Cloud bandwidth pricing usually follows one of two ideas.
One model says:
Pay only for the data you send.
The other says:
Pay for the infrastructure or network capacity, and transfer is included.
Both can be economical.
Both can also become expensive.
The difference appears when the workload moves from hundreds of gigabytes to multiple terabytes every month.
A small API may barely notice egress charges.
A video platform, backup service, software distribution system, AI media application or data pipeline may build its entire cloud economics around them.
That is why the useful question is not:
Which provider has free bandwidth?
It is:
At what traffic level does included bandwidth become cheaper than metered egress for this workload?
For the provider-by-provider rate comparison, first see Cloud Egress Costs in India. This article focuses specifically on the pricing-model decision.
There Are More Than Two Bandwidth Pricing Models
“Including bandwidth” can mean several different things.
That distinction matters before comparing providers.
| Pricing Model | How It Works | Main Cost Risk |
|---|---|---|
| Pay per GB/GiB | Outbound traffic is metered after any free allowance | Bill rises directly with traffic |
| Included transfer quota | Plan includes a fixed amount such as 1 TB; overage may be charged | Cost jumps after allowance |
| Unmetered transfer | No per-GB charge under normal terms | Port speed, fair-use or sustained-use conditions |
| Transfer included in SKU | Data transfer is included as part of a specific VM/GPU/product price | Higher base price and SKU restrictions |
| Fixed bandwidth capacity | Buyer pays for Mbps/Gbps capacity rather than data volume | Paying for unused capacity |
These models should not be treated as synonyms.
For example, unmetered means volume is not billed per GB. It does not mean the connection has infinite throughput.
Likewise, 1 TB included is not the same as unlimited data transfer.
Pay-Per-GB Egress Works Well When Traffic Is Small
For a low-egress workload, paying for a large fixed bandwidth allowance can be unnecessary.
Examples include:
- Internal dashboards
- Development environments
- Small APIs
- B2B applications with light payloads
- Early-stage SaaS products
- Admin systems
- Low-volume inference APIs
AWS currently provides 100 GB/month of internet data transfer out free across eligible AWS services and regions, excluding China and GovCloud.
Azure also includes the first 100 GB/month of internet egress. For traffic from Asia over Microsoft’s Premium Global Network, the next 10 TB is currently priced at $0.12/GB.
Google Cloud Standard Tier currently includes the first 200 GiB/month per account, with traffic from regions including Mumbai and Delhi then priced at $0.085/GiB through 10 TiB.
At a few hundred gigabytes per month, the absolute egress bill may remain small compared with compute, databases, engineering or managed-service costs.
That is where metered pricing has a clear advantage:
low traffic produces low network cost.
The Economics Change at Multi-Terabyte Scale
Metered egress scales linearly with outbound data.
That is useful when traffic falls.
It becomes more painful when traffic grows.
Azure Example at 10 TB
Azure’s current Asia Premium Global Network pricing is:
- First 100 GB: Free
- Next 10 TB: $0.12/GB
Assume 10,000 GB total monthly internet egress.
Billable transfer:
10,000 GB - 100 GB = 9,900 GB
Estimated egress charge:
9,900 × $0.12 = $1,188/month
That is internet transfer alone.
It does not include NAT Gateway processing, cross-region traffic, CDN, load balancers or taxes.
Google Cloud Standard Tier at 10 TiB
Google Cloud Standard Tier currently lists:
- First 200 GiB: Free
- 200 GiB to 10,240 GiB: $0.085/GiB
Assume 10 TiB total:
10 TiB = 10,240 GiB
Billable transfer:
10,240 - 200 = 10,040 GiB
Estimated charge:
10,040 × $0.085 = $853.40/month
This is not an apples-to-apples Azure comparison because:
- Azure is using decimal GB/TB
- Google is using GiB/TiB
- Standard Tier uses a different network-routing model
But the underlying point is clear.
Once traffic reaches several terabytes per month, outbound transfer can become a major infrastructure cost.
The Break-Even Formula Is More Useful Than “Free Egress”
Suppose two otherwise acceptable cloud configurations look like this:
Metered plan: Lower monthly infrastructure price + per-GB egress
Included-bandwidth plan: $300/month higher base price + no ordinary per-GB egress charge
The useful calculation is:
Then add any free transfer allowance to determine the total monthly traffic break-even point.
Against Azure Premium Global Network
Extra monthly cost:
$300
Paid egress rate:
$0.12/GB
Paid traffic needed to recover the premium:
$300 ÷ $0.12 = 2,500 GB
Azure also includes 100 GB free.
