Together AI Raises $800M at $8.3B Valuation as Neocloud Bets Heat Up

TL;DR: Together AI closed an $800 million Series C on July 1, 2026, at an $8.3 billion valuation, roughly 2.5x its $3.3 billion mark from February 2025. Aramco Ventures led the round. Nvidia, Vista Equity Partners, General Catalyst, Salesforce Ventures, Emergence Capital, March Capital, Pegatron, and SentinelOne's S Ventures all wrote checks. The company says its annual bookings crossed $1.15 billion last quarter.

According to TechCrunch, Together AI raised the $800 million Series C at an $8.3 billion post-money valuation, more than doubling the value investors assigned the company just 16 months earlier. The deal closed Wednesday, July 1, and Together AI announced it the same day.

I want you to sit with one number before we go further: $8.3 billion. That's the price tag on a company that rents out Nvidia GPU clusters and other AI infrastructure so other startups don't have to buy their own. Together AI doesn't build a frontier model you've heard of. It doesn't have a consumer app. It sells the picks and shovels behind open-source AI, and investors just told you that business is worth more than most publicly traded regional banks.

The Deal Mechanics: Who Put In Money and Why

Aramco Ventures led the round through its Prosperity7 Ventures arm, the Diversified Venturing Program of Saudi Aramco's investment vehicle. That's not a typo. The world's largest oil company, through its venture arm, just anchored an $800 million check into a company that resells Nvidia compute. Abhishek Shukla, Managing Director of Prosperity7 Ventures US, put it plainly in the company's own announcement, carried by Morningstar via Business Wire: "Together has built the platform that makes open source models genuinely usable at enterprise scale, and the team's ambition matches the scale of the opportunity in front of them."

The rest of the syndicate reads like a venture capital greatest-hits list: Vista Equity Partners, General Catalyst, Emergence Capital, Nvidia itself, Salesforce Ventures, March Capital, Pegatron, and S Ventures, the strategic vehicle backed by cybersecurity firm SentinelOne. Several of these names already had positions in Together AI, a lineup confirmed independently by SiliconANGLE. General Catalyst led the $305 million Series B in February 2025. Nvidia and Emergence Capital were in the $102.5 million Series A back in 2023, led by Kleiner Perkins, per TechCrunch's reporting. This is a round dominated by insiders doubling down, not new money discovering the company for the first time. That matters, and I'll get to why in a minute.

Founded in 2022 by Vipul Ved Prakash, Percy Liang, and Ce Zhang, Together AI calls itself a "neocloud," a term for infrastructure providers that rent GPU capacity and AI tooling without owning the hyperscale data center footprint of Amazon Web Services or Microsoft Azure. Prakash previously sold his startup Topsy, a social media search platform, to Apple in 2013 for a reported $200 million-plus. Liang is a Stanford computer science professor. Zhang holds a joint appointment at ETH Zürich and the University of Chicago. This isn't a founding team learning venture math for the first time.

The company will use the capital to expand its inference business, the process of running already-trained AI models in production, and to scale computing capacity roughly 50-fold over the next five years, according to the official press release. Together AI's pitch to customers: open-source models like DeepSeek, Nemotron, MiniMax, and Kimi deliver comparable performance to closed frontier models such as GPT or Claude at a fraction of the token cost. The company claims customers see 6x to 60x cost savings. It names Decagon as an example, saying the customer-service AI company cut inference costs sixfold after migrating.

Why This Round Matters More Than the Headline Number

Here's the part that should catch your attention if you're an accredited investor weighing venture exposure in 2026. This round wasn't the round Together AI originally went out to raise. According to Data Center Dynamics, citing a March 2026 report from The Information, Together AI was in talks to raise approximately $1 billion at a $7.5 billion pre-money valuation. It ended up taking $800 million, or 20% less capital, and closing at an $8.3 billion post-money valuation, which is higher than the target it was reportedly shopping four months earlier.

