Print Icon
 

February 28, 2026

Anthropic is:
a) A SaaS killer
b) A disruptor of multiple industries with its Claude large language model and the agentic apps built around it
c) An ingredient brand for enterprise software
d) Already part of the enterprise technology establishment via a well-worn SaaS playbook
e) All of the above


The correct answer is "e," as multiple things about Anthropic can be true at the same time. Anthropic has been portrayed as a SaaS assassin. Anthropic launches connectors and plugins to various enterprise apps, and SaaS stocks tank. Anthropic doubles down on Claude Code and Cowork, and suddenly, we're all expected to generate ERP, HCM, and ERP systems on the fly. Anthropic rolls out security code checks, and cybersecurity stocks stumble. Anthropic talks about modernizing COBOL, and IBM shares collapse. 


By next week, Anthropic will be making us dinner, collapsing restaurant shares, brewing beer on the fly, and building houses with robots that run Claude as their brain. 


Simply put, we're nearing peak Anthropic hype if we're not there already. What's lost in all of these storylines is the nuance. After all, nuance isn't as much fun. Here's a nuanced view of Anthropic because multiple things are true about the LLM giant at once. 


Anthropic isn't killing SaaS. 


What?!? Of course, Anthropic is killing SaaS, fat margins, and the existing state of enterprise software. Do we know how Anthropic will affect enterprise software? No. That's why SaaS stocks are a hard sell. There's a lot of uncertainty ahead. 


Not surprisingly, SaaS CEOs have made their cases. Speaking on Workday's fourth quarter earnings call, CEO Aneel Bhusri said:


"You've all heard the narrative out there that HR and ERP will be replaced or relegated to the background by AI. I personally just don't see that happening. Our application domains are really, really hard to build. I've been working in the HR and ERP space for over 30 years. These are true systems of record that must process transactions with absolute accuracy and speed, enforce complex security models, and comply with statutory and regulatory requirements worldwide. That kind of complexity is very hard to replicate. No amount of vibe coding is going to produce an HR or an ERP system."


Bhusri's argument is that business processes are deterministic and AI is probabilistic. The win is to marry AI and enterprise apps to redefine user experience, improve business processes by melding humans and AI agents, and drive insights. 


If anything, Anthropic's Claude could become the interface and door to enterprise applications and systems of record. The issue is that SaaS vendors often use Anthropic as an ingredient brand to retool their platforms. In that world, Anthropic is more of a SaaS enabler than a killer. 


Hereโ€™s a look at where Claude is being used.

Anthropic is an ingredient brand. 


The number of enterprise software companies working with Anthropic is vast. At a recent enterprise agent event where Anthropic announced new features and industry extensions for Claude Cowork, executives touted partnerships with SaaS vendors and outlined how models are embedded in the background. 


Anthropic launched Model Context Protocol (MCP), quickly stood up an industry standard, and now serves as the glue of agentic AI. 


In addition, Claude is available on all three large hyperscalers. Simply put, Anthropic is everywhere. Ingredient brands generally don't destroy the partners that are embedding them. 


Salesforce CEO Marc Benioff went with the apps, agents, and models are better together theme. Benioff said Salesforceโ€™s Agentforce platform uses multiple LLMs, including Anthropic, as part of the architecture. Now itโ€™s possible that these LLM players develop platforms, but today these vendors are more of an ingredient. 


โ€œOur job, as a software company, is to help our customers to create success and to take that and help them connect with their customers in a whole new way,โ€ said Benioff. โ€œWe'll deliver our products, our capabilities, our value proposition with our customer relationships. The current reality is about humans and agents working together. Our job is to take what's available today and make it successful.โ€


The list of SaaS vendors that count Anthropic as an ingredient includes ServiceNow, Salesforce, Snowflake, IBM, DocuSign, Microsoft, Palantir, Intuit, and a host of others. 


Anthropic is a SaaS company. 


Anthropic has followed and executed the SaaS playbook better than most SaaS companies. 

Consider:

  • Much of Anthropic's revenue is recurring, driven by seats and subscriptions. 
  • The company has rapidly expanded its partner network for sales throughput. 
  • Anthropic has systematically expanded beyond software development to financial services and healthcare. 
  • The company has hired executives who are experienced in building out ecosystems and broader platforms. 
  • Anthropic is a friend and foe for SaaS partners. One thing is clear: Anthropic's playbook rhymes with both SaaS friends and foes. 

