AI Funding Roundup: July 2026’s Biggest Deals

Nuclear startup Valar Atomics is in talks to raise a $1 billion equity round at a $6 billion valuation, with Sequoia expected to lead the deal, according to TechCrunch’s July 2026 report. Separately, Microsoft and 3M announced a strategic partnership on July 15 targeting AI data center infrastructure, while the pre-seed funding environment for AI startups grows increasingly competitive heading into fall.

Valar Atomics Seeks $6B Valuation in Sequoia-Led Round

Valar Atomics, a three-year-old small modular reactor (SMR) startup based in El Segundo, California, is negotiating a new capital raise that would nearly triple its last known valuation. The company previously raised $450 million — comprising $340 million in equity and $110 million in debt — at a $2 billion valuation, per a Bloomberg report from March 2026.

The Information first reported the funding discussions. TechCrunch subsequently confirmed through three independent sources that part of the $1 billion target has already been raised at a lower valuation — a deal structure that is becoming increasingly common in AI-adjacent fundraising, where tranches close at different prices and times, creating the appearance of a single uniform valuation.

Earlier in July, Valar demonstrated that its nuclear reactor could supply power to an NVIDIA AI chip — a proof-of-concept that coincided with a partnership announcement between Valar and NVIDIA to explore nuclear energy as a power source for future AI data centers. The timing underscores the acute pressure on data center operators: electricity demand from AI infrastructure is projected to outpace utility capacity additions by years in many regions, creating an opening for alternative energy suppliers.

Sequoia and Valar Atomics both declined to comment on the funding talks.

Microsoft and 3M Partner on AI Data Center Infrastructure

Microsoft and 3M announced a strategic partnership on July 15, 2026, targeting AI data center infrastructure and enterprise-scale digital operations, according to Microsoft’s official announcement. The collaboration combines 3M’s materials science portfolio with Microsoft’s cloud and AI platform capabilities.

The partnership reflects a broader pattern of industrial manufacturers aligning with hyperscalers as data center build-out accelerates. Thermal management, cabling, and physical infrastructure — areas where 3M has established product lines — are increasingly bottlenecks as GPU-dense racks generate more heat per square foot than traditional server hardware.

Neither company disclosed financial terms of the arrangement in the July 15 announcement.

Pre-Seed Funding Grows Harder as AI Raises the Bar

AI startups are absorbing a disproportionate share of seed-stage capital in 2026, raising the evidence threshold that pre-seed investors now demand from founders who lack a working product. TechCrunch is addressing this directly at its Disrupt 2026 conference — scheduled for October 13–15 in San Francisco at Moscone West — with a dedicated session titled “Winning Pre-Seed Without a Product.”

The panel will feature Sandhya Venkatachalam, founder and managing partner of Axiom Partners, a newly launched $52 million early-stage venture fund focused on connecting founders with AI practitioners. Venkatachalam previously served as a general partner at Khosla Ventures and Social Capital, where she made the first institutional investment in Groq and led deals into GalileoAI, ForethoughtAI, and FirefliesAI — all of which have since been acquired or reached significant scale.

The session is part of Disrupt’s Builders Stage, which covers operational decisions, fundraising mechanics, and go-to-market strategy.

Cybersecurity M&A: AI Hype Is Inflating Valuations

In cybersecurity specifically, acquirers are being warned to look past AI-inflated valuations toward actual integration fit and measurable outcomes. Rohit Dhamankar, VP of M&A and AI Strategy at Fortra, wrote in Forbes Technology Council in July 2026 that compelling acquisitions must fill a specific capability gap, integrate cleanly with existing tools, and serve a defined customer need — not simply add AI features.

Dhamankar cited Gartner data showing that only 20% of cybersecurity teams report highly beneficial results from GenAI use cases, arguing that “more AI” does not constitute a strategic rationale for an acquisition. His framing applies equally to investors evaluating AI-native startups: headline model capabilities rarely translate directly into enterprise outcomes without significant implementation work.

The warning is particularly relevant as deal volume in AI-adjacent security companies has risen alongside valuations that may not reflect production-ready deployments.

Enterprise AI Agents Are Shipping Faster Than Evaluations Can Keep Up

A new VentureBeat Pulse Research study of 157 enterprises found that half have already deployed an AI agent or LLM feature that passed internal evaluations and then caused a customer-facing failure in production, according to VentureBeat’s July 2026 report. A quarter of respondents said it had happened more than once.

Despite this, 66% of organizations already permit or are actively building toward fully automated, zero-human-in-the-loop deployment for at least some agent changes. Only 5% of technical leaders said they fully trust automated evaluation today, and the most-cited limitation — named by 29% of respondents — is that evaluations align poorly with real-world outcomes.

The finding has direct implications for enterprise AI investment decisions: organizations are committing budget to agentic infrastructure before the reliability tooling needed to govern it has matured.

What This Means

The deals and data points from July 2026 point to a single underlying tension: capital is moving faster than validation. Valar Atomics’ structured round — where different investors may be paying different prices for the same company — illustrates how AI-adjacent energy infrastructure is attracting the same valuation dynamics that characterized pure AI software rounds in 2023 and 2024. The Microsoft-3M partnership signals that physical infrastructure suppliers are now considered strategic partners, not just vendors, in the AI build-out.

On the startup funding side, Axiom Partners’ $52 million fund and the Disrupt pre-seed panel both acknowledge that the bar for early-stage capital has risen sharply — but that conviction and narrative can still substitute for product in the right context. Meanwhile, Dhamankar’s Gartner-backed caution and VentureBeat’s enterprise survey data suggest the same problem from two angles: AI adoption decisions are outrunning the evidence needed to justify them. Investors and acquirers who build evaluation discipline now are likely to avoid the correction that follows.

FAQ

What valuation is Valar Atomics seeking in its 2026 funding round?

Valar Atomics is targeting a valuation of approximately $6 billion in a new round expected to total $1 billion in equity, with Sequoia reportedly set to lead the deal. The company’s prior raise valued it at $2 billion, according to Bloomberg.

What is Axiom Partners and who founded it?

Axiom Partners is a newly launched $52 million early-stage venture fund founded by Sandhya Venkatachalam, a former general partner at Khosla Ventures and Social Capital. The fund focuses on connecting founders with AI practitioners to support applied AI development.

Why are enterprise AI agent evaluations failing in production?

According to VentureBeat’s survey of 157 enterprises, the primary reason is that internal evaluations do not align with real-world outcomes — cited by 29% of respondents. Half of organizations surveyed had already experienced at least one customer-facing failure from an agent that had passed all internal tests before deployment.