On November 20, 2026, AWS said Amazon EC2 T9g instances are generally available. The new burstable, general-purpose sizes run on AWS Graviton5 processors and the sixth-generation AWS Nitro System. AWS positions them for workloads that need only low to moderate CPU most of the time, and it says they deliver up to 45% better price performance than the previous-generation T4g instances.
This write-up stays with that What’s New post and the official T9g product page. It does not invent On-Demand prices, a complete language or AMI matrix, or Regions beyond the list in the announcement.
What T9g is for
T9g comes in four sizes: nano, micro, small, and medium. The What’s New post lists intended uses as data processing, login gateways, small databases, microservices, low-traffic websites, freemium services, staging and development, and short-duration compute such as batch processing, event-driven functions, and CI/CD pipelines.
Compared with T4g, AWS says T9g delivers up to 2.3x the compute performance, 1.5x higher network bandwidth, and 2.8x higher EBS bandwidth. The product page adds a footnote on the price-performance claim: it is based on On-Demand pricing and real-world compute benchmarks in Standard mode, and price performance at sustained utilization varies with Unlimited mode charges. AWS also calls T9g the best price-performance option among general-purpose EC2 instances, with the same footnote. Those are AWS’s published comparisons, not independent tests.
For workloads that need larger sizes or higher sustained compute, the product page points to Amazon EC2 M9g instances, which AWS says provide up to 30% better price performance than previous-generation equivalent instances and can scale up to a 48xlarge size. That is a separate instance family; this article does not treat the T9g announcement as an M9g launch note.
CPU credits, Unlimited, and Standard mode
T9g uses the same CPU credit system as earlier T instances, including Standard and Unlimited modes. Instances provide a baseline level of CPU performance and can burst above that baseline. They accumulate CPU credits when utilization is below the baseline and spend credits when it is above. One CPU credit equals one vCPU running at 100% utilization for one minute, according to the product page.
T9g instances launch in Unlimited mode by default. AWS says that for most general-purpose workloads, Unlimited mode provides ample performance without additional charges: if average CPU utilization stays below the baseline over a 24-hour period, the hourly instance price covers interim spikes. Workloads that need to run at higher CPU utilization for a prolonged period incur a small additional charge per vCPU-hour. Customers can instead choose Standard mode, in which the instance bursts only until it exhausts its earned credits. Existing T customers can move to T9g by changing the instance type.
Official size specifications
The T9g product page publishes the following specifications. Network and EBS figures are burst bandwidth.
| Size | vCPUs | Memory (GiB) | Baseline performance / vCPU | CPU credits earned / hour | Network burst (Gbps) | EBS burst (Gbps) |
|---|---|---|---|---|---|---|
| T9g.nano | 2 | 0.5 | 5% | 6 | Up to 7.5 | Up to 6 |
| T9g.micro | 2 | 1.0 | 10% | 12 | Up to 7.5 | Up to 6 |
| T9g.small | 2 | 2.0 | 20% | 24 | Up to 7.5 | Up to 6 |
| T9g.medium | 2 | 4.0 | 20% | 24 | Up to 7.5 | Up to 6 |
T9g.micro and T9g.small are available under the AWS Free Tier. The announcement does not place T9g.nano or T9g.medium in the Free Tier.
Nitro, isolation, purchase options, and Regions
The What’s New post says T9g instances are built on the sixth-generation AWS Nitro System and feature the Nitro Isolation Engine, which AWS describes as using formal verification to provide mathematical assurance that customer workloads are isolated from each other and from AWS operators. The product page describes Nitro as offloading traditional virtualization functions to dedicated hardware.
T9g instances can be purchased with Savings Plans, On-Demand, and Spot. AWS did not publish a price table in the What’s New post or on the T9g product page excerpt used here, so this article does not invent hourly rates.
According to the November 20 What’s New post, T9g is available in US East (N. Virginia, Ohio), US West (N. California, Oregon), Canada (Central), South America (São Paulo), Europe (Frankfurt, Spain), and Asia Pacific (Malaysia, Singapore, Sydney, Tokyo). This article does not add Regions that are absent from that list.
How this differs from other recent AWS news
T9g is a burstable EC2 instance-family launch, not a change to object storage, Lambda runtimes, or the Quick workspace. Amazon S3’s higher general-purpose read request rate is a storage throughput change. Lambda’s Node.js 26 runtime applies to functions, not to these EC2 sizes. AWS Quick’s new web app and plugins are an AI workspace update.
Key facts
| Item | What official sources say |
|---|---|
| Announcement | AWS What’s New post dated November 20, 2026, plus the official Amazon EC2 T9g product page. |
| Status | Generally available. |
| Processor and platform | AWS Graviton5 processors; sixth-generation AWS Nitro System and Nitro Isolation Engine. |
| Sizes | Nano, micro, small, and medium only, as published. |
| AWS comparison vs T4g | Up to 45% better price performance; up to 2.3x compute, 1.5x network bandwidth, and 2.8x EBS bandwidth. Price-performance footnote applies to Standard-mode On-Demand benchmarks. |
| Credits | Same T-family credit system; Unlimited mode by default; Standard mode optional. |
| Free Tier | T9g.micro and T9g.small. |
| Purchase options | Savings Plans, On-Demand, and Spot. |
| Regions in the What’s New post | US East (N. Virginia, Ohio), US West (N. California, Oregon), Canada (Central), South America (São Paulo), Europe (Frankfurt, Spain), and Asia Pacific (Malaysia, Singapore, Sydney, Tokyo). |
FAQ
When did AWS announce EC2 T9g instances?
AWS published the general availability What’s New post on November 20, 2026. The official T9g product page carries the instance specifications quoted in this article.
What does T9g run on?
AWS says T9g instances are powered by AWS Graviton5 processors and built on the sixth-generation AWS Nitro System, including the Nitro Isolation Engine.
Which T9g sizes are available?
Four sizes: nano, micro, small, and medium. AWS has not listed larger T9g sizes in these sources. For larger or more sustained general-purpose compute, the product page points to the separate M9g family.
How do T9g CPU credits work?
T9g uses the same credit system as earlier T instances. It accumulates credits below the baseline and spends them above the baseline. Instances launch in Unlimited mode by default; Standard mode stops bursting when earned credits run out.
Which sizes are in the AWS Free Tier?
T9g.micro and T9g.small. The announcement does not include T9g.nano or T9g.medium in the Free Tier.
Did AWS publish T9g prices?
The What’s New post and the product page describe purchase options — Savings Plans, On-Demand, and Spot — and a relative price-performance claim versus T4g. They do not include a dollar price table in the material used for this article, so no hourly rates are listed here.
Where is T9g available?
The November 20 What’s New post lists US East (N. Virginia, Ohio), US West (N. California, Oregon), Canada (Central), South America (São Paulo), Europe (Frankfurt, Spain), and Asia Pacific (Malaysia, Singapore, Sydney, Tokyo). Check the latest AWS documentation if you need a Region that is not on that list.
Is T9g a drop-in upgrade from T4g?
AWS says existing T customers can upgrade by changing the instance type and keep the same Standard and Unlimited credit modes. Confirm image, architecture, and application compatibility for Graviton5 in your own environment before you move production traffic.

Leave a Reply
You must be logged in to post a comment.