Nvidia’s writing checks into sovereign AI now. So what exactly did it buy? For anyone joining us mid-arc, here’s the short version: AI venture funding is clustering around companies claiming foundational roles in model platforms, infrastructure, applied tooling, physical AI, and sovereign AI. Recent rounds include Augment Code’s $227 million Series C, Zenity’s $125 million Series C for AI-agent security, Olix’s $312 million chip round, and Horizon3’s $250 million autonomous-security Series E. This is Startup Fundraising. Today: Nvidia joins Sarvam, agent security finally brings customer numbers, and a bank-core startup gets a seed check with some serious expectations attached. This story isn't over: AI venture market concentration. Follow us wherever you're listening, and the next chapter comes to you. Here's Tasneem Kanchwala at Free Press Journal:
In a milestone development, Bengaluru-based Sarvam AI has become the first Indian company to bring on global chipmaker Nvidia as a strategic investor, with the artificial intelligence startup closing a $75 million extension of its Series B funding round led by the company. Sarvam AI cofounder Pratyush Kumar has confirmed this development.
Sarvam’s filing gives us unusually clean math: Nvidia put in about $25 million for 1.66%, implying roughly a $1.51 billion valuation. A strategic check with an actual denominator—what a concept. And Nvidia is now backing a sovereign-AI distribution bet in India, not just selling it chips. What does it get for that $25 million: preferred access, a flagship customer, or a front-row seat to a market it can’t build alone? The $75 million extension comes after Sarvam’s earlier $234 million Series B, led by HCL Technologies. Nvidia is the extension’s biggest investor, putting it directly behind Sarvam’s sovereign-AI push. Fine, this one has evidence. But a $1.5 billion price means Sarvam has to become more than India’s well-funded AI emblem. It needs customers still buying once the strategic-investor headlines fade. From SiliconANGLE:
Obsidian Security Inc. has raised an $85 million Series D funding round at a post-money valuation of $1.1 billion as enterprises increasingly look to secure autonomous artificial intelligence agents accessing cloud applications, Chief Executive Hasan Imam said today in an exclusive interview with theCUBE with me at our NYSE Wired studio in NYC.
Obsidian’s $85 million Series D gets my attention because 14 customers are already paying more than $1 million a year. Fine—there’s actual enterprise spend here, not just an agent-security slide deck. And another 100 customers clear $100,000 annually. At a $1.1 billion post-money, Crescent Cove is paying for a business with visible spending tiers—more useful than the valuation headline. But 65% of its enterprise customers have already let agents into third-party SaaS data. That’s the sales pitch and the danger: every OAuth token becomes a tiny, tireless employee with terrible judgment. Zenity’s $125 million round showed investors will crowd into this attack surface. Obsidian brings better underwriting: named customer thresholds and a stated R&D and go-to-market use for the $85 million. GlobeNewswire writes:
Maximum today emerged from stealth with a $30 million Seed round, one of the largest Seed financings in fintech, led by CRV with participation from Pear VC, Restive, Plug and Play Ventures, Anthemis and others. Founded by Randy Fernando, Maximum is building an AI-native operating system for banks designed to replace the fragmented, decades-old infrastructure that powers most of the banking industry today.
$30 million at seed to replace a bank’s core system? Good. That’s an actual hard problem, not another chatbot wearing a compliance badge. CRV led, joined by Anthemis and investors who’ve backed Fernando before. Randy Fernando sold Vault to Acorns in 2017 and Power to Marqeta in 2023, so this check is plainly underwriting his execution record. And he’d better know the procurement maze. More than 70% of nearly 5,000 U.S. banks still run legacy cores—because swapping one is like changing an airplane engine mid-flight, except the airplane is everybody’s checking account. “AI-native operating system” is the pitch here. The investable claim is core replacement: if Maximum can move banks off decades-old infrastructure, $30 million will look restrained; if it merely layers agents on top, that seed round gets expensive fast. Here's TechCrunch:
Rapoport and her three co-founders — Ohad Hen, Barak Goldstein, and Idan Tsitiat — have a different idea about how to bring AI into broader use. To pursue it, the company raised $20 million in pre-seed funding led by Marc Benioff’s Time Ventures, with additional backing from tech luminaries like Michael Dell, Aaron Levie, and George Kurtz.
June raised $20 million pre-seed from Marc Benioff’s Time Ventures without sharing a valuation—and, apparently, without even bringing a deck. The founders’ Salesforce pedigree bought them a very short diligence process. They want AI to replace forward-deployed engineers who get enterprise AI working. Fine. But a Fortune 500 deployment fails because of permissions, rotten data, and six internal owners—not because nobody generated enough code. Benioff built a professional-services ecosystem around Salesforce. So his backing of a company arguing services don’t scale says something. June’s four founders also sold Bonobo AI to Salesforce in 2019; this is a team he’s seen operate. A $20 million pre-seed can fund a lot of elegant deployment agents. It cannot make a bank’s security team approve a production model before Christmas. June needs to prove it shortens that clock, not just automates the PowerPoint around it. Here's TechCrunch:
Endeavor Optical Networks, a startup founded in May and emerging from stealth today with $10.75 million in seed funding from General Catalyst and Andreessen Horowitz, is betting on that plan. The co-founders, CEO Charlie Horowitz and CTO Tyler Presser, aim to launch a network of laser-equipped spacecraft to link data centers from orbit.
EON raised $10.75 million to put laser links in orbit and wants a 2027 demo. They’re starting at 2.4 terabits per second; undersea fiber runs at 200 terabits. That gap contains, oh, only the atmosphere and all of orbital mechanics. General Catalyst and a16z are funding a company founded in May to challenge ocean cable economics. Seed capital is usually for finding product-market fit; here it’s also buying a physics experiment. And the buyer is supposedly a hyperscaler that already knows how to procure cable, build data centers, and wait out vendors. EON has to prove the laser link is reliable enough that somebody reroutes serious traffic through it—not a nice-looking demo packet. TechCrunch points to NASA and private companies demonstrating optical links, but those were around 2.5 gigabits per second. EON is aiming roughly a thousand times higher from the opening pitch. Modest ambition has left the building. If you’re into startup fundraising, you may also enjoy The Data Center Daily, a daily briefing on AI compute, hyperscaler capex, the power grid, semiconductor supply, and energy markets reshaped by intelligence at scale. Find it wherever you listen to podcasts.
You’ll find links to every story in today’s show notes, with more detail on anything that caught your attention. Thanks for listening, and we’ll be back tomorrow. That’s Startup Fundraising for today. This is a Lantern Podcast.