Ex-Palantir Execs Launch Dream With $17M as Proton & Nvidia Face Tests

Physical-world AI foundation models for architecture and construction, aggressive internal AI software engineering mandates at encrypted-services provider Proton, and a U.S. counterintelligence investigation involving Singapore AI cloud firm Megaspeed and Nvidia headlined enterprise technology reporting across California, Switzerland & Singapore on Thursday, October 29, 2026.

Ex-Palantir Executives Launch Dream With $17M From a16z for AI Construction World Models

Former Stanford computer science roommates and decade-long Palantir veterans Andy Elder and Arjun Mathur emerged from stealth Thursday with Dream, announcing $17 million led by Andreessen Horowitz; trained on 100 million sq. ft. across 10 active projects in California and Texas, as reported by TechCrunch's Marina Temkin:

• Founder Background & Investors: Prior to launching Dream early last year, Co-founder Andy Elder led product design for a team of 800 people; Arjun Mathur was commercial SVP at Palantir. The $17 million seed round was led by Andreessen Horowitz partner Kimberly Tan, with participation from SV Angel, XYZ, and Zigg Capital. "We've gone from dial-up internet to superintelligent AI, but our ability to build the physical world around us has actually decreased," Mathur told TechCrunch.
• "Structural Cognition" vs. Legacy AEC Tools: Seeking to replace legacy architecture, engineering, and construction (AEC) tools such as Autodesk's AutoCAD and Revit—and competing in a sector alongside startups Motif, Snaptrude, and Arcol—Dream trained its physics-grounded world model on more than 100 million square feet of building data collected from U.S. permit archives. "Our model actually understands what's within those walls. What materials are used, what the beams are made of, and how much weight they can support," Elder said, describing the approach as "structural cognition" that accounts for internal plumbing and HVAC systems to conceptualize whole buildings and estimate overall construction costs.

Inside Proton's Aggressive Push Toward AI-Generated Code

In enterprise software engineering, an investigation by 404 Media reporter Joseph Cox revealed 90% of code in October written by AI after interviewing 8 current and 1 former employee at Swiss privacy company Proton (maker of Proton Mail, founded in 2014, and the Lumo 2.0 LLM chatbot launched in June):

• Executive Mandates & Code Tracking: Reviewing internal Slack messages, audio recordings, and documents, 404 Media reported that CEO Andy Yen—a former CERN particle physicist—established an internal email address to "ruthlessly smash everything that stands in the way of AI velocity," installed tracking software on employee computers to measure AI code generation, and told the security Slack channel at the end of August to "stop lovingly handcrafting all our code and use AI more," adding: "This will be the last year that we hire engineers, and starting next year we hire what we used to call product managers, to build things with AI."
• CTO Manifesto, Claude Usage & Re-Interviews: According to meeting audio obtained by 404 Media, CTO Bart Butler stated that 90% of Proton's code in October was written by AI—much of it using Anthropic's Claude—and published an internal AI manifesto stating that engineers' jobs are now to help LLMs write code faster and have AI review most code. While an internal document noted Proton is not on track to hit its annual recurring revenue target by year-end due to "insufficient velocity," the company notified staff this week of "consequential" re-interviews regarding their AI usage. A Proton spokesperson did not dispute 404 Media's reporting, stating: "In our determination to move fast, we have sometimes been too blunt or too slow in explaining ourselves to our people. We will work on that, as their questions are the ones we most need to answer."

Nvidia Responds to Report on U.S. Surveillance of Singapore Cloud Executive

Separately, TechCrunch's Julie Bort reported on 1 year of U.S. intelligence surveillance of Megaspeed executive Alice Huang following 2025 chip inquiry, citing a New York Times report:

• Surveillance of Megaspeed Executive: According to anonymous sources cited by the Times, U.S. intelligence agencies surveilled Alice Huang (also known as Huang Le), an executive at Singapore AI cloud firm Megaspeed—which the Commerce Department investigated in 2025 over allegations of reselling restricted AI chips to Chinese companies—for about a year after she was photographed alongside Nvidia CEO Jensen Huang at regional business dinners in Singapore, examining whether she was gathering information on Nvidia's technology plans and export-control discussions with the Trump Administration.
• Nvidia Statement: An Nvidia spokesperson denied any unusual relationship, telling TechCrunch in an emailed statement: "Jensen's interactions with Ms. Huang are not unusual or unexpected. Meetings with customers are part of a C.E.O.'s job; Jensen talks with thousands of customer executives every year. Our regional cloud customers receive ordinary business information—not sensitive NVIDIA trade secrets—and zero information relevant to national security."