Meanwhile, among AI 100 winners, who’s got the most investors on their cap table?
Together AI.
With 45 total investors, Together AI, which offers infrastructure to support open-source model development and deployment, has the busiest cap table. (It’s followed by AV startup Wayve with 35 investors.)
Together AI raised a $305M Series B (co-led by General Catalyst) in February at a $3.3B valuation.
There are 9 investor pairs that have co-invested in 3 AI 100 winners. And General Catalyst is in 5 of those pairs, co-investing most frequently with Lightspeed, SV Angel, Kleiner Perkins, Greycroft, and Amazon’s AWS Generative AI Accelerator.
Corporate venture is increasingly involved in the AI sphere, claiming 3 of the top 10 AI 100 investors — Nvidia’s NVentures; AWS Generative AI Accelerator; and Samsung NEXT.
Baby bumps and lobsters? Nvidia blasted Anthropic over “tall tales” of AI chip smuggling — including claims that Chinese operatives hid semiconductors in prosthetic baby bumps and lobster shipments. The bizarre allegations came from Anthropic’s blog post urging tighter US export restrictions on AI hardware. Nvidia, whose chips power Anthropic’s models, called the claims unserious and said policy shouldn't be used to “manipulate regulators” or throttle competition. Meanwhile, Anthropic stood by its national security pitch, citing the need to protect American AI leadership as global rules tighten.
Ruff endorsement:NutriCanine got called out by entrepreneur Sam Parr for using AI to fake a video ad implying he backed their dog food brand. Parr said he never mentioned the company and slammed the move as dishonest, especially since NutriCanine’s tagline is “an honest approach that caters to your dog.” The ad was taken down after backlash, and Parr requested a $2K donation to a dog rescue in lieu of legal action — which the company made.
CultGPT: A viral Rolling Stone story recently chronicled how ChatGPT is fueling a wave of spiritual delusions — and wrecking relationships in the process. One woman said her husband, convinced the bot had “awakened,” declared himself God and nearly left her to “grow at a higher frequency.” Others say their partners received cosmic titles like “spark bearer” and “river walker,” or believe the AI granted them access to ancient archives and teleporter blueprints. Psychologists warn that ChatGPT’s tendency to validate user beliefs makes it a potent enabler of existing mental health issues.
Nonprofit or never? OpenAI is walking back its for-profit ambitions, saying the nonprofit will stay in charge after all. The about-face follows mounting pressure from former employees, legal experts, and civil society groups, who warned that a restructuring would gut the company’s original mission. Under the new plan, OpenAI’s LLC becomes a public benefit corp — but the nonprofit keeps control. CEO Sam Altman said the move keeps “everyone aligned” and downplayed the ongoing beef with Elon Musk, whose xAI is trying to block the for-profit shift (and offered $97.4B to buy OpenAI outright). “You all are obsessed with Elon,” Altman said. “More power to you."
Leaderboard illusion: Cohere’s VP of Research Sara Hooker is calling out LM Arena for allegedly letting top model makers — including Google and OpenAI — quietly test dozens of variants and cherry-pick the best scores. A new paper from researchers at Cohere, Princeton, Stanford, MIT, and AI2 backs her up, saying the practice gave Meta, Amazon, and others a “significant” ranking boost. LM Arena fired back on X, claiming providers just “make different choices” — but the flap has reignited calls for more transparency in open model benchmarking.
Source: X
Engagement > answers? Instagram co-founder Kevin Systrom criticized AI chatbots this week for prioritizing user engagement over usefulness. Speaking at StartupGrind, Systrom said many models are designed to keep conversations going — not necessarily to deliver quality answers — likening the trend to growth tactics used by social media platforms. “Every time I ask a question, it asks another little question to get yet another one out of me,” he said. OpenAI has acknowledged the issue, attributing it to short-term user feedback and emphasizing that its models ask clarifying questions only when needed.
.png?upscale=true&width=1040&upscale=true&name=AI100%20Investors%20-%20043025%201%20(1).png)
.png?upscale=true&width=1040&upscale=true&name=AI100%20Investors%20-%20043025%202%20(1).png)
.png?upscale=true&width=1040&upscale=true&name=AI100%20Investors%20-%20043025%203%20(1).png)
