Venture capitalists invested 200 billion to bet on AI disrupting everything, but are they prepared to be disrupted by AI themselves?

When AI drastically reduces startup costs, founders may no longer need VC funding at all.

Author: WIRED

Translation: Deep潮 TechFlow

Deep潮 Summary: Venture capitalists are the biggest believers in AI, collectively pouring over $200 billion into AI tracks last year. But an awkward question looms: will AI also disrupt VCs themselves? A platform called ADIN has already replaced human analysts with AI Agents to conduct investment due diligence, completing what used to take days or weeks in just one hour. Even more threatening is a second layer of risk—when AI makes starting a company so cheap that founders might not need VC money anymore. This article interviews several well-known VCs, revealing real industry disagreements and anxieties.

Full Text:

Last fall, venture capitalists flooded into the AI sector with record-breaking investments, with a group of investors gathered to evaluate a new project. The company is called Infinity Artificial Intelligence Institute, developing software for automatic tuning of AI models to make them faster and cheaper. The founding team looks promising, and the market is rapidly expanding. Half of the investors are cautious, while the other half see dollar signs. One called this deal “an absolute blockbuster.”

This company is real, and the $100,000 seed round invested by these VCs is real. But these VCs are actually all AI Agents, belonging to a new platform called ADIN (Autonomous Deal Investing Network).

Launched in 2025, ADIN uses AI to replace human analysts in venture deals. Input a startup’s pitch deck, and it outputs a detailed business model and team analysis, a due diligence checklist including compliance risks, TAM estimates, and a suggested valuation. ADIN has over a dozen different Agents, each with unique personas and investment theses. Tech Oracle examines underlying technology, Unit Master assesses financial fundamentals, Monopoly Maker looks for market monopoly opportunities based on Peter Thiel’s style. When most Agents favor a project, they recommend how much capital ADIN’s fund should allocate. The entire process takes about an hour, compared to days or weeks for traditional VC analysts.

“Venture investing is a game with a low success rate,” says Aaron Wright, co-founder of Tribute Labs, the parent company of ADIN. The current approach—an intuitive, gut-feeling method of picking tomorrow’s unicorns—has only about a 1% chance of hitting a “home run” (i.e., a project returning 10x or more). Three-quarters of VC deals don’t even recoup the initial investment.

Wright believes AI models can significantly improve this success rate. He sees venture capital entering its own Moneyball era, where quantitative methods will surpass human intuition, leading everyone to hit more home runs. “These systems will increasingly filter out bad projects, focus on more successful ones, and reduce operational costs,” Wright says. He believes that within a few years, AI Agents could become the world’s best venture investors.

And then what? “Sand Hill Road might not exist anymore.”

No group is more bullish on AI than venture capitalists. They collectively invested over $200 billion into AI last year. Advances in AI models are changing how investors view nearly every company and industry. Vinod Khosla, founder of Khosla Ventures, recently predicted that by 2030, AI will replace 80% of jobs. But many VCs seem to underestimate how much AI could disrupt their own work.

Marc Andreessen—star VC and co-founder of Andreessen Horowitz—said on his podcast, The Ben & Marc Show, that once AI handles everything else, venture investing might be “one of the last few domains humans still do.” He sees the job as more than just writing checks; it’s about choosing the right ideas and people at the right time and guiding them to success.

“This isn’t science; it’s art,” Andreessen added. “If it were science, someone would be able to fine-tune it to an 80% success rate. But the real world isn’t like that. You’re dealing with chance events. There’s an ineffable quality, a taste factor.”

Many VCs I interviewed share similar views. Keval Desai, managing partner at Shakti, compares early-stage investing to “picking Michael Jordan in kindergarten.” A project with no product, no revenue, only potential. “You can have all the computing power and algorithms, but without data, there’s nothing to analyze,” Desai says. (He admits that when entering unfamiliar markets, he sometimes has Gemini “act as a VC analyst” to get opinions.)

