Hello and welcome to Eye on AI. In this edition…Anthropic enters the enterprise AI game, OpenAI execs (and former execs) advance billion-dollar funding deals, the DOJ investigates Nvidia, and Elon Musk spreads election disinformation.
Anthropic yesterday announced it’s launching Claude for Enterprise, a new tier of its Claude chatbot aimed at large organizations. While the company’s Team plan launched earlier this year introduced features for collaboration (such as the ability to create projects and share chats across teammates), the Enterprise plan goes a step further by allowing businesses to work with their own proprietary information inside Claude. It also has the type of enterprise-grade controls and security features big businesses need to even consider using such a tool, including the ability to securely manage user access and the promise that the company won’t train AI models on customer data and interactions with Claude.
For the most part, the offering is about the same as ChatGPT Enterprise, OpenAI’s plan for enterprises that was launched a year ago. Both offer enterprises a more secure working environment and the ability to turn the chatbot into a company-specific assistant that can analyze the business’s own data, in addition to helping with general tasks like brainstorming and drafting content.
Enterprise is where the AI smart money is
It’s no surprise Anthropic is following suit by trying to court enterprises—as I’ve been reporting over recent weeks, enterprise AI is big business and turning out to be the most significant slice of the AI pie thus far. OpenAI said users of its Enterprise tier grew 4x in just a few months from 150,000 in January of this year to 600,000 in April. OpenAI is also gearing its investment and M&A strategies around enterprise AI, while at the same time, AI startups with B2B business models are scooping up the vast majority of VC deals compared to AI startups with B2C models. After all, enterprises have the most resources available to experiment with AI, especially as many shareholders continue to push for AI adoption.
Anthropic may be playing catch-up to OpenAI, but the increase in choices can only be good for customers and the overall state of enterprise AI. It will be interesting to see if customers of ChatGPT Enterprise that chose the chatbot because it was the only option at the time switch to Anthropic’s new offering, especially with the rival startup’s reputation for having a safer approach to AI.
Anthropic wants to compete on context length
Anthropic is offering users a few reasons to pick its product over ChatGPT. First, there’s the context window, referring to the number of tokens (words or parts of words) a model can handle at once. Anthropic for Enterprise boasts a 500,000 token context window, which the company says is “equivalent to hundreds of sales transcripts, dozens of 100+ page documents, or medium-sized codebases.” This is more than double that of ChatGPT Enterprise, which offers a 128,000 context window.
Anthropic’s Enterprise tier also introduces a native GitHub integration for syncing GitHub repositories with Claude. This gives the model direct access to customers’ codebases so their engineering teams can tap the chatbot for assistance with debugging issues, iterating on new features, or onboarding new engineers.
The plan is available now, but one detail that’s not clear is pricing. An Anthropic spokesperson tells Eye on AI that Enterprise pricing is customized for each company based on factors like the number of users, volume of queries, and specific feature requirements such as the depth of integration with company systems. ChatGPT Enterprise uses a similar, customized pricing model.
Corporate adoption of gen AI remains challenging
While the number of options for businesses to incorporate AI into their workflows is growing, it’s important to note that doing so still poses considerable challenges. Companies have to train employees on these tools, decide what processes they’re actually good for, and ensure they’re being used within the bounds of what the company permits. For some, there are also copyright and compliance concerns, plus the fact that generative AI is still plagued by hallucinations and opens new security vulnerabilities. And yet, the great AI adoption continues.
And with that, here’s more AI news.
Sage Lazzaro
[email protected]
sagelazzaro.com
Correction: Due to an editing error, Tuesday’s edition of Eye on AI misspelled the last name of Honeywell CEO Vimal Kapur in several instances.
AI IN THE NEWS
OpenAI execs court global investors for tens of billions to build U.S. AI infrastructure. After spending the earlier part of the year seeking the U.S. Government’s help to fund the project, Bloomberg reports that Sam Altman and other OpenAI executives have been meeting with global investors in recent weeks to advance a deal for tens of billions of dollars to build data centers and chip manufacturing facilities in the U.S. Backers are likely to include investors from Canada, Korea, Japan, and the United Arab Emirates, as well as private companies. Microsoft, OpenAI’s biggest investor and strategic partner, would not comment specifically on the project but told Bloomberg it’s involved in any broad infrastructure efforts involving OpenAI. The planned investments could draw U.S. Government scrutiny. Sources tell Bloomberg that OpenAI held meetings with the U.S. National Security Council about the project in recent months.
Safe Superintelligence (SSI) raises $1 billion. It’s the first big news out of the startup, which was cofounded by former OpenAI chief scientist and cofounder Ilya Sutskever in June. Investors include Andreessen Horowitz and Sequoia Capital, and sources tell Reuters the three-month-old company was valued at around $5 billion. SSI was launched shortly after Sutskever resigned from OpenAI in May. While OpenAI has moved away from its initial research model and has begun rapid commercialization of products, SSI says its mission is to develop artificial general intelligence (AGI) and not to commercialize any products until it’s done so.
Nvidia denies reports it was subpoenaed by the DOJ. A Tuesday report from Bloomberg stating that the U.S. Justice Department subpoenaed Nvidia in an escalation of its investigation into the company’s dominance over the AI chip market caused the company’s stock to slip in after-hours trading. But Nvidia yesterday denied the report. The company said it’s inquired with the department but hasn’t been subpoenaed.
FORTUNE ON AI
A 22-year-old is betting on ‘swarms of AI agents’ to radically transform the humble spreadsheet —by Sharon Goldman
Nvidia’s Jensen Huang lost $10 billion in a day after historic stock rout—more pain may be in store —by Christiaan Hetzner
Canva says its new AI features justify raising subscription prices by 300% —by Sydney Lake
Dating apps not working? Let an AI wingman do your flirting. —by Jane Thier
1 in 5 workers are ‘underground’ AI users—here’s why they’re keeping it a secret —by Emma Burleigh
AI CALENDAR
Sept. 10-11: The AI Conference, San Francisco
Sept. 10-12: AI Hardware and AI Edge Summit, San Jose, Calif.
Sept. 17-19: Dreamforce, San Francisco
Sept. 25-26: Meta Connect in Menlo Park, Calif.
Oct. 22-23: TedAI, San Francisco
Oct. 28-30: Voice & AI, Arlington, Va.
Nov. 19-22: Microsoft Ignite, Chicago, Ill.
Dec. 2-6: AWS re:Invent, Las Vegas, Nev.
Dec. 8-12: Neural Information Processing Systems (Neurips) 2024 in Vancouver, British Columbia
Dec. 9-10: Coins2Day Brainstorm AI San Francisco (register here)
EYE ON AI NUMBERS
60 million
That’s how many times an AI-generated image depicting Kamala Harris in communist garb, posted by Elon Musk on X, was viewed in less than 24 hours, according to CNN. Musk shared the deceptive image along with the false caption: “Kamala vows to be a communist dictator on day one. Can you believe she wears that outfit!?” The post appears to violate X’s policy on manipulated content and still remains up on the platform, which of course is owned by Musk.
The image is not especially convincing, but it highlights how easy AI makes it to create false and inflammatory images. It’s also important to note that content doesn't have to be real or even look real for users to spread it and for people’s perceptions to be influenced by it. In the replies to the post, many users called out Musk for using AI to spread disinformation and noted how it’s ironic that he’s been one of the loudest voices warning about the dangers of AI.
