Pomegra
Together AI's Aramco-led round comes with 500 MW of separately committed compute capacity, signaling the open-source inference market is competing directly with hyperscalers for enterprise AI workloads.
- Together AI raised $800M at an $8.3B valuation in a Series C led by Aramco Ventures, announced July 1, 2026.
- Investors committed over 500 MW of compute capacity independently, not drawn from the company's balance sheet.
- The round values Together AI more than 2.5x above its $3.3B Series B valuation from February 2025.
Lead
Together AI, the open-source AI inference platform, closed an $800 million Series C round on July 1, 2026, at a valuation of $8.3 billion. Aramco Ventures - the investment arm of Saudi Aramco - led the deal. The round also secured separate commitments for over 500 megawatts of compute capacity, structured outside the company's balance sheet and capitalized independently by investors.
What Does Together AI Actually Do?
Together AI provides inference infrastructure for open-source AI models, letting enterprises run frontier-class models without routing workloads through closed, proprietary APIs from OpenAI, Google, or Anthropic. Its platform handles the compute-heavy task of serving large language models at scale, positioning it as a neocloud alternative to hyperscaler AI offerings.
The company's pitch is straightforward: open-source models have closed much of the performance gap with proprietary ones, but running them reliably at enterprise scale requires infrastructure that most organizations cannot build alone. Together AI sells that infrastructure.
Why Is Aramco Leading This Round?
Aramco Ventures leading a deep-tech AI infrastructure deal reflects a broader Saudi strategy to diversify away from hydrocarbon revenue by seeding critical compute capacity. The kingdom has been aggressive in AI infrastructure commitments throughout 2025 and 2026, and anchoring a round in a company that controls both software and physical compute fits that pattern.
The 500 MW compute commitment - structured separately from the equity round - is the more unusual element. A single large hyperscaler campus draws roughly 100 to 500 megawatts. Locking in that capacity at formation, before it appears on Together AI's books, keeps the company's balance sheet light while securing supply for multi-year enterprise contracts.
Investor Lineup
The round drew a broad mix of financial and strategic investors. Vista Equity Partners and General Catalyst anchored the financial side. NVIDIA, Salesforce Ventures, Emergence Capital, Lux Capital, March Capital, Pegatron, Geodesic, PSP Partners, and SE Ventures - the venture arms of both SentinelOne and Schneider Electric - also participated.
NVIDIA's presence matters: Together AI's infrastructure depends on GPU supply, and a strategic investment from the dominant chip supplier creates alignment on both pricing and hardware allocation. Schneider Electric's involvement signals the power and data center infrastructure angle explicitly.
What Does the Valuation Imply About the Last Round?
The jump from a $3.3 billion Series B valuation in February 2025 to $8.3 billion roughly 16 months later is steep - a 2.5x step-up that implies meaningful revenue growth or a significant re-rating of AI infrastructure multiples, most likely both. Inference-as-a-service businesses benefit as model usage compounds. The revenue base that justifies each new valuation tier matters more than the headline multiple, and Together AI has not disclosed current ARR.
Still, the investor roster suggests this is not a story built purely on market enthusiasm. Strategic participants with direct supply-chain relationships - NVIDIA, Schneider Electric - are writing checks that imply operational alignment, not just financial exposure.
Scale Ambitions
Together AI has stated plans to grow its infrastructure footprint roughly 50-fold over the next five years. That goal requires both capital and power - two resources the Series C appears designed to secure simultaneously. The 500 MW commitment gives the company credibility with enterprise buyers evaluating whether an inference provider can sustain long-term workloads.
The timing also aligns with a wider shakeout in the neocloud segment. Several GPU cloud startups that emerged in 2023 and 2024 have struggled to convert capacity into stable, recurring enterprise revenue. Together AI's focus on open-source inference - rather than raw GPU rental - gives it a higher-margin angle and a more defensible product layer.
Outlook
Together AI enters the second half of 2026 with a war chest large enough to compete with the infrastructure spend of midsize hyperscalers, and with physical compute commitments that de-risk its supply chain. The core question is whether enterprise adoption of open-source inference accelerates fast enough to justify the 50-fold growth target. The Series C suggests a credible cohort of investors believes it will. The 500 MW side arrangement is the more telling signal - that is infrastructure capital, not speculation capital, and it bets on durable, large-scale demand.
