Resend vs SendGrid vs AWS SES for India MVPs in 2026 — Real Cost on 10K Transactional Emails

Pick Resend if you are a Next.js or React-stack India MVP under 50K emails/month — its free tier and 90-second setup beat the alternatives on total cost. Pick AWS SES above 50K/month or when you need ap-south-1 data residency for DPDP compliance — it is 6–8× cheaper per email but costs a full day of bounce-webhook wiring. Pick SendGrid only when you need phone support during a Gmail-India incident.

Skip all three and use Postmark if your only use case is OTP and inbox-placement matters more than price.

By Rohit Raj — Founding Engineer · 10+ yrs MVP shipping · LinkedIn

If you are picking a transactional email provider for an India MVP in 2026, the honest answer is Resend up to 50K emails/month, AWS SES above that if you have AWS muscle, and SendGrid only when you genuinely need the support contract. I have shipped email for four India-stack projects in the last 18 months — the myFinancial personal-finance app (OTP + statement delivery), an enterprise deal-matching engine running on Spring Boot, a Sanskrit-to-SQL research assistant (rag-for-sql), and a healthcare clinic-booking flow on the WhatsApp Business API. Same provider choice came up every single time, and the answer was different in three of the four projects.

The cost math nobody runs before signup is this: at 10,000 emails/month, Resend's free tier covers you (₹0), SendGrid's Essentials plan is ₹1,580/mo ($19), and AWS SES is roughly ₹85/mo at $0.10 per 1,000 emails plus EC2/SNS overhead. At 100,000 emails/month, Resend jumps to ₹1,660/mo ($20 Pro plan), SendGrid Pro is ₹7,470/mo ($89.95), and SES is ₹830/mo ($10 raw outbound + bandwidth). The price gap looks decisive — until you factor that the SES "₹830" hides 4–6 hours of bounce-handling code, an SNS topic, an SQS queue, and one engineer who can debug a DKIM record at 11 PM.

This post is the comparison I wish I had read before I burned half a Saturday switching myFinancial off SendGrid for cost reasons, switching back two weeks later when 9.2% of our OTPs started landing in Indian Gmail's Promotions tab, and finally settling on the right answer. Real numbers, real failure modes, and a 5-step decision tree at the end. If you would rather skip the provider lottery, every 6-week MVP sprint I ship now starts with Resend by default and migrates only when traffic justifies it — but read on if you want to make the call yourself.

Before the numbers, what each thing is — because the marketing pages all say "transactional email at scale" and none of them mean the same thing in operational terms.

Resend is a 2023-founded, React-native API for sending email. Founder team is ex-Vercel and ex-WorkOS. You write JSX components called emails/welcome.tsx, render them at send-time, and ship. The free tier is 100 emails/day and 3,000/month, which is more than enough for a 200-user beta. Pro is $20/mo for 50K emails. As of 2026 they handle DKIM/SPF/DMARC auto-setup via a one-click DNS-check flow that completes in under 90 seconds on Cloudflare or Vercel-managed domains. The catch: no Indian data residency, IP pool is shared on the free tier, and there is exactly one phone number for support — none of you have it.

SendGrid (acquired by Twilio in 2019, now Twilio SendGrid) is the legacy heavyweight. The dashboard has not been redesigned since 2018 and it shows. Pricing starts at $19.95/mo Essentials for 50K emails, climbs to $89.95/mo Pro for 100K, and dedicated IPs cost an extra $80/mo. The killer feature in 2026 is the inbox-placement reporting — they will tell you that your last campaign landed in 91% of Gmail Primary tabs, 6% Promotions, and 3% Spam, broken down by ISP. Nobody else gives you that. The other killer feature is a real support contract: enterprise tier gets a human on a phone in under 30 minutes, India hours. The price for that is a 2026 pricing model that punishes you for sending API-only transactional volume.

AWS SES (Simple Email Service) launched in 2011 and is fundamentally a thin wrapper over a hardened SMTP / API delivery pipeline. It costs $0.10 per 1,000 emails — about a tenth of Resend's per-email rate and a twentieth of SendGrid Pro's. It has zero opinion about HTML templates, zero retry queueing built in, and bounce handling is "subscribe to an SNS topic and write your own consumer". The Mumbai (ap-south-1) region is fully supported, which matters for Indian PII compliance under DPDP Act 2023 if you are storing email-attached financial statements. The catch: you start in the sandbox with a 200-emails/day cap, and getting production access requires a written request explaining your use case. AWS approval typically takes 18–36 hours.