So total traffic break-even is approximately:
2,500 GB + 100 GB = 2,600 GB
or:
2.6 TB/month
Below roughly 2.6 TB, the metered plan is cheaper under these simplified assumptions.
Above it, the $300 included-bandwidth premium begins to pay for itself.
Against Google Cloud Standard Tier
Extra monthly cost:
$300
Paid rate:
$0.085/GiB
Paid traffic break-even:
$300 ÷ $0.085 = ~3,529 GiB
Add the current 200 GiB free allowance:
3,529 + 200 = ~3,729 GiB
That is approximately:
3.64 TiB/month
The important detail is that a statement such as “the break-even point is 3.5 TiB” can be misleading unless it says whether that refers to billable traffic or total traffic.
Break-Even Points for Different Monthly Premiums
The same formula can be reused for any included-bandwidth plan.
| Extra Monthly Cost of Included Plan | Azure at $0.12/GB | Google Standard at $0.085/GiB |
|---|---|---|
| $100 | ~0.93 TB total | ~1.35 TiB total |
| $200 | ~1.77 TB total | ~2.49 TiB total |
| $300 | ~2.60 TB total | ~3.64 TiB total |
| $500 | ~4.27 TB total | ~5.94 TiB total |
| $1,000 | ~8.43 TB total | ~11.68 TiB total |
These figures include the current free allowance in each example.
They deliberately exclude other differences in compute, storage, NAT, performance, support and tax.
That is why the table should be used as a screening tool, not a purchasing verdict.
Included Bandwidth Changes the Cost of Growth
A metered plan makes network cost variable.
If monthly egress doubles from 2 TB to 4 TB, the egress line rises with it.
An included or unmetered model can behave differently.
AceCloud, for example, currently says bandwidth is unmetered for most normal workloads, with no ordinary per-GB charge for traffic leaving VMs.
Its FAQ adds that sustained CDN-scale traffic in the tens of terabytes per month from a single instance may be discussed during onboarding.
Yotta Shakti uses a different version of the included model.
Its current AI Workspace, AI Lab, GPU, cluster and serverless GPU pricing pages list unlimited data ingress and egress on a number of specific SKUs.
That does not mean every Yotta product has free transfer. The same broader pricing page also lists separate ingress and egress add-ons for other use cases.
The distinction is important:
Included bandwidth can be a product feature, not a provider-wide rule.
Unlimited Transfer Does Not Mean Unlimited Speed
Data volume and network throughput are different measurements.
Data volume answers:
How many TB can I transfer?
Throughput answers:
How quickly can I transfer them?
An unmetered plan can still use a fixed 50 Mbps, 100 Mbps, 500 Mbps or 1 Gbps interface.
The theoretical maximum data moved over a 30-day month is:
| Continuous Port Speed | Theoretical Maximum in 30 Days |
|---|---|
| 50 Mbps | ~16.2 TB |
| 100 Mbps | ~32.4 TB |
| 200 Mbps | ~64.8 TB |
| 500 Mbps | ~162 TB |
| 1 Gbps | ~324 TB |
For example:
50 Mbps × 30 days ≈ 16.2 TB theoretical maximum
That assumes the connection runs continuously at full line rate.
Real applications usually transfer less because of:
- Protocol overhead
- TCP behavior
- Request patterns
- Congestion
- Application bottlenecks
- Shared interfaces
- Server capacity
- Packet loss
- Idle periods
A Fixed Port Can Be a Natural Cost Control
With pay-per-GB egress, a traffic surge can become a billing surge.
With fixed network capacity, the port itself places an upper bound on how quickly traffic can leave.
That can make spending more predictable.
It can also create a performance ceiling.
Consider a product launch where user demand increases fivefold.
Metered Egress
The platform may continue serving additional traffic if compute and network capacity are available.
The bill scales with usage.
Fixed or Included Bandwidth
The monthly data-transfer bill may remain stable.
But the application can eventually hit:
- Port saturation
- Queueing
- Higher latency
- Packet loss
- Throughput limits
Predictability therefore has a trade-off.
A fixed cost is useful only when the included network capacity can support peak demand.
Which Workloads Fit Pay-Per-GB Better?
Pay-per-GB is generally attractive when outbound traffic is:
Examples include:
- Internal SaaS
- Admin applications
- Development workloads
- Small APIs
- Low-volume B2B systems
- Text-heavy AI inference
- Data-processing systems where results remain inside the cloud
In these workloads, paying an extra fixed amount for bandwidth that remains mostly unused may reduce cost efficiency.
Which Workloads Fit Included Bandwidth Better?