One nuance is that Anthropic's business model relies on API connections and consumption. The model isn't per-user, but more per-outcome and per-token.  

Anthropic is a friend and foe (depending on the use case)


Anthropic is both a friend and foe to SaaS, and that nuance is often lost in play-by-play news and analysis. Consider:

  • Integration: Claude is a friend of SaaS because it serves as a natural language layer for data. Anthropic is a foe because Claude Cowork allows users to bypass SaaS interfaces and lower seat counts. 
  • Ecosystem: Anthropic needs SaaS data to ground its reasoning, so it's friendly with the enterprise software ecosystem. However, Anthropic's MCP standard makes it easier to bypass SaaS vendors. 
  • Innovation: Salesforce was cited as a key customer by Anthropic, which is used to enable Agentforce. However, Anthropic's specialized plugins for legal and HR could help it take market share in the SaaS market. 

Anthropic is playing for the thinking engine layer of the emerging AI stack. Sometimes, that thinking engine approach will make SaaS vendors better. Going forward, using Claude as a thinking engine can commoditize SaaS.

Anthropic's launch of new Cowork updates featured customers including Steve Hasker, CEO of Thomson Reuters, Sridhar Masam, CTO of NYSE, and Seth Hain, SVP of R&D for Epic. What's interesting about that customer panel is that Anthropic is an enterprise ingredient brand to those companies, yet also viewed as a rival to something like Thomson Reuters Co-Counsel.


Key themes from the customer panel.

  • Claude is being integrated internally. Masam said the NYSE has been using Claude Code to rewire "our engineering process with coding, writing tests, legacy codebases, refactoring documentation."
  • Masam added that the NYSE has also built internal AI agents based on Claude. "Claude models are particularly effective in processing large documents and applying roles," said Masam, who noted that the NYSE has built new AI-driven workflows for proxy filing, auditing, SEC filings, and news classification.
  • Technology is outpacing change management. Hasker said AI developments are outpacing change management. "A general counsel's office, a law firm, a tax accounting firm, or an audit firm needs to rewire the processes to be able to take advantage of the benefits that the tools provide," said Hasker. "And I think that work is ongoing, but I think it's 18 months away before that sort of change management catches up with the standard of the tool."
  • Claude can drive ideas from the bottom up. Epic's Hain said AI has shifted from being more assistive to becoming more of a collaborator. "As we're shifting into 2026, particularly with Claude 4.6 and Claude Code, they're able to do code exploration, compose, and have a draft of their support ticket and their response right out of the gate. Giving tools where employees can create on their own is a real way to open up opportunities," said Hain.
  • Trust is critical. Hasker said that Co-Counsel has more than 1 million users, and trust is a big reason. Anthropic's tools leave an audit trail, which is a big feature for regulated industries.


Where Anthropic is really headed.


Anthropic operates like an enterprise software company but has characteristics of an ingredient brand. Anthropic doesn't own the system of record but wants to own the reasoning and orchestration layer that sits across systems. 


Given that Anthropic is on all the cloud hyperscaler platforms, has deep integrator partnerships, and is developing its own enterprise surfaces, the company has the heft to pivot. It can be very disruptive yet also behave like an enterprise software establishment. 


In the end, Anthropic will help systems-of-record providers while taking away their user interfaces and potentially commoditizing them. Anthropic will be a foe to providers that are workflow players over other systems and use case-specific vendors. 


Ultimately, Anthropic is likely to be an ingredient brand with expanding enterprise SaaS ambitions and doesn't fit neatly into a category.  In the future, Anthropic likely become a platform. 

     

NOTEBOOK


๐Ÿ“’ For the latest on Dell, CoreWeave, and quantum, see Constellation Insights and my live Notebook.


๐Ÿ‘พ Salesforce has a new metric: the Agentic Work Unit (AWU). The big question is whether enterprises will adopt it. An AWU is one discrete task accomplished by an AI agent. These tasks can include decisions made, records updated, and workflows triggered. In the fourth quarter, Salesforce delivered 771 million AWUs. Hereโ€™s what you need to know. 


๐Ÿ“ˆ You'll likely hear a bit of consternation about the lack of pop in Nvidia shares following blowout fourth quarter earnings. Now I can quibble about inventory, investment gains from Nvidia's Intel stake, and a few other spots, but the reality is the company keeps delivering stellar results. Here's what's happening: There's no one left to buy Nvidia, and it doesn't matter what the company does. 