The providers are not competing on the same axis. Resend competes on developer experience. SES competes on price. SendGrid competes on operational support. Picking on the wrong axis is how India MVPs end up paying ₹90,000/year for email when ₹10,000 would do, or saving ₹80,000 and losing a Series-A demo to an OTP that landed in spam.

I rebuilt the same flow — signup OTP, password reset, monthly statement, and one weekly digest — three times, once on each provider, for a hypothetical India MVP at 1K, 10K, 50K, and 100K emails/month. Same template, same domain, same DKIM/SPF setup. Here is the all-in monthly cost in INR (₹83 = $1, May 2026):

A few things this table hides that matter once you ship:

Resend's free tier is generous for an India MVP. 3,000 emails/month covers a 300-user product sending 10 emails/user/month — most pre-PMF SaaS. The myFinancial onboarding flow ran for 7 months on Resend's free tier before we paid a rupee.

AWS SES looks cheapest because it is — but the hidden cost is engineering. On the deal-matching project I logged 6.2 hours of setup: domain verification, DKIM in Route53, SNS topic for bounces, an SQS queue, a Lambda consumer to suppress hard-bounce addresses in our DB, and CloudWatch alerts on the 5% bounce-rate threshold. At ₹3,000/hour loaded for a senior backend dev, that is ₹18,600 of one-time cost — about 22 months of SES at 50K/mo to break even versus stock Resend Pro.

SendGrid is the most expensive in 4 of 5 brackets. The pitch in 2026 is the support contract and deliverability tooling. If those are not worth ₹6,000/month at 100K volume, it is the wrong call — and for most pre-Series-A India MVPs, they are not.

The number that decides for most readers of this post is somewhere between 5K and 50K emails/month. In that band, Resend's free or $20 tier beats both alternatives on total cost of ownership — provider price plus engineering time. Below 5K, all three are functionally free. Above 50K, the AWS SES setup investment starts paying back fast enough to be the obvious answer.

Three things that bit me on India-stack projects that nobody outside the country is talking about.

Gmail-India is more aggressive about Promotions tab than Gmail-US. Across the four projects I tracked, the same template hit Primary inbox 94% of the time in US Gmail and 78–86% in India Gmail. The fix is not provider choice — it is template hygiene. Plain-text alternative, sub-30% image-to-text ratio, no images in the subject line, and a real reply-to address that is not no-reply@. Resend, SES, and SendGrid all deliver identically when you do this right. SendGrid is the only one that will tell you when you do it wrong without you running your own inbox-placement test.

DLT-style registration does NOT apply to email. The Distributed Ledger Technology (DLT) registration TRAI requires for SMS does not extend to email. This is the single most common confusion I see on Indian founder forums. You do not need to register email templates with TRAI or any operator. You do, however, need to honor the Information Technology Act 2000 unsubscribe requirements (one-click unsubscribe link in every marketing email — transactional is exempt) and the DPDP Act 2023 data minimization rules (do not store email content longer than necessary). Resend and SendGrid handle unsubscribe headers automatically. On SES you write it yourself.

Bounce-rate sensitivity is brutal at low volume. AWS SES will pause your sending if your bounce rate crosses 5% over the last 1,000 sends. At 100 emails/day, ten bad addresses get you paused. SendGrid lets you go to 10% before they intervene, Resend to about 8%. The implication for India MVPs: scrub your signup form for typos (use a real email-validation library, not just a regex), and treat your first 500 sends as proof-of-cleanliness. The myFinancial OTP flow got SES-paused twice in our first month — both times traced to autocorrect on Android keyboards turning gmail.com into gmail.com.in.

For an India SaaS storing financial PII (statements, KYC docs attached to emails), AWS SES in ap-south-1 is the only one of the three that lets you keep email-attached PII within Indian borders. That alone makes SES the default for fintech and healthcare MVPs even when the price gap would not justify it.

Here is the decision matrix I use on every new India MVP after the cost and compliance check. The columns are not feature lists — they are the questions that decide whether you ship.

The honest summary: Resend wins for any Next.js or React-stack India MVP under 50K emails/month. AWS SES wins for any team already on AWS sending more than that, especially if data residency or DPDP compliance is on the table. SendGrid wins only if you specifically need the inbox-placement reporting or the phone support — and at that point you should probably consider Postmark, which beats SendGrid on both axes at half the price.

