Ex-Palantir execs launch Dream to build AI world models for construction; plus, Upwind raises $400M; and Mintive lands Cowboy Ventures, Coatue, angels Jeff Dean, Andrej Karpathy

Speaker A: This is techcrunch.

Speaker B: This episode is brought to you by Subaru. You know that place? Your mind always wanders to the lake, the mountains, the homes of the people you love most. The completely redesigned 2026 Subaru Outback is built to help you get there. With legendary capabilities, standard symmetrical all wheel drive and the confidence to go further, the 2026 Subaru Outback love never looked so good. Visit subaru.com outback to learn more.

Speaker A: Andy Elder and Arjun Mather were Stanford roommates who both studied computer science before spending a decade at Palantir. Elder led product design for a team of 800 people building software for US defense and intelligence agencies, while Mather served as a senior vice president in the company's commercial practice. When the friends decided to build a startup early last year, there was no question they would do it together. Despite having no direct background in architecture, Elder and Mather were troubled by a growing problem. Fewer buildings are constructed in the US Today than a few decades ago. As development costs have surged, Mather told TechCrunch, We've gone from dial up Internet to superintelligent AI, but our ability to build the physical world around us has actually decreased. Believing AI can boost productivity across the architecture, engineering and construction industry, they set out to build Dream, a startup developing built world models that automate building design and engineering. Dream has emerged from Stealth with $17 million in seed funding. The Led by Andreessen Horowitz alongside other backers including Svangel, XYZ and Zig Capital, the startup aims to replace cumbersome, outdated AEC tools like Autodesks, AutoCAD and Revit. If you sit next to an architect or a structural engineer, you will notice that they are still drawing every piece of their designs part by part on legacy software, mather says. And AI really plays no role in that equation. While a Standard LLM like ChatGPT can generate an image of a building, the result will not be structurally sound. Dream, by contrast, is grounded in real physics and construction standards, just as LLMs learn from text. Dream's world model is trained on more than 100 million square feet of building data gathered from permit archives across the U.S. elder said. 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. The startup calls its approach Structural Cognition because its model generates structurally precise designs that account for internal building elements like plumbing and H Vac. Dream has already begun selling its product to AEC professionals and developers who use the model to conceptualize building design and estimate overall UH costs. The platform is already deployed across 10 active projects in Texas and California. While Dream is not the only company attempting to update AEC software with AI, other startups working on this include Motif, Snaptrude and Arkl. Elder and Mather claim that they are the only ones that developed a world model that instantly designs whole buildings. Kimberly Tan, a partner at Andreessen Horowitz, said that while the firm first invested in Dream over a year ago, when the startup model was just an idea, she has been extremely impressed with their progress. I have not seen anybody take the technical approach, she said. They're doing everything based on their own models. Upwind is a San Francisco startup that uses AI to secure cloud infrastructure, and the company has raised a $400 million Series C round at a $3.8 billion valuation, led by Bessemer Venture Partners, with participation from existing backers including Greylock Partners and Kraft Ventures. The round comes nine months after Upwind closed its previous $250 million financing at a $1.5 billion valuation. The fresh funding, which has been in the works for months, nearly fell apart after Datadog attempt to a person familiar with Upwind. Datadog is a publicly traded company that sells tools for monitoring how software runs, known as observability. It did not offer a premium above the $3.8 billion valuation set by venture capital investors, the person said, so Upwind decided to stay independent. 4 year old upwind has grown rapidly amid rising AI driven cybersecurity threats. The startup has been adding $20 million to $30 million in annualized recurring revenue each quarter and is projected to end the year at roughly $90 million, according to a person once again familiar with Upwind's financials. That means that VCs and Datadog valued the startup at more than 40 times its forecast ARR, a common yardstick for pricing software companies. The high multiple is also well below the ADX ARR fetched by Sierra, another fast growing cybersecurity startup. Cybersecurity is poised to be the third big wave of AI software that companies will pay for, after AI coding tools and customer support, Theory Ventures general partner Tomas Tungus recently wrote in his widely read blog. By walking away from an acquisition offer, Upwind clearly believes that the company will ultimately be worth more than $3.8 billion that Datadog was ready to pay for it. However, given how few cybersecurity companies have gone public in recent years, an eventual sale remains the most likely outcome for the startup. Upwind continues to draw acquisition interest, according to a person familiar with the company. The startup was founded by Amaram Shachar, who sold his previous company, the cloud spend management startup Spot IO to NetApp for $450 million. Upwind calls its approach to thwarting threats runtime security, which means it alerts and remedies attacks and vulnerabilities in active services in real time.

Speaker B: This episode is brought to you by Subaru. You know that place your mind always wanders to. The lake, the mountains, the homes of the people you love most. The completely redesigned 2026 Subaru Outback is built to help you get there. With legendary capability, standard symmetrical all wheel drive and the confidence to go further. The 2026 Subaru Outback love never looked so good. Visit subaru.com outback to learn more.

Speaker A: Well, here's another early stage agent management startup to watch. Because of the backgrounds of the founders and the sheer number of big names it has already ATT as investors. Mentev has raised $11 million total, including a newly announced $7 million seed led by Cowboy Ventures. Additional investors include CO2, Operator Collective and Hustle Fund. Angels include Okta co founder Frederic Carest, Discovery Loop co founder and former Google chief scientist Jeff Dean and OpenAI co founder Andre Karpathy, among others. Mintive is joining a crowded field of products that promise to govern provision and offer identity management to enterprise AI agents. Its platform includes an MCP gateway that allows IT professionals to authorize what the cloud services agents may access and what actions they can perform on them. It also offers an LLM gateway that routes to different models and can cap costs, handles agent identities and authorizations that may differ from the employees that they serve, and an agent monitor that tracks all agents across the company. The company also allows enterprises to create agents that proactively perform long running tasks. Still, it's hard to imagine a field with more competition. Agent management systems are offered by Amazon, Microsoft, OpenAI, Anthropic plus as point products galore from companies as diverse as LangChain, New Relic and Rippling, to name just a few. Then there are countless startups working on the idea. 40 from Y Combinator alone. So what's Mintev's secret sauce? Uh, given the founders long history in building AI for enterprises, they have designed their system to be especially easy to use. Setup amounts to clicking a few buttons. The pitch has won over 50 enterprise customers so far since the startup launched its product back in February. These include Roe, Modern, treasury, the University of Michigan, Coursera, Apax and Stability AI. But it was the founder's industry experience that won over Cowboy Ventures managing partner Eileen Lee. She said AI is going to be able to power incredible things, but we need more infrastructure that's going to give us governance and visibility and control. Li also says she particularly respects that the company has been careful by only raising $11 million. This compares to other agentic building startups raising vast amounts at unicorn valuations. As just one example, Prime Intellect raised $130 million for its Series A at a $1 billion valuation. That's all for now. For the latest in startup tech news, go to techcrunch.com.

Speaker C: They're running a AH Cover 2 defense, but you're running to get Game day deals on Uber Eats Keep everyone at tonight's Watch Party fed with deals from your Go to restaurants or stock up with deals on your favorite snacks and groceries. Whatever the occasion, get gameday deals on Uber Eats all season long. It's not football without food. Uber Eats Order now. Terms apply. Product availability varies by region. See app for details.