Included, bundled or unmetered bandwidth becomes more interesting when outbound traffic is:
Examples include:
- Video delivery
- Large software downloads
- Image-generation platforms
- Video AI
- Backup restores
- Dataset distribution
- CDN origin workloads
- Customer file delivery
- Public artifact repositories
For these workloads, eliminating a per-GB charge can materially change unit economics.
Four Example Workloads
| Workload | Monthly Egress Pattern | Pricing Model Worth Testing First |
|---|---|---|
| Internal dashboard | <100 GB | Metered |
| Small SaaS API | 200 GB–1 TB | Metered |
| Growing media platform | 3–20 TB | Compare break-even carefully |
| Video / dataset distribution | 20 TB+ | Included/unmetered deserves serious testing |
These are directional ranges rather than universal rules.
A workload with 20 TB of traffic might still belong on a hyperscaler if its CDN, global backbone, security, managed services or engineering ecosystem produces a lower total cost.
Predictability Has Economic Value Too
Cloud cost is not only about the lowest expected invoice.
It is also about variance.
Assume two providers have the same expected annual infrastructure cost.
One fluctuates between:
₹3 lakh and ₹8 lakh per month
depending on traffic.
The other stays close to:
₹5 lakh per month
because network transfer is bundled.
Finance, procurement and engineering may value the second model even if its average bill is slightly higher.
Predictability can improve:
- Budgeting
- Gross-margin forecasting
- Customer pricing
- Capacity planning
- Reserved-spend decisions
- Investor forecasting
- Internal cost allocation
This becomes especially important for SaaS companies whose own customer pricing is fixed while infrastructure traffic varies.
But Included Bandwidth Can Be Wasted
Bundled bandwidth is not automatically economical.
If a plan costs more because it includes generous network capacity and the workload uses almost none of it, the organization is effectively paying for idle bandwidth.
That is similar to buying an oversized VM.
The goal is not to maximize what is included.
The goal is to minimize:
Total cost for the capacity actually required.
This is why the break-even point matters.
Do Not Compare Egress Without NAT and Other Network Charges
A metered cloud can have more than one per-GB network meter.
AWS and Azure both document managed NAT Gateway charges that can sit on top of ordinary internet transfer.
Other architecture components can include:
- Cross-zone transfer
- Inter-region transfer
- Load balancers
- Firewalls
- Transit gateways
- Public IPs
- CDN
- Private connectivity
A better formula is:
An included-bandwidth plan should be checked in the same way.
“No egress fee” does not necessarily mean that load balancers, firewalls, public IPs or private links are free.
For the detailed provider comparison, see Cloud Egress Costs in India.
CDN Can Move the Break-Even Point
Suppose an application sends 10 TB/month of static media.
Without a CDN, most of that traffic may come directly from origin infrastructure.
With a high cache-hit ratio, the origin may serve only a fraction of the total.
That can reduce the value of switching clouds purely to avoid origin egress.
The relevant calculation becomes:
Origin Egress + CDN Requests + CDN Delivery
rather than:
Origin Egress Alone
A CDN can also improve latency for geographically distributed users.
That performance consideration matters when comparing a lower-cost network path with a premium backbone.
The Cloud Latency in India 2026 article covers why network cost and network quality should be tested together.
Included Bandwidth Needs a Procurement Checklist
Before treating an included-bandwidth plan as predictable, confirm what “included” actually means.
- Is the transfer allowance measured in GB, GiB, TB or TiB?
- Is data volume truly unmetered?
- What is the port speed?
- Is the port dedicated or shared?
- Are ingress and egress both included?
- Does the rule apply to every service or only the selected SKU?
- Are cross-region transfers included?
- Is object-storage egress included?
- Does CDN traffic have separate pricing?
- Are NAT or gateway processing charges separate?
- Is sustained high-volume traffic subject to fair-use terms?
- Can the provider throttle traffic?
- What happens during traffic spikes?
- Are there overage charges after a hidden or stated threshold?
- Are bandwidth upgrades available?
- Does the contract define sustained-use limits?
- Is the network SLA different from the compute SLA?
- What is the expected cost at 5 TB, 20 TB and 100 TB/month?
For the broader procurement process, use the Cloud Provider Selection Checklist for India.
A Simple Decision Framework
Choose Metered Egress When
Traffic is low, bandwidth demand is uncertain on the low side, and paying for unused network capacity would outweigh the expected transfer bill.
Choose Included Bandwidth When
Traffic consistently exceeds the break-even point and the included throughput is sufficient for peak demand without problematic fair-use restrictions.
Benchmark Both When
Traffic sits near the break-even point, changes rapidly, or network performance and managed services are as important as the transfer price.
A Better Way to Compare the Two Models
Use this sequence.