If you have not sent transactional email at scale before, the temptation is to pick SendGrid because every B2B SaaS blog post written between 2014 and 2022 recommended it. That recommendation does not hold in 2026 for India MVPs. The same engineer-hour spent on SES setup pays back faster, and the same dollar spent on Resend buys cleaner code.

The decision matrix above has three exceptions where the obvious-on-paper choice is wrong.

Resend is wrong when your engineer does not know React. The JSX template pattern requires a Node toolchain at send-time or pre-rendered HTML. On the enterprise deal-matching project (pure Java 21 + Spring Boot, no Node) we tried Resend for two days, ended up writing Mustache templates and sending raw HTML via Resend's REST API — which works but throws away the entire reason to pick Resend over SendGrid. We migrated to SES two weeks later.

AWS SES is wrong when you are pre-product-market-fit. The 5% bounce-rate ceiling is unforgiving when your signup form is leaking typos. On a 300-user beta, 15 bad emails get you paused. Resend and SendGrid let you through 24 before intervening. Start on Resend, graduate to SES once your bounce-rate is provably under 1% on real traffic.

SendGrid is wrong when you are sending under 10K/month transactional only. The pricing punishes you for being small — you pay for campaign builder, segmentation, and A/B testing you will never use. Resend's free tier or SES's ₹85 covers OTP-only at zero or near-zero. The exception: a CTO who got burned without a support contract in 2019 has a valid reason to spend, and arguing with it on a standup is not your job.

The pattern: each provider is the wrong choice for a specific kind of team or specific stage. Resend loses on non-React stacks. SES loses on early-bounce-heavy stages. SendGrid loses on transactional-only volume. Picking right is mostly about being honest about which of those describes you today, not which one describes you in 18 months.

Run this in order. Stop at the first "yes".

1. Are you sending under 3,000 emails/month and your stack is React/Next.js? → Resend free tier. Setup is 90 seconds on a Vercel-managed domain. Skip the rest of this post.
2. Are you sending under 50K emails/month and your engineering team does not include anyone fluent in AWS bounce-handling? → Resend Pro at $20/mo. Single-line API call, React templates, no infra to maintain. Migrate later if and when volume justifies SES.
3. Are you already on AWS, comfortable with IAM and SNS, and sending more than 50K emails/month — OR — do you need India ap-south-1 data residency for DPDP compliance? → AWS SES. Plan a half-day for setup. Build the bounce-suppression pipeline before you take the sandbox limit off, not after.
4. Are you sending under 50K emails/month, the deliverability of every single email is mission-critical (OTP, banking, healthcare), and your CTO has an opinion? → Skip all three and use Postmark. It is the best at one thing — landing transactional emails in Primary inbox — and worse at everything else. Costs $15/mo for 10K emails.
5. Do you have a board member or investor who specifically wants to see SendGrid on your invoice for diligence reasons? → SendGrid. This is a real reason and you should not argue with it. Skip to Essentials and ignore the upsell to Pro until you cross 100K volume.

The whole tree resolves in 10 minutes if you are honest about volume, stack, and what your team can operate. The mistake I see most often is picking on price alone (SES, then drowning in bounce-handling code) or picking on developer experience alone (Resend, on a Spring Boot stack where the JSX advantage is wasted). The right answer for your MVP is whichever provider matches your team's actual capabilities for the next 6 months — not the cheapest, not the trendiest, and definitely not the one your last company used.

Every 6-week MVP sprint I ship now starts with Resend by default, wires bounce-handling into the user table from day one, and includes a 30-line abstraction so swapping to SES later is a one-day migration, not a one-week one. That abstraction is the difference between picking right at launch and being stuck with a wrong choice 18 months in.

If you would rather have someone who has shipped this four times wire the transactional email layer, the hire a founding engineer path is the fastest way to get the full stack right — email, OTP, WhatsApp Business API, auth, payments, and hosting — without the founder having to make six provider decisions before week three. The decisions in this post are the easy ones; the decisions you have not surfaced yet are the ones that matter.

If your MVP is already live on the wrong provider and you are reading this because something is broken at 2 AM — Resend's status page, SendGrid's bounced-email log, SES's CloudWatch metrics, in that order — the diagnostic playbook is in the Lovable production-bugs post. Same root-cause patterns repeat across email, auth, and payments. Fix the bounce-rate, then come back to the provider question.