Brian Nichols, co-founder of Angel Squad—a network of angel investors affiliated with early-stage VC firm Hustle Fund—told me he wouldn’t trust AI to do the “screening” part of investing. Ultimately, VC is a relationship business: who you know and who you can personally vouch for matters most. But he also thinks AI might replace other parts of the process. When we spoke, he had just returned from a Hustle Fund team-building event, where a partner used Claude Code to create a tool to filter founders’ emails. “We spend hours every day replying to founders’ pitches,” he said. “That time could be better spent elsewhere.” Afirm founder and managing partner Aydin Senkut of Felicis told me he believes most VCs are experimenting with AI in some way to stay competitive. His firm is testing chatbots to write investment memos, improve deal sourcing, and help partners score founders.

Projects like ADIN aim to automate more of the underlying work. Due diligence—the process of investigating a project’s feasibility, risks, and growth potential—is one of the most time-consuming parts of venture investing, especially with companies in emerging markets. ADIN compresses this step to minutes, quickly flagging regulatory or compliance issues that could derail a deal. When evaluating a mining tech company, ADIN identified a series of export control laws and cross-border data transfer concerns. “These aren’t questions most investors would think to ask,” says Priyanka Desai, a partner at ADIN. She adds that AI “never gets tired, doesn’t have blind spots from inertia, and can surface long-tail risks that are easy to overlook.”

Humans still need to do a few things. First, ADIN’s deal sourcing comes from a venture scout network. While ADIN functions like a traditional VC fund with LPs, it offers scouts an unusual financial incentive—50% of carried interest, typically reserved for GPs (general partners). “Basically, it’s giving GP-level economics to someone who just submits deals and leverages their network,” Desai explains.

Humans also handle the “last mile,” including meeting founders and making the final decision to invest. “We know these systems aren’t perfect, so we need a second confirmation,” Wright says. Sometimes, AI Agents are too aggressive in their recommendations: he showed me a project all Agents liked, but after meeting the founders and discovering issues with existing competitors, ADIN decided not to invest.

On the other hand, Wright says he’s used ADIN to evaluate some companies that have already raised over $20 million, and some Agents were unanimously negative. “Our challenge is figuring out whether that’s accurate or a misjudgment,” he says. In some cases, investors fall into a common human trap: relying on gut feelings to praise a project or founder.

Whether AI can outperform human investors is one thing. But there’s another existential threat: the same AI technology that makes VC work faster and more efficient is also making it easier and cheaper to start software companies. Over the past decade, most VC money has gone into SaaS. But a project that once needed a $2 million seed round to hire a professional engineering team can now reach the same speed with a few vibe coders and less than six figures. The math behind big checks no longer holds.

Until recently, only a tiny fraction of unicorns were self-sustaining. According to SaaStr’s data on SaaS companies, the average software unicorn raised $370 million. Now, companies like AI image generator Midjourney—whose core team is just a few dozen people—have reached unicorn status. (According to PitchBook, Midjourney has about 100 employees. Court documents related to a copyright lawsuit show annual revenue exceeding $300 million. Midjourney did not respond to WIRED’s request for comment.)

This scenario—where some founders no longer need VC funding—is the most likely to terrify investors. “That’s the real existential threat,” Nichols of Angel Squad says. “The money is there, but founders don’t need it.” AI may not directly replace investors, but it could make their investments unnecessary.

Beyond robotics, biotech, or hardware companies, we may soon see fewer startups requiring the large-scale funding that the VC industry has traditionally relied on. This could bring the industry back to its roots: a small, specialized niche bridging scientific breakthroughs and commercial applications. (Major companies building foundational models still exist, likely continuing to use VC funds to pay for enormous compute, data centers, and staff.)

If starting a company becomes cheap enough, we might see this industry rapidly shrink. This could also cause a different kind of unemployment for investors: not replacement, but a shift in business models. “If these funds sit idle, competing for the very few deals that still need funding, that creates another problem,” Nichols says. “That’s what keeps investors awake at night.”