Together AI, the four-year-old startup that has quietly become one of the most-used engines behind open source artificial intelligence, has raised an $800 million Series C at an $8.3 billion valuation. The company announced the round on July 1, and the numbers land as a clear statement about where a growing slice of enterprise AI spending is heading: away from the priciest closed models and toward the leaner, cheaper open source stack that Together AI rents out by the hour.
%20002.png){kind=link}
The financing was led by Aramco Ventures, the venture arm of the Saudi energy giant, with participation from a notably broad syndicate: Vista Equity Partners, General Catalyst, Emergence Capital, Nvidia, March Capital, Pegatron, and SentinelOne's S Ventures, among others. It is the kind of investor list that blends deep-pocketed strategic backers with the chip supplier whose hardware sits at the center of Together AI's business.
Founded in 2022, Together AI operates what the industry has taken to calling a "neocloud" — a specialized cloud provider that rents Nvidia GPU clusters and other AI-specific infrastructure rather than the general-purpose computing offered by the hyperscalers. Its pitch is straightforward: give developers a place to run, fine-tune, and serve open source models at a fraction of the cost of paying per-token premiums for closed frontier systems.
Why It Matters
The valuation jump tells its own story. Together AI last raised a $305 million Series B at a $3.3 billion valuation roughly 16 months earlier, in early 2025. More than doubling that figure in little over a year puts the company among the fastest-appreciating infrastructure names in the AI buildout — and it did so, according to reporting, by taking less money than the roughly $1 billion it had reportedly sought in the spring while securing a higher valuation than the $7.5 billion floated at the time.
.jpg){kind=link}
Behind the numbers is a shift in how companies actually buy AI. Together AI claims annual bookings of more than $1.15 billion as of its most recent quarter, and it credits the surge to businesses increasingly reaching for capable open source models instead of paying the premium on tokens for closed frontier models across every workload. The company says open source usage has roughly tripled across the industry over the past year, a trend echoed by adjacent players such as AI gateway OpenRouter.
That matters because it reframes the competitive map. For most of the past two years the narrative centered on which lab had the single most powerful model. The neocloud thesis is different: as open weights close the quality gap for a widening set of tasks, the strategic value shifts toward whoever can serve those models most efficiently, reliably, and cheaply at scale.
The Reaction
The syndicate itself is the loudest signal. Aramco Ventures leading the round underscores how sovereign-linked capital is treating AI compute as core infrastructure worth owning a piece of, not a passing software cycle. Nvidia's repeat participation — it also backed Together AI's 2023 Series A alongside Kleiner Perkins and Emergence Capital — reflects the chipmaker's habit of investing in the customers who buy its silicon, deepening a relationship that runs in both directions.
Customers appear to be voting with their workloads. Together AI says it has thousands of paying customers and names fast-growing developer-facing companies including Cursor, Cognition, and Decagon among them. Those are exactly the kinds of AI-native businesses whose economics live and die on inference cost, and their presence lends weight to the argument that open source serving is becoming a default layer rather than a budget alternative.
The founding team gives the round additional credibility with investors. Together AI was co-founded by Vipul Ved Prakash, who previously sold social search startup Topsy to Apple in 2013, alongside Stanford professor Percy Liang and Ce Zhang, an associate professor affiliated with ETH Zürich and the University of Chicago. That academic-plus-operator pedigree has helped position the company at the intersection of open research and commercial infrastructure.
What's Next
Together AI says it will use the capital to expand its product lineup and scale its computing capacity, which it expects to grow roughly 50-fold over the next five years. That is an aggressive target, and it points to the central constraint of the neocloud business: growth is gated by access to power, data-center space, and the steady supply of high-end accelerators that make the whole model work.
The company is far from alone in chasing that opportunity. Neoclouds have become a favored destination for venture capital more broadly — Upscale AI recently reached roughly $500 million in total funding at a $2 billion valuation, and TensorWave, which focuses on AMD-powered GPU clusters, raised a $350 million Series B at a $1.55 billion valuation. The category is filling with well-capitalized specialists, which means Together AI's edge will rest on efficiency, software, and the breadth of models it can serve rather than on being first.
The open question is durability. If open source models keep closing the gap with closed frontier systems, the neocloud tailwind strengthens. If the frontier labs pull decisively ahead again, or if the hyperscalers undercut specialists on price, the thesis gets harder. For now, the bookings trajectory and the caliber of the syndicate suggest investors are betting the open source wave has room to run.
Closing Thoughts
Together AI's raise is less about one company and more about a structural bet on how AI gets consumed. The early chapters of the generative AI era were written by the labs building the biggest models. The next chapter is increasingly being shaped by the infrastructure layer that decides whether those models — open or closed — can be run affordably enough to reach everyday products.
An $8.3 billion valuation for a company that mostly rents out someone else's chips would have sounded implausible a few years ago. It sounds a lot less so now that open weights have become a serious commercial option and inference cost has become a boardroom concern. Whether the neocloud model becomes a lasting pillar or a stepping stone, Together AI's Series C is a useful marker of where the money — and the workloads — are moving in mid-2026.
한글 요약
오픈소스 인공지능을 구동하는 핵심 인프라 업체로 자리 잡은 투게더 AI(Together AI)가 7월 1일 8억 달러 규모의 시리즈 C 투자를 유치했다고 발표했다. 기업가치는 83억 달러로 평가됐으며, 사우디 에너지 대기업의 벤처 부문인 아람코 벤처스가 라운드를 주도하고 비스타 에쿼티, 제너럴 캐털리스트, 엔비디아, 페가트론 등이 참여했다. 2022년 설립된 이 회사는 엔비디아 GPU 클러스터를 시간 단위로 빌려주는 '네오클라우드' 사업자로, 폐쇄형 프론티어 모델의 비싼 토큰 요금 대신 저렴한 오픈소스 모델을 실행·서빙할 수 있는 환경을 제공한다.
이번 투자는 기업의 AI 구매 방식 변화를 보여준다. 투게더 AI는 지난 분기 기준 연간 예약 매출이 11억5천만 달러를 넘어섰다고 밝혔으며, 오픈소스 모델 사용량이 업계 전반에서 지난 1년간 약 세 배로 늘었다고 설명했다. 약 16개월 전 33억 달러 가치로 시리즈 B를 유치했던 것과 비교하면 기업가치가 두 배 이상 뛴 셈이다. 커서(Cursor), 코그니션(Cognition), 데카곤(Decagon) 등 추론 비용에 민감한 AI 네이티브 기업들이 주요 고객으로 이름을 올렸다.
회사는 이번 자금을 제품 확장과 컴퓨팅 용량 확대에 투입해 향후 5년간 연산 능력을 약 50배로 늘리겠다는 목표를 제시했다. 다만 업스케일 AI, 텐서웨이브 등 자본력을 갖춘 네오클라우드 경쟁사들이 빠르게 늘고 있어, 투게더 AI의 강점은 '최초'보다는 효율성과 지원 모델의 폭에서 판가름날 전망이다. 오픈소스 모델이 폐쇄형과의 격차를 계속 좁힐지가 이 사업 모델의 지속성을 가를 핵심 변수로 꼽힌다.