in defense."

Jill Hazelbaker

Description: Uber president, now OpenAI's chief corporate affairs officer.
Why mentioned: Senior comms and policy hire as OpenAI deepens government ties.
Quote: "will oversee global communications and corporate positioning. She previously held senior communications and policy roles at Snapchat and Google."

Zohran Mamdani

Description: New York City Mayor.
Why mentioned: Backing biometric-surveillance restrictions.
Quote: "criticized Madison Square Garden owner James Dolan's use of facial-recognition and other surveillance technology, backing proposed city legislation that would restrict the collection of biometric data in public places."

James Dolan

Description: Madison Square Garden owner.
Why mentioned: Target of the surveillance criticism above.
Quote: See above.

Yvonne Bajela

Description: Founder of Evertrue Capital, former Phoenix Court partner.
Why mentioned: Track record backing notable fintech companies.
Quote: "Bajela previously backed companies including Wise, Moove, and Marshmallow."

Robin Haak

Description: Solo GP at Robin Capital.
Why mentioned: Raised a second fund focused on defensible European technologies.
Quote: "led by solo GP Robin Haak, has raised a €15 million second fund."

Markus Persson

Description: Minecraft creator.
Why mentioned: Listing a $70M Beverly Hills mansion.
Quote: "Minecraft creator Markus Persson is asking $70 million for his 22,300-square-foot mansion in Beverly Hills' tony Trousdale Estates enclave."

5. Operating Insights

Treat known threat tactics as operational risk, not background noise The Binance case shows that recruiter-themed malware lures, a tactic the industry had warned about for years, can still succeed against a company that had already been targeted dozens of times. Operators should run recurring, role-specific training and controls for high-privilege staff (such as wallet and infrastructure engineers).

"...tricking a wallet engineer into downloading malware despite years of industry warnings about the tactic and at least 57 suspected North Korean attacks against the company itself."

Take internal security dissent seriously, including physical security OpenAI's experience shows the cost of letting competitive urgency override repeated employee warnings. Create a channel where security concerns escalate to the board, and extend controls to physical spaces.

"...some concluding that management prioritized keeping pace with rivals over tightening defenses. The concerns gained new urgency after an OpenAI engineer was charged last month with planting a hidden recording device in an executive conference room."

Sell data control and embedded engineering to enterprises, but structure contracts for durability Palantir's wins come from combining security guarantees with forward-deployed engineers. The caution is that fast adoption through trials may not convert to long-term commitments, so build a path from pilot to multi-year deployment.

"Palantir has long offered what is today known as 'forward-deployed' software engineering, in which its own engineers work directly with customers to customize the software." / "...they are often signing short-term trials."

6. Overlooked Insights

Fal's valuation doubled in about seven months This is easy to miss amid the large headline numbers, but it signals how quickly late-stage AI infrastructure pricing is moving.

"...raised a $1.3 billion round at a $16 billion valuation, double the valuation it commanded in April."

Google says it has dropped the DeepMind non-disparagement clause for new hires but not for existing staff contracts The company's carve-out leaves ambiguity for current employees, which matters for anyone assessing AI-safety disclosure norms and talent retention.

"Google says it has dropped the post-employment clause for new hires and does not enforce it against former employees."