Read that twice. A company sought $1 billion, took $800 million instead, and still landed a richer valuation than the one it was pitching in March. That's not a company running out of leverage. That's a company that had multiple term sheets and could afford to give less equity away for a better price. In a market where a lot of AI startups are begging investors to just show up, Together AI apparently had the opposite problem: too much demand for the round.

Jeff's take: this is the clearest signal I've seen this quarter that the AI infrastructure layer, specifically the "arms dealer" tier that sells compute and tooling rather than building the models themselves, is where institutional capital wants to sit right now. Together AI reports annual bookings crossed $1.15 billion last quarter, per Reuters' coverage via AOL, up from an annualized revenue run rate the company put at roughly $1 billion as of February 2026, according to research firm Sacra. If those bookings figures hold, Together AI is trading at roughly 7.2x forward revenue. That's not cheap, but it's a lot less frothy than the 15x-plus multiples some foundation model labs command on far less predictable revenue.

This deal also fits a pattern. Together AI isn't the only neocloud raising money at a blistering pace. The same week saw a wave of AI-adjacent rounds close across categories, and TechCrunch notes that Upscale AI closed a Series A-plus extension totaling $500 million at a $2 billion valuation last month, while TensorWave, which focuses on AMD GPU clusters, raised a $350 million Series B at a $1.55 billion valuation in the same window. Three infrastructure resellers, three nine-figure-plus rounds, all within about 30 days of each other. Our Q2 2026 recap found AI took 81% of all venture dollars deployed, and this round is a data point confirming that concentration didn't ease up in the back half of the quarter. It got sharper, and it moved further downstream from model-builders into the infrastructure and tooling layer underneath them.

The Contrarian Case: What Could Go Wrong Here

I'll play it straight with you. Neoclouds carry a structural risk that model labs don't: they're capital-intensive resellers sitting between Nvidia's supply and enterprise demand, and both ends of that trade can move against them. If Nvidia decides to sell directly to more enterprise customers, or if hyperscalers like AWS and Azure drop pricing to compete for the same open-source workloads Together AI is chasing, the company's margin gets squeezed from two directions at once. Together AI doesn't manufacture chips. It leases them, packages them with software, and resells capacity. That's a real business, but it's not a moat the way owning a foundation model with a defensible dataset can be.

There's also the valuation math itself. Together AI went from a $1.25 billion valuation in March 2024 to $3.3 billion in February 2025 to $8.3 billion now, a roughly 6.6x increase in under 28 months. Revenue grew fast too, but not quite as fast as the valuation multiple implies if you're being conservative. Sacra pegged annualized revenue at about $618 million at the end of 2025, climbing to roughly $1 billion by February 2026. That's real, strong growth. It is not, however, growth that obviously supports a valuation more than 8x that revenue figure without assuming the current trajectory holds for years, in a market where open-source model economics could shift overnight if a major lab drops its own inference pricing.

And remember: the insiders who priced this round, Nvidia and General Catalyst chief among them, have every incentive to keep marking Together AI up. Nvidia sells the chips Together AI rents out. A higher Together AI valuation validates Nvidia's own thesis that GPU demand keeps compounding. That's not a conspiracy, it's just how strategic investing works, but you should discount insider-led up-rounds slightly relative to rounds where a genuinely new, disinterested investor sets the price. This round had one new name at the top, Aramco Ventures, and a long list of people marking their own homework underneath it.