๐Ÿซฑ๐Ÿฝโ€๐Ÿซฒ๐Ÿพ AMD is investing $150 million in Nutanix shares at $36.26, as the two companies partner on enterprise AI. The two companies are pursuing a joint roadmap to integrate AMD ROCm and AMD Enterprise AI software into the Nutanix Cloud Platform and the Nutanix Kubernetes Platform, which will use EPYC CPUs and AMD Instinct GPUs. The deal equates to more distribution for Nutanix. AMD will also fund up to $100 million to support joint engineering and go-to-market efforts. Separately, Nutanix reported better-than-expected fiscal second-quarter results.


๐Ÿš€ ServiceNow launched Autonomous Workforce, an AI specialist on the Now Platform designed for specialized roles in service desk, employee service, and security operations. The company also launched EmployeeWorks, which combines Moveworksconversational AI and enterprise search with ServiceNow's workflows and governance.


๐Ÿ‘Ž๐Ÿผ Citrini Research said AI will absolutely tank the economy in a dystopian way.  CitidelSecurities called BS. 


๐Ÿ“ข OpenAI announced Frontier Alliances, which is a set of system integrators and consultants aimed at deploying "AI coworkers" across enterprises. The ironic โ€” somewhat comical โ€” reality is that large language model (LLM) giants like OpenAI and Anthropic are expected to upend SaaS incumbents, yet the playbook they're using isn't original in any way.


๐Ÿ”€ Accenture and Mistral AI said they will co-develop and deliver enterprise AI. The Accenture partnership gives Mistral AI more distribution, but the European LLM player already has multiple partners, including SAP and IBM. Accenture will become a customer of Mistral AI.


โ„๏ธ Snowflake reported better-than-expected fourth-quarter results and said it added 740 net new customers, up 40% from a year ago.


๐Ÿ’ฐ SAP customers are being more selective in their spending, moving slowly to S/4HANA and often deciding to extend maintenance for on-premises deployments, according to the German-speaking SAP User Group (DSAG) 2026 investment report. And if these SAP customers are pursuing AI, they are often going with non-SAP vendors.


๐Ÿš— Uber launched Uber Autonomous Solutions, and the upshot is that the company is looking to be a data and monetization layer for autonomous vehicles.


๐ŸŒช๏ธ Researchers took autonomous AI agents for a spin and found they could wreak havoc. The paper, Agents of Chaos, was penned by researchers from Harvard, MIT, Stanford, Carnegie Mellon, Northeastern University, and other institutions.


๐Ÿ”” Workday reported a solid fourth quarter, but its subscription outlook was light.


โŒ C3.ai is cutting 26% of its workforce and looking to cut non-employee costs by 30% by the second half of fiscal 2027. C3.ai said it is looking to cut $135 million in non-GAAP operating expenses. The restructuring plan was outlined in an SEC filing following disappointing third-quarter results and outlook.


๐Ÿซฑ๐Ÿฝโ€๐Ÿซฒ๐Ÿพ Pure Storage said it is changing its name to Everpure and acquiring 1touch, a data intelligence and orchestration company.


๐Ÿช™ Zoom reported fourth-quarter earnings of $2.22 a share on revenue of $1.25 billion, up 5.3% from a year ago. Non-GAAP earnings were $1.44 a share. The earnings results were a nickel a share below estimates.


๐Ÿ“ˆ HP reported better-than-expected first-quarter results with earnings of 58 cents a share on revenue of $14.4 billion, up 6.9% from a year ago. Non-GAAP earnings in the quarter were 81 cents a share, or 4 cents better than estimates. Bruce Broussard, Interim CEO at HP, said the company saw strong demand for AI PCs. Personal systems revenue was up 11%, with consumer up 16% and commercial up 9% from a year ago.


๐Ÿ”’ CrowdStrike, in its 2026 Global Threat Report, said the average e-crime breakout time fell to 29 minutes in 2025, with the fastest at 27 seconds. Adversaries are leveraging AI at a rapid clip, and nation-state attacks have surged.

     

Like getting our weekly round-up? Subscribe to Constellation Insights for consistent news and analysis on technology trends, earnings, enterprise updates and more.

Want to work with Constellation analysts? Our Sales team would love to get in touch!

You are receiving this email as you signed up for our newsletters.

Want to change how you receive these emails?

You can Unsubscribe or Update your preferences