On 28 May Anthropic announced it had raised a landmark US$65 billion in Series H funding, bringing the artificial intelligence (AI) company to an eyewatering valuation of approximately US$965 billion, edging out rival OpenAI’s valuation and demonstrating the power of storytelling.
What’s this got to do with storytelling, you ask?
Just weeks earlier, in April, Anthropic announced Claude Mythos Preview, a new AI model so “strikingly capable at computer security tasks” it was deemed too dangerous to release generally without first using it to help close security holes in the world’s “most critical software”.
As part of this effort, Anthropic launched Project Glasswing, a program that sees Anthropic work with select industry partners who have limited access to Claude Mythos Preview to find vulnerabilities across “the most systemically important software in the world.”
Accompanying the initial announcement of Claude Mythos Preview and Project Glasswing was a leviathan blog post, nearly 14,000 words long, telling readers all about the model’s capabilities without telling us much about the model itself at all.
“We’re limited in what we can report here,” the company insists, before going on to tell us that 99% of the vulnerabilities found by Claude Mythos have not yet been patched. The blog post finishes by invoking previous initiatives to deal with things like post-quantum cryptography, while suggesting that, this time, “the threat is not hypothetical. Advanced language models are here.”
Whether this is hyperbole or a reasonable response to an unanticipated threat, whether Claude Mythos turns out to be a legitimately game-changing AI model or an elaborate form of vapourware, the rhetoric used to discuss it is a grand display of vision storytelling.
Simply put, a vision story describes a desired future state. For example, Anthropic’s claim that Project Glasswing (and Claude Mythos) will be a key first step to giving defenders a “durable advantage in the coming AI-driven era of cybersecurity.”
These vision stories represent just one type of storytelling model in the corporate grab-bag of messaging templates. Also in that bucket are tried-and-tested narrative frameworks such as:
Each story type has its own strengths and applications.
Vision stories work to enhance and influence the audience’s view of the future and, by default, the role of the organisation telling the story in creating that future. They are particularly effective at anchoring a future-state vision to something in the present. In this way, they effectively pull hypothetical future outcomes or value toward the present in the mind of the audience.
Elon Musk is a notable practitioner of vision storytelling. For years he’s been touting the goal of seeing humans become a multiplanet species and articulating a vision of permanent settlements on Mars. Despite a continued lack of commercial imperative for such outcomes, these stories have captured the public’s imagination, even while rockets were exploding on the launchpad.
Vision stories appeal to customers. They also appeal to investors. Musk’s vision stories have no doubt helped SpaceX build its onramp to the blockbuster IPO that’s expected to result in a valuation of between US$1.75 trillion and US$2 trillion (at the time of writing). Now, Anthropic is doing the same, with the company eyeing up its own blockbuster IPO.
It is, in part, the power of vision storytelling to anchor future outcomes in the present that allows companies like SpaceX to even consider the astronomical implied IPO price-to-revenue multiple of over 90x that it’s aiming for. Even Anthropic’s far more reasonable implied price-to-revenue multiple of 20x or more following its latest capital-raising effort is an exceptional figure.
Not a bad outcome if the investor confidence in that 20x figure is influenced in any way by news of the new AI model that Anthropic believes could “reshape cybersecurity”. Investors swayed by Anthropic’s vision story in this context are betting on something many are yet to see for themselves. But then, that’s the nature of vision narratives.
The vision narratives at the heart of companies like Anthropic and SpaceX are, importantly, not about Anthropic or SpaceX. They are not about Elon Musk or Anthropic CEO Dario Amodei. Musk’s story is about all of humanity. Anthropic’s story is about the software that runs our world.
Effective vision stories are about the reader, even if they embody the goals or aspirations of a brand and its leader. This is true for many narrative frameworks. They also tend to have one foot in the present and the other in the promised land of their future state vision, the vivid picture they paint of what the outcomes could be if they achieve their goals.
In the words of Anthropic: “In the long run, we expect that defence capabilities will dominate: that the world will emerge more secure, with software better hardened—in large part by code written by these models. But the transitional period will be fraught. We therefore need to begin taking action now.”
As this example also demonstrates, the best vision stories also acknowledge the obstacles on the path to the future. It will not be an easy journey, they suggest, but stick with us and we’ll get you there. Importantly, vision stories also evolve, changing over time to prevent stagnation and shift with market dynamics. This helps to maintain energy and urgency over time.
The trick is to make the case for taking action now, as Anthropic tells the security community in its gargantuan blog about Claude Mythos Preview. Our future-state depends on your current state, the message insinuates.
“There’s no denying that this is going to be a difficult time,” Anthropic warns, while also making the case that “Mythos Preview is only the beginning”.
And this, perhaps, is the crux of the vision story: the future starts here, and you get to choose whether you come with us or not.
You don’t want to make the wrong choice, right?