Claude Opus 4.6, launched by Anthropic on February 5, 2026, arrived in the middle of a rout in software stocks. A week later, Anthropic raised $30 billion. The Claude Opus impact split in two: the market sold the risk, professionals bought the gain, and both reactions are rational. At Fast Growth Advisors, we draw one lesson from it for marketing teams: the more powerful the model, the more the framing makes the difference.
What does Claude Opus 4.6 actually bring?
Three advances: agent teams working in parallel, a one-million-token context in beta, and top-tier results on high-value tasks.
With Agent Teams, first offered as a research preview in Claude Code, you can spin up several agents that work in parallel and coordinate autonomously, according to Anthropic’s announcement. Scott White, Anthropic’s Head of Product, compares the capability to assembling a capable workforce in which each member owns a piece of the work (TechCrunch).
No external orchestration required.
On the memory side, Opus 4.6 is the first Opus-class model with a one-million-token context window. An entire codebase, a stack of contracts or dozens of research papers fit in a single request.
Above all, the model improves where economic value is highest. GDPval-AA evaluates knowledge work tasks in finance, legal and other domains: Opus 4.6 beats GPT-5.2 there by around 144 Elo points. On Terminal-Bench 2.0, which measures agentic coding, it posts the highest score. In cybersecurity, Anthropic’s team says it has found and validated more than 500 high-severity vulnerabilities in open-source code with the model (Anthropic, February 2026).
At Fast Growth Advisors, where AI agents work on editorial production with NOMO IA, that last line is the one that stands out.
The progress that matters here is not the prettier sentence. It is the more complete task.
Why did the market sell off software stocks?
Because investors see Anthropic’s tools as a threat to per-seat licensing, and the selling had started even before Opus 4.6.
Tuesday, February 3, 2026: Anthropic unveils a productivity tool for in-house lawyers, a Claude Cowork plugin. According to Bloomberg, as quoted by TechStartups, it sparks a $285 billion rout across software, legal information and financial services stocks. Thomson Reuters loses 16% in the session.
“Get me out”: that is how Jeffrey Favuzza, of Jefferies’ equity trading desk, sums up the style of selling that day (Bloomberg, February 3, 2026).
By the next day, the iShares Expanded Tech-Software exchange-traded fund (ETF) is down nearly 4%, taking its loss since the start of 2026 to 22% (Business Insider, February 4, 2026). On February 5, the day Opus 4.6 launches, the sector has lost about $830 billion in market value since January 28 (Nasdaq).
For sellers, the reasoning is straightforward. If ten AI agents replace the work of a hundred salespeople, a company no longer needs a hundred Salesforce licences, and the per-seat revenue model of software as a service (SaaS) starts to wobble.
Not everyone agrees.
Jensen Huang, Nvidia’s chief executive, calls the idea that AI will replace software “the most illogical thing in the world”: AI uses software as its tools. So the market is not saying “everything will vanish”. For Fast Growth Advisors, it is saying something else: what used to be a product advantage is becoming an execution advantage.
Why did Anthropic raise $30 billion a week later?
Private investors read the same signals the other way round: where the stock market sees a risk for software vendors, they see revenue growing tenfold a year.
Anthropic announces on February 12, 2026 a $30 billion Series G, led by GIC (Government of Singapore Investment Corporation, Singapore’s sovereign wealth fund) and Coatue, at a $380 billion post-money valuation (Series G announcement). According to Crunchbase, it is the second-largest venture round of all time, behind OpenAI’s $40 billion in 2025. Microsoft and Nvidia take part as strategic investors.
The figures published that day explain the appetite.
Run-rate revenue: $14 billion, growing more than 10x a year in each of the past three years. Claude Code alone exceeds $2.5 billion in run-rate revenue, a figure that has more than doubled since the start of 2026. Eight of the Fortune 10 are customers. Two years ago, a dozen customers spent over $1 million a year with Anthropic; in February 2026, there are more than 500 (the trajectory is dizzying).
For Fortune, Anthropic now sits with OpenAI and SpaceX among the three biggest names that could go public in 2026. Two opposite reactions to the same event, then, and that is the reading Fast Growth Advisors takes from that February: both are rational.
How did professionals react to the Claude Opus impact?
Through adoption rather than selling: practitioners do not have a valuation-multiple reflex, they have a workflow reflex.
One term circulates from launch day: “vibe working” (Geo News, February 6, 2026). After “vibe coding”, which means building software by intent rather than syntax, the same logic extends to knowledge work: finance, law, research, analysis.
You define what you want. The agents execute it.
Law offers the clearest example. At Dentons Europe, chief technology officer Matej Jambrich explains that better model reasoning reduces rework and improves consistency, so lawyers can focus on higher-value judgement (Microsoft, February 5, 2026). Harvey reports 90.2% on BigLaw Bench, with 40% perfect scores. That kind of signal speaks to quality and stability, not hype.
Twelve days later, the capability goes mainstream.
