Bold Partnerships Fueling Anthropic’s AI Growth
Last week, Anthropic PBC made an announcement that went largely unnoticed. The company, having ignited selloffs in the equity market through a continuous flow of product updates, aimed to prevent further unsettling of investors. The livestream was not broadly publicized; there was an absence of a related blog post. However, the launch was as significant as any recent offering from the artificial intelligence lab. The company, which has made significant strides in the enterprise sector with “agents” that perform various tasks, has announced new updates to its tool Claude Cowork, enhancing its capabilities in fields such as human resources, investment banking, and design. To achieve this, Anthropic collaborated with various companies to enable those capabilities via what are known as plug-ins. A plug-in functions as a mini-application within Claude Cowork, enhancing the AI tool by introducing a new capability, such as financial analysis. According to Anthropic, FactSet co-developed three plug-ins for Claude Cowork that perform tasks such as drafting research notes, comparing consensus estimates, and updating financial models.
This is beneficial for Claude — but what advantages does it offer to FactSet or these other third parties? Anthropic merely hints that integrating its AI will enhance the appeal of partner products. “The more connected Claude is to an organisation’s existing software, the more useful it becomes — and the more valuable those tools become in turn,” Anthropic states in its press release. However, this still does not present a financially viable option for Anthropic’s partners, who are offering their expertise through detailed instructions that enable an AI tool to mimic certain aspects of their work, such as the methods equity analysts use to compile corporate reports or the processes human-resources staff follow to create performance summaries. The motivation for companies like FactSet and Thomson Reuters Corp., another partner, is straightforward: If Claude is set to transform the dynamics of their industry, they prefer to be part of the process rather than merely observing from the sidelines.
However, for the time being, the arrangement resonates with the partnerships that Facebook established with publishers nearly a decade ago, where news articles were shared freely with the social network in exchange for increased traffic. In 2018, when Facebook altered its algorithm to deprioritize news, the influx of new traffic disappeared entirely, leaving publishers with a failed investment and advertising ventures that had diminished during that period. Anthropic asserts that its arrangement differs significantly from Facebook’s, emphasizing that its plug-ins provide “bidirectional value” and enhance the utility and engagement of its partners’ products. Providing a plug-in for Cowork comes at no cost. Nonetheless, the tangible benefits for participants offering their expertise remain ambiguous, particularly as Anthropic reaps all the financial rewards via subscriptions to Claude Cowork, which range from tens to hundreds of dollars monthly, based on usage, for each “seat” allocated to corporate teams. Let us extend a measure of trust to Anthropic for the time being. What if it were establishing the foundation for a more significant future financial opportunity for its partners by, for instance, sharing revenue? Such a scenario would not be unprecedented in the evolving landscape of AI models; for instance, OpenAI receives a portion of the revenue generated by Microsoft Corp. from the sale of subscriptions to Copilot, as it supplies the foundational AI system for that product.
A more effective and ambitious strategy for Anthropic would involve establishing a marketplace where partners receive a commission based on the usage of their applications. Such a move would establish a financial incentive that draws in additional partners and possibly increases the subscriber base, reminiscent of the network effect initiated by Apple Inc. with the launch of its App Store in 2008. It would expand Anthropic’s revenue streams beyond solely subscription fees. It would also be a constructive move toward addressing what tech media publisher Tim O’Reilly refers to as “the missing half of the AI economy.” Motivated by the aggressive growth strategies of Silicon Valley and the competitive landscape, leading AI laboratories have concentrated on capitalizing on their models exclusively for their own benefit through avenues such as subscriptions or the sale of API tokens. The partners contributing to Claude’s increased value are receiving minimal compensation in return. “Market-making is the best way to justify all those capital expenses,” O’Reilly states, referencing the billions allocated for chips and data centres by AI companies. He highlights that Alphabet Inc.’s Google was the pioneer in developing pay-per-click advertising, whereas YouTube implemented ContentID, a tool designed to detect copyrighted music, along with revenue sharing to address conflicts with the music industry. Both cases resulted in innovative opportunities for others to generate income.
No significant AI laboratory has accomplished that as of now. Anthropic has emerged as a prominent player, making waves in the markets and taking a stand against the US Department of Defence regarding potentially questionable applications of its technology. The latter has bestowed a noble sheen upon the company recently, yet leveraging the expertise of other firms could yield a contrasting effect, particularly if it exacerbates the white-collar jobs bloodbath that Anthropic’s CEO foresees. It might be prudent to maintain a degree of skepticism regarding all of these disruptive possibilities. Anthropic’s plug-ins might struggle, similar to how OpenAI’s “GPT Store” faltered in 2024. Sam Altman’s marketplace for custom-built ChatGPT agents struggled to find traction due to quality control issues and, unsurprisingly, a lack of opportunities for partners to generate revenue. Perhaps that is now a matter worthy of reconsideration.
