Daily Funding Roundup:
Mar 27, 2026
Defense AI titan Shield AI raised a massive $2B at a $12.7B valuation, the largest defense tech round ever. GPS-alternative satellite startup Xona Space closed $170M. Military drone maker PDW pulled in $110M. SoftBank secured a record $40B bridge loan for its OpenAI stake. Anthropic's secret Mythos model was accidentally leaked. Defense and aerospace dominated with $2.3B+ raised across seven deals.
Rounds
Defense AI juggernaut Shield AI closed a staggering $2B raise at a $12.7B valuation, more than doubling from $5.3B a year ago. Advent International and JPMorgan Chase's Security and Resiliency Initiative co-led a $1.5B Series G, with Blackstone investing $500M in preferred equity (plus a $250M delayed draw facility). Shield AI builds Hivemind, autonomous pilot software that enables teams of aircraft to fly missions without GPS, communications, or human pilots. The company is projecting $540M+ in revenue for 2026 (80% growth YoY). A portion of the proceeds will fund the acquisition of Aechelon Technology, a defense simulation company. Founded in 2015 by former Navy SEAL Brandon Tseng, his brother Ryan Tseng, and Andrew Reiter. Total raised: $3.3B.
Satellite navigation startup Xona closed $170M to build a GPS alternative that actually works indoors, under tree cover, and in contested environments. The Pulsar constellation will deploy 258 satellites flying 20x closer to Earth than GPS, delivering signals 100x stronger. A distributed clock architecture eliminates the need for expensive atomic clocks used in traditional GPS. First satellites launch by end of 2026 from the company's new Burlingame, California factory. Founded in 2019 by Stanford graduates Brian Manning (CEO) and Tyler Reid (CTO). Total raised: $289M.
Huntsville-based military drone manufacturer PDW raised $110M to scale production of its modular combat UAS platform. The portfolio includes the C100 quadcopter, Attritable Munition strike drone, SIM flight simulator, and CORE mission planning software, all built for contested environments where jamming and GPS denial are the norm. Founded in 2019 by former Special Operations veterans and professional drone racing pilots, PDW has 336 employees with over 20% military veterans. The defense investor mix (Hanwha, Booz Allen) signals institutional confidence in U.S. drone production capacity at a time when Ukraine has demonstrated the decisive role of small UAS in modern warfare. Total raised: $135M.
Also Noted
AI-powered agentic hiring marketplace raised $40M led by Scale Venture Partners with Felicis and Liquid 2 Ventures. Connects companies with a decentralized network of specialized recruiters. Angel investors include the CTO of Palantir, COO of Stripe, President of Shopify, and co-founders from Canva and YouTube. 1,000+ companies served. Total raised: $65M.
AI psychiatry platform raised $20M led by Headline with Village Global, TA Ventures, and Operator Partners. The copilot augments clinical decision-making and automates admin tasks, saving psychiatrists 30+ minutes per patient visit. Emerged from stealth in October 2025 with $18.4M seed. Total raised: $38.4M.
Healthcare retention-driven recruitment platform raised $20M led by Crosslink Capital and Digitalis Ventures. Uses an 'ROTC for healthcare' model where employers fund education in exchange for service commitments. Delivers retention rates 2.5x higher than traditional hiring with $130M+ in employer commitments.
Digital funeral planning platform raised $9M led by Lachy Groom and Haystack. Founded by Stripe alum Sam Gerstenzang, Meadow lets families manage funeral logistics, documentation, and memorials entirely online. 3x revenue growth in 2025, now the largest independent funeral home in California, expanding to Texas and Washington.
News & Signals
SoftBank's $40B bridge loan signals an OpenAI IPO is imminent
SoftBank signed a record $40 billion bridge loan arranged by JPMorgan, Goldman Sachs, Mizuho, SMBC, and MUFG to fund its $30B follow-on investment in OpenAI. The non-collateralized loan matures in just 12 months, which only makes sense if the lenders believe OpenAI's IPO will happen before March 2027. SoftBank has already injected $30B+ into OpenAI and holds a ~90% stake in Arm Holdings. TechCrunch reported that the short maturity structure strongly suggests the loan is designed to bridge to a liquidity event. If OpenAI does IPO in late 2026, it would be the largest tech listing since Alibaba in 2014, and Masayoshi Son's biggest bet since SoftBank's $32B Arm acquisition.
