Open source AI just received a massive vote of confidence. Together, AI raised $800 million on July 1, 2026, according to a report from TechCrunch. This brought its valuation to $8.3 billion, giving founders a new reason to bet on open models. The round was led by Aramco Ventures, with support from NVIDIA, Vista Equity Partners, General Catalyst and Salesforce Ventures.
Here’s why you should care. Together AI runs a cloud that allows companies to train and deploy AI on open models like Llama, Mistral and Qwen. This means cheaper AI, less lock-in, and more control over your data. For a young founder counting every dollar, this change changes what you can build this year.
“Usage of open models has tripled and customers have reduced inference costs up to 60x compared to closed models.”
As part of the $800 million round
The numbers are huge and tell a story. Together AI recorded annual bookings in excess of $1.15 billion in its most recent quarter. This puts a company that most people hadn’t heard of in the same revenue bracket as established business software names.
Demand is the reason. The company claims that usage of open models has tripled and customers have significantly reduced inference costs, up to 60 times to be exact, compared to closed models. When computation becomes much cheaper, more ideas become fundable. I used to tell founders to wait for better tools. Now that the tools are here, the wait is over.
| Metric | Figure |
|---|---|
| Round size | 800 million dollars |
| Assessment | $8.3 billion |
| Annual reservations | More than $1.15 billion |
| Validated calculation | More than 500 megawatts |
Backers matter as much as the title number. Aramco Ventures contributes a large amount of energy and capital, while NVIDIA provides the chips that make large-scale inference possible. When strategic investors like these align behind open models, it signals that change is structural, not fleeting. For founders, this is a green light to confidently plan for open tools.
Why Open Source AI Has Become Cheaper for Founders
Closed models are powerful, but they come with a bill and a leash. You pay per token and build on someone else’s terms. Open weight models flip this script. You keep your data, you adjust the model and you avoid surprise price increases.
It’s the same logic that explains why the strongest arguments now lead with urgency, a point we addressed when discussing seed funding in 2026. Cheaper AI infrastructure is a “why now” in itself. Because costs are falling rapidly, the window to launch something lean is now open.
Consider what a 60x cost reduction really means. A workload that once cost thousands of dollars per month in closed model fees could drop to the price of a modest software subscription. This is not a rounding error. For a first team, this can mean the difference between delivering a feature and shelving it.
So the takeaway is simple. If your product needs AI, evaluate an open model before committing to a closed model. The savings can fund another rental or another month of track.
How start-ups can implement open models
Start small and specific. Choose a time-wasting workflow and test an open model against it. Support responses, data cleansing, and early release content are all good candidates.
Then measure honestly. Compare quality, speed and cost with your current tool. Founders winning with AI see it as a practical lever, not a magic wand, a theme we explored in AI for small businesses. Because the barrier keeps dropping, your first release may ship sooner than expected.
Here’s a simple test to do this week. Take a repetitive task, give it to an open model for three days, and record the hours you save. If the results hold up, expand slowly and reinvest those savings into products or people. Small, measured bets beat a risky overhaul.
You can also read Together AI’s own breakdown in its C Series Announcementwhich lays out the open model strategy in simple terms. Study how they define value, then borrow the elements that fit your story.
The computing race behind the headlines
There is a quieter story to this deal. Together, AI has secured commitments for more than 500 megawatts of computing, built independently by its investors. This scale allows it to promise constant supply to enterprise customers, the kind of guarantee once reserved for AWS, Azure and Google Cloud.
For founders, this is important because reliable, affordable computing is the new oil. As money continues to flow into AI infrastructure, expect more competition and better pricing. We followed the same change through unicorn startups in 2026.
Look at two things next. First, whether open model costs will continue to fall through the end of 2026. Second, whether more startups default to open rather than closed. Both trends favor small teams with big ambitions.
Questions asked by the founders of Together AI
What does Together AI actually sell? It sells cloud access to run and refine open AI models at a lower cost than closed systems.
Why is the $800 million round a big deal? This indicates that open source AI is now a serious and well-funded option for businesses, not a choice for amateurs.
Should a small startup use open models? Often yes, because they reduce cost and lock-in, but test the quality against your specific use case first.
