1. Anthropic’s $1B to $19B growth run: how Claude became the fastest-growing AI product in history — Lenny's Newsletter

  • Why read: A rare inside look at how Anthropic scaled ARR by 19x in 14 months using autonomous growth experiments.
  • Summary: Anthropic's Head of Growth, Amol Avasare, details their unprecedented trajectory, driven largely by an internal AI tool called "CASH" that automates growth experiments. The team indexes heavily (70/30) toward high-leverage big bets rather than incremental tweaks. Avasare argues that activation is the most critical growth lever in AI, as the technology fundamentally shifts the ratio of product managers to engineers. Their focus on AI coding not only grew the product but created a research flywheel that accelerated underlying model development.
  • Link: https://www.lennysnewsletter.com/p/anthropics-1b-to-19b-growth-run

2. The protégé problem today — David Hoang

  • Why read: A sharp analysis of how the AI transition is disrupting traditional mentorship and career progression in tech.
  • Summary: The shift toward multi-modal AI is squeezing mid-career professionals, creating a unique dynamic where traditional mentors lack foundational skills in new AI toolchains. This "protégé problem" means junior talent often outpaces their seniors in raw technical execution and modern workflows. Software development is fundamentally changing, with massive implications for how companies structure teams and how leaders remain relevant. Staying ahead requires actively adapting to new interface paradigms rather than waiting for the dust to settle.
  • Link: https://www.proofofconcept.pub/p/the-protege-problem-today

3. Grow or Die — SaaStr

  • Why read: A stark wake-up call for SaaS founders about the massive reallocation of IT budgets toward AI.
  • Summary: While overall IT spending is projected to cross $6 trillion in 2026, the growth is highly concentrated, with AI spending surging by 44% year-over-year. This effectively siphons budget away from legacy software categories as CIOs consolidate their spend into high-ROI, AI-native winners. Markets are severely punishing non-growth companies, valuing private AI firms at massive premiums (61x ARR) while traditional public SaaS languishes (4x ARR). Getting profitable isn't enough; if a company isn't actively capturing the AI budget in its space, it is slowly dying.
  • Link: https://www.saastr.com/grow-or-die/

4. $18B in pharma deals and AI 40% trade growth — Chamath Palihapitiya

  • Why read: Highlights massive macroeconomic shifts driven by AI hardware and accelerated biotech M&A.
  • Summary: AI-related trade, including semiconductors and data-center hardware, grew by nearly 40% in 2025, accounting for a massive 30% of total global trade growth. This explosion is reshaping global supply chains, with countries like Vietnam and India absorbing displaced US-China trade. Concurrently, the pharma industry is seeing a massive wave of M&A, driven by companies buying up AI drug-discovery platforms and near-term assets to combat impending patent cliffs. These multi-billion dollar deals signal a profound reallocation of capital toward AI-enabled biotechnology.
  • Link: https://chamath.substack.com/p/what-i-read-this-week-178

5. #278 | Request for Startups, Infra Roadmap, & more — Ali Afridi

  • Why read: A curated snapshot of what top venture capitalists are prioritizing and funding in 2026.
  • Summary: This newsletter aggregates investment theses from top VC firms, highlighting key frontier areas for the coming year. Major themes include the rise of "Deployed Intelligence Companies" and comprehensive AI infrastructure roadmaps. It also points to massive technological whitespace in physical industries, such as brick-and-mortar retail and industrial resilience. The consensus indicates that investors have moved past superficial AI wrappers and are looking for foundational infrastructure and deep vertical disruption.
  • Link: https://www.sandhill.io/p/278-request-for-startups-infra-roadmap

6. On Cooling America Out — Contraptions

  • Why read: A fascinating sociological lens on the current American identity crisis using Erving Goffman's theory of the "long con."
  • Summary: Applying Goffman's 1952 paper "On Cooling the Mark Out," this essay argues that America is experiencing the psychological fallout of being the "mark" in its own historical con. The core damage of a con isn't just material loss, but the destruction of a deeply held self-image and the resulting social humiliation. The mechanisms meant to "cool out" the populace and help them accept their new reality are currently failing. This provides a profound psychological framework for understanding widespread cultural and political instability.
  • Link: https://contraptions.venkateshrao.com/p/on-cooling-america-out

7. The Only Productivity System You Actually Need — Regina Gerbeaux

  • Why read: A grounding reminder that tooling and software can't fix a lack of foundational alignment.
  • Summary: Executives constantly seek the perfect app, morning routine, or time-blocking method to optimize their output. However, implementing a productivity system without first defining core values is just a visually appealing waste of time. True productivity isn't about the mechanics of a color-coded calendar; it’s about aligning your daily actions with what you genuinely care about optimizing for. Establishing this bedrock of personal values is the only way to make any operational system stick.
  • Link: https://force-multipliers.beehiiv.com/p/the-only-productivity-system-you-actually-need

8. Vertical AI founders, don't worry - Claude Cowork is not coming for your lunch. YET. — Sathya

  • Why read: A critical framework for vertical AI founders on why they must ignore their customers' existing processes.
  • Summary: Frontier models optimize for general intelligence, but true enterprise transformation requires deep vertical integration and an outside-in approach. When talking to customers, founders must ruthlessly separate the underlying organizational pain from the current process. Existing processes were designed around human constraints that AI makes obsolete, so building a product around them means building on an eroding foundation. Instead, vertical AI must impose a better, autonomous workflow that eliminates those constraints entirely.
  • Link: https://twitter.com/sathyanellore/status/2040550670575792441/?rw_tt_thread=True

9. Context Graphs: an IAM Problem at Scale — Allie Howe

  • Why read: Explains the technical infrastructure required for AI agents to truly learn from enterprise decisions.
  • Summary: Context graphs represent the future of enterprise AI, moving beyond simple retrieval to storing the reasoning and negotiation behind business decisions. By capturing fragmented context from Slack, CRMs, and codebases, agents can build a predictive world model of an organization. However, this creates a massive Identity and Access Management (IAM) challenge. Agents need deep access to discover context, while enterprises need strict controls on who can retrieve these shared insights; solving this IAM bottleneck is the key to unlocking true autonomy.
  • Link: https://twitter.com/vtahowe/status/2040832207087194448/?rw_tt_thread=True

10. One of the core things we’re going to have to... — Aaron Levie

  • Why read: A fundamental truth about why the "context layer" will always be the most important part of enterprise AI.
  • Summary: Even the most advanced frontier models cannot possess all the relevant knowledge needed for enterprise use cases. Every organization—and every user within it—has entirely different workflows and access permissions for corporate data. Continual learning strictly at the model weight level is nearly impossible in enterprises due to strict data silos and catastrophic forgetting. Therefore, the "harness" and the "context layer" are where the actual magic happens, turning general models into highly specialized, secure agents.
  • Link: https://twitter.com/levie/status/2040650799550722243/?rw_tt_thread=True

Themes from yesterday

  • The Bifurcation of Enterprise AI: IT budgets are concentrating entirely around AI-native winners, punishing traditional SaaS companies and forcing a "grow or die" paradigm.
  • Agentic Infrastructure over General Intelligence: True enterprise value is shifting from frontier models to "context graphs" and IAM frameworks that securely capture organizational decision-making.
  • The Obsolescence of Legacy Workflows: Vertical AI is moving away from digitizing existing human processes, instead fundamentally restructuring how work gets done by removing historical constraints.
  • The Protégé Problem: The rapid adoption of multi-modal AI is disrupting traditional career ladders, leaving mid-career tech workers struggling to adapt while junior talent commands new toolchains.