Step 1: Measure Current Outbound Traffic
Record at least three months of:
- Total internet egress
- Peak Mbps/Gbps
- 95th-percentile throughput if available
- Traffic by workload
- Traffic by destination
Step 2: Forecast Growth
Model:
- Normal month
- High-growth month
- Traffic spike
- 12-month expected volume
Step 3: Calculate Metered Cost
Use:
Free Allowance + Paid Egress Tiers + NAT + Other Network Charges
Step 4: Calculate Included-Bandwidth Cost
Use:
Extra Base Price + Network Add-ons + Required Port Upgrade
Step 5: Find the Break-Even Point
Use:
Extra Fixed Monthly Cost ÷ Metered Egress Rate
Then add any free allowance.
Step 6: Test Performance
Compare:
- p50 latency
- p95 latency
- p99 latency
- Throughput
- Packet loss
- Jitter
- Geographic coverage
Step 7: Check Commercial Terms
Confirm:
- Fair-use policy
- Overage
- Port limit
- SLA
- Contract length
- Upgrade path
- Cancellation terms
Only then is the bandwidth comparison complete.
FAQs
Is included bandwidth always cheaper than pay-per-GB egress?
No. Included bandwidth usually has a fixed cost built into the plan. If outbound traffic is low, a metered provider can be cheaper because little egress is billed. Included bandwidth becomes more attractive when traffic regularly exceeds the plan's break-even point.
How do I calculate the bandwidth break-even point?
Divide the extra monthly cost of the included-bandwidth plan by the metered egress rate. Then add any free egress allowance to calculate the total monthly traffic level where the two simplified costs become equal.
What is the break-even point for a $300 premium against Azure's $0.12/GB rate?
The $300 premium equals 2,500 GB of paid egress at $0.12/GB. After adding Azure's first 100 GB free, the simplified total traffic break-even is about 2,600 GB, or 2.6 TB per month.
What is the break-even point for a $300 premium against Google Cloud Standard Tier?
At $0.085/GiB, $300 buys about 3,529 GiB of paid traffic. Adding the current 200 GiB free allowance gives roughly 3,729 GiB total, or about 3.64 TiB per month.
Does unlimited bandwidth mean unlimited speed?
No. An unmetered or unlimited-transfer service can still have a fixed port speed such as 50 Mbps, 100 Mbps or 1 Gbps. That port limits how quickly data can move even when there is no per-GB charge.
How much data can a 50 Mbps connection transfer in one month?
A 50 Mbps connection running continuously at full line rate for 30 days can theoretically transfer about 16.2 TB. Real usable transfer is lower because applications do not sustain perfect line-rate utilization and network overhead also applies.
Does AceCloud include egress?
AceCloud currently says bandwidth is unmetered for most normal workloads and that standard SaaS, API and web traffic is not separately billed per GB. Its FAQ notes that sustained CDN-scale traffic in the tens of terabytes per month from a single instance may be discussed during onboarding.
Does Yotta Shakti include bandwidth?
Several current Shakti AI Workspace, GPU, AI Lab and cluster SKUs state that data transfer is included or unlimited. However, Yotta also publishes separate transfer add-on pricing for other services, so buyers should verify the exact SKU rather than assume a provider-wide rule.
Which pricing model is better for a video or media platform?
Included or unmetered bandwidth can become attractive because media workloads can produce large and unpredictable outbound volumes. However, CDN pricing, cache hit ratio, port throughput and global delivery performance should be included before deciding.
Which pricing model is better for a small SaaS application?
Pay-per-GB is often simpler for a small SaaS workload with low outbound traffic. If the application sends only a few hundred gigabytes per month, the egress bill may be lower than the fixed premium required for a large included-bandwidth plan.
Final Takeaway
Included bandwidth is not automatically cheaper.
Pay-per-GB egress is not automatically more expensive.
The decision depends on three numbers:
Monthly Traffic + Metered Egress Rate + Extra Cost of the Included Plan
Then add the constraints that pricing tables do not capture:
Port Speed + Fair-Use Terms + Network Performance + Other Network Charges
For low-egress workloads, metered pricing often remains the rational choice.
For workloads consistently moving several terabytes per month, included or unmetered bandwidth deserves serious evaluation.
And for workloads near the break-even point, neither headline is enough.
Calculate the threshold, test the network, and compare the total production architecture.
Use the broader Cloud Egress Costs in India comparison for provider-specific pricing, then use this break-even framework to decide which billing model better fits your traffic.
Sources and Pricing References
Bandwidth rates, free allowances, included-transfer terms and fair-use policies can change. Verify current service-specific pricing and contract terms before making a production purchasing decision.


