Daily Funding Roundup:
Apr 9, 2026
Nvidia's hand is in everything. SiFive raised $400M at $3.65B for RISC-V chips with Nvidia participation. Australian AI data center builder Firmus pulled in $505M at $5.5B (also Nvidia-backed). Ex-OpenAI researchers at Applied Compute raised $80M Series A at $1.3B unicorn status. OpenAI eyes a $1 trillion IPO. Markets digested the ceasefire as oil swung between $97 and $100.
Rounds
RISC-V processor IP company SiFive raised $400M Series G at a $3.65B valuation, led by Atreides Management with Nvidia participation. The funding will accelerate development of high-performance RISC-V CPUs for AI data centers, positioning SiFive as the open-standard alternative to Arm Holdings. SiFive does not sell chips; it sells blueprints that customers like Google customize for their own internal designs. The strategic timing is critical: Arm unveiled its own chips last month, transforming from neutral IP licensor to potential competitor to its longtime customers. SiFive is pitching itself as the safe alternative for hyperscalers who do not want to be dependent on Arm's increasingly competitive trajectory.
Australian AI data center builder Firmus raised $505M at a $5.5B valuation, the third equity round in six months bringing total recent fundraising to $1.35B. Project Southgate aims to build multiple AI-focused data centers across Australia in collaboration with Nvidia and CDC Data Centres. Firmus is preparing for an ASX IPO later in 2026 backed by a $10B Blackstone debt facility. The Asia-Pacific AI infrastructure thesis is becoming a venture-scale category: Australia's grid capacity, regulatory environment, and proximity to growing Asian AI demand position it as a key supply node for the global AI buildout.
Ex-OpenAI researchers' enterprise AI agent company Applied Compute raised $80M Series A at a $1.3B valuation, just months after closing an $80M seed. The company calls its approach 'Specific Intelligence,' focused on bridging the gap between smart and useful for production enterprise workflows. The valuation jump (effectively pricing the seed at ~$200-300M and the Series A at $1.3B) reflects the AI talent premium for OpenAI alumni and the bet that enterprise agent infrastructure will be the next category to produce generational outcomes. Total raised: $160M in roughly 18 months.
News & Signals
Markets digest the ceasefire: oil swings as Strait reopening proves messy
Wednesday's euphoria gave way to Thursday's uncertainty. The S&P 500 added 0.6% to 6,824, the Dow gained 275 points, and the Nasdaq climbed 0.8%, but oil prices were volatile. WTI surged above $100 intraday before settling at $97.87 (up 3%) as traders learned that Iran was limiting traffic through the Strait of Hormuz despite the ceasefire agreement. The two-week pause is technically holding, but the practical reality is messier than the headline: tankers are not yet flowing freely, insurance markets are still pricing war risk premiums, and Iran is asserting partial control over Strait traffic. A glimmer of additional good news: Israel agreed to direct talks with Lebanon, suggesting the broader Middle East de-escalation might be more comprehensive than just the U.S.-Iran component.
OpenAI eyes $1 trillion IPO valuation, plans retail allocation
The Information reported that OpenAI is laying the groundwork for a U.S. listing that could value the company at up to $1 trillion. CFO Sarah Friar confirmed that OpenAI plans to reserve part of its eventual IPO for individual investors, building on the $3 billion raised from retail in the recently closed $122B private round. At $1T, OpenAI would be worth more than every other AI company combined and would represent the largest IPO in history by a factor of more than 13x (vs Saudi Aramco's $25.6B). The CFO concerns reported earlier this week (about Altman's $600B 5-year spend plan) have apparently been resolved or set aside as the IPO timeline accelerates following the ceasefire-driven market recovery.
Nvidia is now the dominant strategic investor in AI infrastructure
Today's three biggest deals all share a common thread: Nvidia is an investor in all three. SiFive (RISC-V chips), Firmus (data centers), and earlier this week ThinkLabs AI (grid optimization), Cognichip (chip design), and Aria Networks (data center networking via Sutter Hill, an indirect Nvidia ally). Nvidia's NVentures portfolio is rapidly becoming the most consequential corporate venture arm in tech, displacing GV and Microsoft's M12 in importance for AI infrastructure investments. The strategic logic: Nvidia wants to ensure that the entire AI stack (from chips to data centers to networking to power) is built in ways that maximize Nvidia GPU adoption. Strategic investing achieves this more cheaply than building everything in-house.
Amazon's AI cloud business hits $15B run rate, chip business doubles to $20B
Amazon disclosed that AWS's AI revenue run rate exceeded $15 billion in Q1 2026, and that Amazon's chip business (Graviton CPUs and Trainium AI accelerators) reached an annual run rate above $20 billion, roughly double the figure cited earlier this year. This is significant for two reasons. First, it demonstrates that hyperscalers are not just buying Nvidia chips, they are also building competitive in-house silicon at scale (Graviton/Trainium) and generating real revenue. Second, it justifies Meta's announced 2026 capex of $115-135 billion (nearly double 2025) and Microsoft's expected matching investment. The hyperscaler AI infrastructure spending cycle is entering a phase where the unit economics are visible in revenue line items, not just capex projections.
VC Mood on X
Thursday's mood was a step down from Wednesday's euphoria but still firmly positive. The day's three big deals (SiFive, Firmus, Applied Compute) totaling $985M demonstrated that the deal pipeline that had been frozen during the war is rapidly thawing. "Every late-stage company that paused fundraising in March is now in active discussions," one growth investor posted. The pattern: companies that had term sheets in hand on February 27 (the day before the war began) are quickly closing those rounds at the originally agreed terms, while companies that were in early discussions are restarting from scratch.
The Nvidia backing pattern dominated the technical discussion. "Nvidia is in 2 of 3 today's deals. They are now the most consequential strategic investor in AI infrastructure, full stop," one infrastructure investor posted. Several VCs noted that being able to claim Nvidia as an investor has become a strong signal for any AI infrastructure startup, sometimes outweighing traditional VC firm names. The downside: Nvidia's strategic interests do not always align with founders' interests, and accepting Nvidia capital can constrain future options (acquisitions by Nvidia competitors become harder, for example).
Applied Compute at $1.3B valuation provoked the most debate. "Eight-month-old company, ex-OpenAI founders, $80M seed and $80M Series A at $1.3B. We are back to peak 2021 valuations for AI talent," one seed investor posted. The bullish view: ex-OpenAI researchers have demonstrated track records in foundation model training and applied AI deployment, and the talent pool that can credibly build production-grade enterprise agents is fewer than 200 people globally. The bearish view: Applied Compute has not yet shipped a product, and a $1.3B valuation pre-revenue requires assumptions about future market share that may not hold up if the enterprise agent category fragments. Both views are reasonable. Time will tell.
Rounds and signals sourced from SEC filings, press releases, and verified news reports. All amounts in USD unless noted. Reporting reflects information available at time of publication.