Claude Just Killed Software Stocks (Here's What Happens Next)

The SAS Apocalypse: A Deep Dive into the AI-Driven Software Sell-Off

Key Concepts:

• SAS Apocalypse: The significant decline in Software as a Service (SaaS) stock values triggered by advancements in Agentic AI.
• Agentic AI: AI systems capable of autonomously performing tasks, including coding, analysis, and problem-solving, reducing the need for human intervention.
• Needle in a Haystack Retrieval: A key capability of advanced AI models (like Anthropic’s Opus 4.6) allowing them to efficiently locate and utilize relevant information from vast datasets.
• Forward Price to Sales (P/S) Multiple: A valuation metric comparing a company’s stock price to its revenue, used to assess market expectations for growth.
• Hyperscalers: Large-scale cloud providers (Amazon, Microsoft, Google) offering extensive computing resources and services.

I. The Trigger: From Legal Plug-in to Market Meltdown

The recent downturn in software stocks, dubbed the “SAS apocalypse,” wasn’t a sudden event but a reaction to a specific catalyst: Anthropic’s January 30th release of a simple, 200-line open-source plug-in for Claude Co-work. This plug-in enabled Claude to perform routine legal tasks – contract review, NDA analysis, compliance summaries – traditionally handled by junior lawyers and paralegals using expensive platforms like Westlaw and LexisNexis.

Within days, SaaS companies collectively lost nearly $300 billion in market capitalization, impacting major players like Adobe, Salesforce, ServiceNow, HubSpot, and Intuit. The core issue isn’t software being obsolete, but the realization that AI can drastically reduce the cost and manpower required for services previously reliant on human labor.

This realization prompted KPMG, a Big Four accounting firm, to leverage AI as bargaining power, securing a 14% reduction in auditing fees from Grant Thornton UK. This sets a precedent: clients will renegotiate entire contracts when AI demonstrates cost savings, challenging the traditional “pay-per-seat” and “pay-per-billable-hour” models.

II. Agentic AI Breakthroughs: Beyond Routine Tasks

The legal plug-in was just the beginning. Recent advancements in Agentic AI demonstrate capabilities far exceeding simple task automation.

• Autonomous Software Development: Anthropic conducted an experiment using 16 Claude Opus 4.6 agents to build a C compiler in Rust. Over two weeks, the agents generated approximately 100,000 lines of code capable of running on mainstream operating systems, handling databases and video tools, and passing standard stress tests – all for around $20,000. This contrasts sharply with the estimated $1 million+ cost and year-long timeframe for a human team.
• Needle in a Haystack Retrieval: The success of these agents hinges on “needle in a haystack retrieval,” specifically Opus 4.6’s ability to scan a million tokens of text and accurately extract relevant information 76% of the time – three times better than competing models. This allows the AI to reason about complex codebases with the understanding of a senior engineer.
• Real-World Application – Rakuten: Rakuten integrated Claude Opus 4.6 into their engineering issue tracker, resulting in the AI autonomously closing 13 tickets and correctly assigning another 12 to developers across six codebases in a single day. This included writing, testing, and deploying code.
• Teamwork & Security: Anthropic’s “teams of agents” feature allows for project breakdown, task assignment, and collaboration, mimicking a small software company. Furthermore, Opus 4.6 identified over 500 previously unknown security vulnerabilities in open-source software without specific instructions, replicating the work of expensive security consultants.

III. Market Dynamics & Future Growth

The global artificial intelligence market is projected to grow nearly 19x over the next nine years, with a compound annual growth rate of 38.5% through 2034 (according to Market US). However, many next-generation AI companies are remaining private for longer periods (averaging 10+ years before IPO), potentially limiting investor access to early-stage gains.

Fundrise is presented as a solution, offering venture capital access to pre-IPO AI companies with a low entry point of $10 and a track record of investing almost $400 million in AI and data infrastructure.

IV. Identifying Vulnerable & Resilient Companies

The market is shifting away from valuing software based on “pay-per-seat” pricing. Companies most at risk share these characteristics:

• Bundled, Commoditized Features: Offering a wide range of basic features behind a slick interface.
• Per-Seat Pricing: Charging based on the number of users.
• User Growth Focus: Prioritizing adding users over increasing value per user.

Examples of vulnerable companies include Salesforce, ServiceNow, HubSpot, Monday.com, and LegalZoom. Broad software indexes are down around 15%, and SAS-focused funds have declined by over 20% year-to-date. Forward P/S multiples have fallen from 9x to 6x, levels not seen in a decade. A hypothetical scenario illustrates the impact: a 20% price cut, 20% seat reduction, and P/S multiple decline could result in a 57% stock price drop.

Companies less at risk are those where the core work happens on their platform, or where the critical data resides:

• Adobe: Creative Suite (Photoshop, Premiere, Firefly) integrates generative AI directly into core workflows.
• Figma: A collaborative design platform serving as a central hub for product teams.
• Palantier: Foundry and AIP platforms act as operating systems for enterprise data, enabling AI workflow integration.

V. Investment Opportunities: Riding the AI Wave

The speaker advocates for investing in the infrastructure supporting Agentic AI:

• Semiconductors:
  • Nvidia (NVDA): Dominates the AI GPU market with over 90% share, benefiting from increased demand as agents scale.
  • AMD (AMD): Provides a competitive alternative to Nvidia, offering cloud providers pricing power and supply chain diversification.
  • Broadcom (AVGO): Focuses on networking chips and ASICs crucial for AI clusters.
  • Samsung, SKH Highix, Micron: Benefit from the surge in demand for high-bandwidth memory.
  • TSMC (TSM): The leading foundry manufacturing chips for Nvidia, AMD, and Broadcom.
• Cloud Infrastructure (Hyperscalers):
  • Amazon (AWS), Microsoft (Azure), Google (GCP): Control approximately 2/3 of the cloud infrastructure market and offer their own AI stacks.
• AI-Focused Data Centers:
  • Coreweave, Nebius, Iron: Provide dedicated AI compute capacity.
• Data Center Hardware:
  • Verdive (VRT): Supplies power systems, cooling racks, and other essential hardware for AI data centers.
• AI-First Software:
  • Palantier (PLTR): Offers platforms (Foundry, AIP) for integrating AI into complex organizations.
  • Crowdstrike (CRWD): Evolves into an AI-powered security system.

Conclusion:

The “SAS apocalypse” is not a temporary setback but a fundamental shift driven by Agentic AI. The market is recognizing that traditional software pricing models are unsustainable in a world of automation and personalized software. Investors who understand this shift and focus on the underlying infrastructure and AI-first software platforms are positioned to benefit significantly. The key is to invest in companies that enable AI, rather than those threatened by it.