Defense tech reaches escape velocity: $2.1B raised in a single day
Shield AI ($2B) and Performance Drone Works ($110M) raising a combined $2.1B represents the single largest day of defense tech startup funding on record. Shield AI's $12.7B valuation makes it the most valuable private defense tech company globally, surpassing Anduril ($14B, but with public-market comparisons). The investor lists tell the story: Blackstone, JPMorgan, Advent, Hanwha Aerospace, Booz Allen Hamilton, Lockheed Martin. These are not venture investors making speculative bets; they are defense establishment players making strategic investments. The through-line is autonomy: Shield AI's pilotless aircraft and PDW's combat drones both operate in GPS-denied, communications-denied environments. Ukraine proved the concept; this capital scales it.
Anthropic's Mythos leak reveals a model with 'unprecedented cybersecurity risks'
A misconfigured content management system accidentally exposed draft blog posts from Anthropic, revealing a previously secret AI model called Claude Mythos (also referred to as Capybara). An Anthropic spokesperson confirmed it represents 'a step change' in performance, with 'dramatically higher scores' on software coding, academic reasoning, and cybersecurity tests. Internal documents warn the model could heighten cybersecurity risks by rapidly finding and exploiting software vulnerabilities. The leak itself, an unsecured data store with 3,000+ assets, is ironic given the model's cybersecurity implications. The bigger signal: if Capybara represents a new tier above Opus, the AI capability curve is not flattening.
GPS alternatives are now a $500M+ category
Xona's $170M raise for a satellite-based GPS alternative comes as governments and militaries globally recognize that GPS is a single point of failure for modern civilization. GPS signals are weak (arriving at Earth's surface at the power level of a 40-watt light bulb from 12,500 miles away), trivially jammed, and increasingly spoofed. Russia has demonstrated GPS disruption affecting commercial aviation across Europe. China has its own BeiDou system. Xona's approach (LEO satellites, stronger signals, no atomic clocks) directly addresses military and commercial needs. Combined with other GPS-alternative investments (including Xona's own earlier SpaceWERX grant from the U.S. Space Force), the category has attracted over $500M in venture and government funding.
VC Mood on X
Thursday's VC conversation was dominated by one number: $12.7 billion. Shield AI's valuation prompted a wave of "defense tech is the new enterprise SaaS" takes, with bulls arguing that Pentagon modernization budgets and allied rearmament create a multi-decade tailwind. Several partners noted that Shield AI's investor list (Blackstone, JPMorgan, Advent) represents a phase shift from venture-funded defense startups to institutional capital flowing into defense technology. "When Blackstone writes $500M checks for drone software, the category is no longer emerging," one GP posted. "It is mainstream."
The SoftBank $40B bridge loan for OpenAI drew more structural analysis than surprise. VCs focused on the 12-month maturity as implicit confirmation of a 2026 OpenAI IPO. The math: SoftBank needs a liquidity event to repay a $40B non-collateralized loan. The only event large enough is an IPO (or a secondary at a dramatically higher valuation). Several investors noted that SoftBank is now leveraged to AI in a way that makes its 2000-era telecom bets look conservative. If OpenAI IPOs at $300B+, Son's bet pays off spectacularly. If AI winter hits, SoftBank's balance sheet is in serious trouble.
The Anthropic Mythos leak generated the most heated debate. AI safety advocates seized on the "unprecedented cybersecurity risks" language in the leaked internal documents, arguing it validates concerns about frontier model capabilities outpacing safety measures. AI accelerationists countered that the leak itself (3,000 unsecured assets in a misconfigured CMS) demonstrates that human error, not AI capability, remains the real security risk. The irony of an AI safety company leaking documents about a dangerous model via a trivially preventable misconfiguration was not lost on Crypto Twitter, which had a field day.
Rounds and signals sourced from SEC filings, press releases, and verified news reports. "Also Noted" covers smaller or less-documented deals. All amounts in USD unless noted. Reporting reflects information available at time of publication.