What Accredited Investors Should Watch For

• Revenue-to-valuation ratio at the next round. If Together AI raises again within 12-18 months at a valuation north of $15 billion without annual bookings crossing roughly $2 billion, treat that as a warning sign of multiple expansion outrunning fundamentals, not durable value creation.
• Nvidia's dual role. Nvidia is both Together AI's supplier and an investor across three consecutive rounds now. Watch whether Nvidia's own earnings calls mention direct-to-enterprise GPU cloud offerings that would compete with the neoclouds it's funding.
• Customer concentration disclosures. The company names Cursor, Cognition, and Decagon as marquee customers but hasn't disclosed what percentage of the $1.15 billion in bookings comes from its top five accounts. Concentrated bookings are a real risk in infrastructure resale businesses.
• SPV and fund access terms. If you're accessing this deal or similar neocloud exposure through a special purpose vehicle or a venture fund, ask about the markup between what the fund paid and what you're paying to get in. Late-stage SPVs on hot rounds routinely carry a 10-20% premium layered on top of the round price.
• Down-round protection language. At an 8.3x step-up from the 2024 valuation, ask whether your entry point includes standard anti-dilution protection if a future round prices lower. Companies moving this fast on paper valuation can also correct fast.
• Whether this is dilutive or debt-backed growth. Infrastructure buildouts of this scale sometimes get financed partly through venture debt rather than pure equity. If you want the mechanics of how that works, read our explainer on venture debt and non-dilutive funding before assuming every dollar in a round like this came from equity investors.

If you're trying to get exposure to this tier of the market without writing an $800 million check yourself, the paths are narrower than they look. Direct access to a round like Together AI's Series C typically requires being an existing LP in one of the participating funds, or getting into a secondary SPV after the fact, often at a markup. Our piece on emerging managers and accredited investor access covers some of the more realistic entry points for individual investors who aren't already LPs at General Catalyst or Kleiner Perkins scale. And if you're evaluating any fund's track record on deals like this, remember that headline IRR on a paper markup means nothing until cash actually comes back to LPs. Our explainer on DPI versus TVPI covers exactly this trap.

Frequently Asked Questions

Q: What is a "neocloud" and how is it different from AWS or Azure?

A neocloud is a company that rents out GPU computing capacity and AI-specific infrastructure without owning the massive, general-purpose data center footprint of a hyperscaler like Amazon Web Services, Microsoft Azure, or Google Cloud. Neoclouds like Together AI, CoreWeave, and Lambda typically lease or buy Nvidia GPUs in bulk, then resell access to that compute, often bundled with software for training and running AI models, at prices competitive with or below what the hyperscalers charge for equivalent AI workloads.

Q: Why would Saudi Aramco's venture arm invest in an AI infrastructure startup?

Aramco Ventures, through its Prosperity7 Ventures program, has been diversifying beyond oil and gas for several years, with AI infrastructure and data center-adjacent investments as a stated priority. Energy and AI compute are increasingly linked: data centers need enormous amounts of power, and companies like Aramco see strategic value in owning equity stakes in the AI infrastructure buildout rather than just supplying the electricity or feedstock behind it. Prosperity7 was also already a Together AI investor at the Series B stage.

Q: Is Together AI profitable?

Together AI has not disclosed profitability figures publicly. The company reports annual bookings crossed $1.15 billion last quarter, but bookings represent contracted revenue, not necessarily cash collected or profit after infrastructure costs. Neoclouds carry heavy capital expenditure for GPUs and data center capacity, and it is common for companies at this stage to prioritize growth and market share over near-term profitability. Accredited investors evaluating similar deals should ask directly about gross margin, not just top-line bookings.

Q: How does this compare to other AI infrastructure deals happening right now?

Together AI's $800 million round is one of several large neocloud raises in mid-2026. Upscale AI closed a $500 million round at a $2 billion valuation, and TensorWave raised $350 million at a $1.55 billion valuation, both within roughly a month of Together AI's announcement. The pattern lines up with broader data on AI's share of total deal volume this year, which we cover in depth elsewhere.

Author Disclosure: Jeff Barnes, MBA has no personal position in any company, fund, or platform named in this article. Angel Investors Network has no current commercial relationship with any party mentioned. AIN provides marketing and education services, not investment advice. Past performance does not guarantee future results. All investments involve risk, including loss of principal.