VC
Value Add VC
⚡HomePulse⚡Helpful Apps📝Blog
← Value Add PulseFUNDING~10x return

GV's Dave Munichiello on Qualcomm's Modular Deal, a Roughly 10x Return, and the AI Software Shift

GV general partner Dave Munichiello discussed Qualcomm's June 24 acquisition of AI infrastructure software startup Modular in a Crunchbase Q&A published June 30, describing a return of roughly 10x on GV's total dollars invested after leading Modular's Series A at a $155 million valuation with a $23 million check. Munichiello framed the deal as validation of a shift toward 'disaggregated inference' — software that lets AI workloads run efficiently across a mix of chips rather than locking into a single hardware vendor.

Qualcomm acquires Modular, announced June 24, 2026
Deal
$23M (Series A lead)
GV's Original Check
$155M
Series A Valuation
~10x on total dollars invested
Return Cited
$15M at $480M (2017) -> $10B round (2026)
GV/SambaNova Comparable
TC
Trace Cohen
Early-stage VC & angel · Founder, New York Venture Partners
June 30, 2026
3 min read
ShareXLinkedInEmail
KEY TAKEAWAYS FOR VCs & FOUNDERS
1

A roughly 10x return on a Series A lead check is a strong outcome that validates GV's early conviction in AI infrastructure software over pure hardware bets

2

Qualcomm acquiring a cross-hardware compatibility layer signals chipmakers see software, not just silicon, as the differentiator in the next phase of AI infrastructure

3

Munichiello's 'disaggregated inference' framing — blending AI chips, CPUs and GPUs — challenges the assumption that Nvidia's GPU dominance is permanent

4

His prediction of 15-20 tech IPOs within six months adds a concrete, checkable forecast to the broader 2026 IPO wave narrative

TC
The VC Read · Trace's TakeTrace Cohen

Munichiello's 'disaggregated inference' framing is the more important story here than the 10x return itself — if the software layer that lets workloads run across any chip becomes the real strategic asset, that quietly undermines the assumption that Nvidia's GPU dominance is a permanent moat rather than a current-generation advantage. Qualcomm buying that capability instead of just out-speccing Nvidia on silicon is a smarter, cheaper way to compete, and I'd expect AMD and the hyperscalers' custom-chip teams to make similar moves. For founders in AI infrastructure, cross-hardware compatibility is looking like one of the most durable, chip-agnostic wedges left in the category. Watch Munichiello's 15-20 company IPO forecast over the next two quarters — that's a specific, checkable prediction from someone with genuinely informed visibility into the pipeline.

⚔️ AI Chip Wars →📈 AI Valuations →

GV general partner Dave Munichiello discussed Qualcomm's acquisition of Modular, the Palo Alto-based AI infrastructure software startup, in a Crunchbase Q&A published June 30, 2026, describing the deal — announced June 24 — as a validating outcome for GV's early bet on cross-hardware AI software rather than a single chip architecture. GV led Modular's Series A with a $23 million check at a $155 million valuation, and Munichiello cited a return of roughly 10x on GV's total dollars invested from the acquisition.

Modular's core product solves what Munichiello called one of the most strategic problems in all of AI: letting the same AI workloads run efficiently across a mix of hardware — AI-specific chips, CPUs and GPUs — rather than forcing developers to lock into a single vendor's silicon and software stack. Munichiello described this as a shift toward "disaggregated inference," where the software layer managing cross-hardware compatibility becomes as commercially valuable as the chips themselves, since it determines how efficiently expensive compute actually gets used.

Qualcomm acquiring that software layer is a notable strategic signal: rather than competing purely on chip specifications against Nvidia, AMD and custom silicon from hyperscalers, Qualcomm is buying the software that could let its own hardware compete more directly for AI inference workloads regardless of which chip a given task ultimately runs on. That mirrors a broader industry pattern in 2026 of infrastructure providers acquiring or investing in the software layer that determines utilization of their underlying hardware — the same logic behind Amazon's investment in TwelveLabs tied to AWS Trainium, and Nvidia's own investment in Together AI.

Munichiello drew a comparison to GV's earlier investment in SambaNova, where the firm led a $15 million Series A in December 2017 at a $480 million valuation; SambaNova has since raised an $800 million round led by General Atlantic at a $10 billion valuation, illustrating the kind of long-duration return profile GV underwrites in AI infrastructure bets that can take the better part of a decade to fully mature.

The numbers in context: a roughly 10x return on total dollars invested in under three years (Modular's Series A to acquisition) is a strong outcome by any venture standard, though notably faster and smaller in absolute terms than GV's SambaNova trajectory, which took closer to a decade to reach its current scale. That contrast underscores how much the AI infrastructure category has accelerated — deals that once took the better part of a decade to mature are now compressing into a few years given the intensity of capital and acquisition interest in the space.

Munichiello also offered a forward-looking data point for the broader IPO market, predicting that 15 to 20 tech companies could pursue public listings within the next six months — a forecast that, if accurate, would keep 2026's already-record IPO pace (32 companies listed above $1 billion in Q2 alone, per Crunchbase's H1 data) running at a similarly elevated rate into early 2027.

For founders building AI infrastructure software, the Modular outcome is a concrete data point that cross-hardware compatibility layers can command premium acquisition interest from chipmakers looking to compete on software rather than silicon alone. For LPs evaluating GV and similar infrastructure-focused funds, both the Modular and SambaNova outcomes support underwriting AI infrastructure bets on a multi-year timeline, even when near-term outcomes (like Modular's relatively quick acquisition) compress that timeline meaningfully.

What to watch: whether Qualcomm's Modular acquisition translates into measurable market-share gains against Nvidia in AI inference workloads, whether Munichiello's 15-20 company IPO forecast proves accurate over the next two quarters, and whether other chipmakers pursue similar cross-hardware software acquisitions to compete on the same terms.

ShareXLinkedInEmail

Originally reported by Crunchbase News. Analysis and editorial commentary by Value Add Pulse.

← Back to Pulse

Read Next

FUNDING$510B H1 2026

Crunchbase: Global Startup Funding Hit a Record $510B in H1 2026 as AI Swallows the Market

Crunchbase News reported July 2 that global startup investment hit a record $510 billion in H1 2026, already exceeding the $440 billion invested in all of 2025 combined, as AI-focused companies captured more than 70% of Q2 global funding. OpenAI and Anthropic alone raised a combined $217 billion in H1 — 43% of every venture dollar deployed worldwide — while exit activity kept pace, led by SpaceX's record $75 billion IPO and its $60 billion all-stock acquisition of Cursor.

FUNDING$800M

Together AI Raises $800M at $8.3B Valuation as Neoclouds Race to Undercut Frontier Labs on Price

Together AI closed an $800 million Series C on July 1 at an $8.3 billion valuation, led by Aramco Ventures with Nvidia, Vista Equity Partners, General Catalyst and Emergence Capital participating — more than doubling its $3.3 billion valuation from February 2025. The 'neocloud' rents Nvidia GPU clusters and hosts open-source models like DeepSeek, Nemotron and Kimi as a cheaper alternative to closed frontier models, reporting annual bookings exceeding $1.15 billion.

FUNDING$65M

Venice AI Becomes a Unicorn With $65M Series A as Its Privacy-First Platform Takes Off

Venice AI closed a $65 million Series A on July 1 at a $1 billion valuation, led by crypto-focused VC firm Dragonfly with Coinbase Ventures and North Island Ventures participating. The platform aggregates access to 200+ AI models with client-side encryption and no server-side data storage, and says it's already profitable at more than $70 million in annualized revenue with 3 million active users.

@Trace_Cohen·t@nyvp.com