Anthropic releases Claude Sonnet 4.6 on February 17, 2026, and it becomes the default model on claude.ai for Free and Pro plans. In testing, users preferred it to Opus 4.5, the November 2025 flagship, 59% of the time, at $3 per million input tokens against $5 for Opus (Sonnet 4.6 announcement). For the European companies Fast Growth Advisors works with, the implication is concrete: the capability level that required an Opus budget is now available in Sonnet.
What does Claude Opus 4.6 mean for marketing and growth teams?
That value shifts to what the model does not “own”: proprietary data and business context, integrations and compliance, orchestration.
First implication: agentic AI moves from demo to daily use. Agent Teams make it possible to parallelise complex workflows (research, structuring, production, validation), and sequential marketing tasks become simultaneous. But the tool alone is no longer enough.
Second, the SaaS business model is changing. If your stack includes fifteen per-seat tools, the consolidation question is on the table now. Agents do not yet replace all software, but investors had already started pricing in this transition by February 2026.
Then comes the decisive point.
Framing quality becomes the competitive advantage. As the model grows more powerful, the brief, the standards and human validation weigh more heavily on the output. A high-performing agent without editorial guardrails produces fast and badly. Inside a structured system, it produces fast and well, and that is exactly what we work on at Fast Growth Advisors with NOMO IA.
Yet all framing starts with the messaging, and that part is still fragile. Our measurements, drawn from the Message-Market Fit Observatory (second quarter of 2026), put a number on it: of 83 detailed audits scored out of 75 points, 75.9% fall short of the critical threshold of 37.5. Hand such a foundation to an agent and you simply produce, faster, a message that fails to convince.
The more AI can do, the more companies pay for control.
Your window to structure AI workflows is closing fast: companies with editorial frameworks and clear standards will absorb this acceleration, and the rest will be subjected to it.
Sources
Links checked on September 30, 2026. Market data dates from February 2026 and comes from the financial press, model characteristics and Anthropic’s figures come from its official announcements, and only the Observatory figures are Fast Growth Advisors’ own data.
- Anthropic, “Introducing Claude Opus 4.6” (February 5, 2026)
- Anthropic, research on zero-day vulnerabilities (February 5, 2026)
- TechCrunch, “Anthropic releases Opus 4.6 with new agent teams” (February 5, 2026)
- Microsoft Azure, Claude Opus 4.6 in Microsoft Foundry (February 5, 2026)
- Bloomberg (via The Wealth Advisor), “‘Get Me Out’: Traders Dump Software Stocks As AI Fears Erupt” (February 3, 2026)
- Business Insider (Yahoo Finance), Jensen Huang on the software selloff (February 4, 2026)
- TechStartups, Claude plugins and the $285 billion selloff (February 5, 2026)
- Nasdaq, “ETFs to Tackle the Anthropic-Led Software Stock Rout” (February 5, 2026)
- Geo News, “vibe working” and Claude Opus 4.6 (February 6, 2026)
- Anthropic, Series G announcement (February 12, 2026)
- Crunchbase News, second-largest round of all time (February 12, 2026)
- Fortune, $380 billion valuation and initial public offering candidates (February 13, 2026)
- Anthropic, “Introducing Claude Sonnet 4.6” (February 17, 2026)
FAQ
What exactly is Opus 4.6?
Anthropic’s Opus model, launched on February 5, 2026.
Claude Opus 4.6 introduces Agent Teams, AI agents that work in parallel, and a one-million-token context window in beta. It posts the highest score on Terminal-Bench 2.0, an agentic coding evaluation.
Why did the stock market drop after the announcement?
The drop had started two days earlier.
On February 3, 2026, a legal plugin for Claude Cowork sparked a $285 billion rout, according to Bloomberg. Investors fear for per-seat licences as AI agents absorb more of the work, and the Opus 4.6 launch reinforced that fear.
Which companies are seen as most exposed?
Those selling a faster deliverable, with no durable advantage.
Proprietary data, deep integrations, regulatory constraints or high switching costs protect a vendor. In February 2026, legal and professional information was hit first: Thomson Reuters lost 16% in a single session.
How did professionals react?
Through adoption and concrete testing.
Harvey reported 90.2% on BigLaw Bench with Claude Opus 4.6, including 40% perfect scores. At Dentons Europe, chief technology officer Matej Jambrich credits the reasoning with reducing rework and improving consistency, which frees lawyers for higher-value judgement.
Does this mean the end of SaaS?
No. It means the end of some easy rents.
A vendor that owns proprietary data, business workflows and real operational accountability is more defensible than one selling an interface over work that has become automatable. AI needs software to run.
What is the relationship between Opus 4.6 and Sonnet 4.6?
Sonnet 4.6 puts these capabilities within everyone’s reach.
Launched on February 17, 2026, Sonnet 4.6 becomes the default model for Free and Pro plans. Users preferred it to Opus 4.5 59% of the time in testing, at $3 per million input tokens.
What should marketing teams take away?
That framing now matters more than the tool.
Models are advancing faster than organisations’ ability to govern them. For Fast Growth Advisors, the difference lies in editorial framing, quality standards and messaging clear enough for an agent to reuse it faithfully.