Anthropic Just Doubled Claude. Here's What 30 Million People Can Now Build — Including Me.

Thirty million people use Claude every month. Six hundred million visits to claude.ai in March alone. Seventy percent of the Fortune 100 are customers. Three hundred thousand businesses on the platform. On Wednesday, May 6th, 2026, on day one of the Code with Claude conference in San Francisco, every one of those users got the same gift. Doubled session limits on Claude Code. Peak-hour throttling removed. And on the API, a sixteen-fold increase in token throughput on the cheapest paid tier.

This is the operator's read on what just changed. What the Claude Code rate limits doubled SpaceX deal actually unlocks. Who benefits most. And what one solo builder in Santos, Brazil is going to do with the headroom.

Table of Contents

• First, the Scale: Who Actually Uses Claude
• What Exactly Changed on May 6
• What This Unlocks for One Person
• My Numbers: What One Solo Operator Is Going to Do With This
• The SpaceX Colossus 1 Deal: Where the Compute Came From
• For Shopify Operators Specifically
• The Window Is Six to Twelve Months
• Summary
• Frequently Asked Questions
• Sources

First, the Scale: Who Actually Uses Claude

Most coverage of this announcement reads it as a developer-tools update. That misses what is actually going on. Claude is one of the largest consumer AI products in the world. Roughly 30 million monthly active users. Around 287 million website visits in February 2026, climbing to 613 million by March. The Claude mobile app crossed 12 million MAU in February, up nearly 50% in a single month. Seventy percent of the Fortune 100 are paying customers. More than 300,000 businesses use Claude. Anthropic's annualized revenue passed $30 billion in April. The company is valued at $380 billion.

That is the population the May 6 announcement applies to. Not a niche of agent builders. Tens of millions of people who write, code, research, learn, run businesses, automate stores, draft documents, build companies. All of them just got more headroom. Pro at $20 a month does twice the work it did on May 5. Max 5x at $100 and Max 20x at $200 doubled their already-huge windows. Tier 1 API does sixteen times the work on Opus. Same money. More capability.

What Exactly Changed on May 6

Three changes, announced simultaneously by Anthropic Chief Product Officer Ami Vora at the Code with Claude developer conference. Free plan excluded.

1. Claude Code five-hour limits doubled

Across Pro, Max, Team, and seat-based Enterprise plans. The rolling session budget that determined how long an agent could keep working before tapping out is now 2x. Long agentic chains that exhausted at lunch now run to dinner.

2. Peak-hours throttle removed for Pro and Max

In March 2026, Anthropic quietly cut peak-hours capacity for free, Pro, and Max users to manage demand. That penalty is gone for Pro and Max on Claude Code. If you scheduled bulk jobs at 3 AM to dodge the cut, you can stop. The cliff is gone.

3. Opus API limits raised by more than an order of magnitude

This is the change with the largest absolute impact. Tier 1 input tokens per minute on Opus went from 30,000 to 500,000. Output tokens per minute climbed roughly 900%. For anyone running Claude Opus 4.7 against a real workload, this is the change that turns "impossible last week" into "viable today."

One thing absent from the announcement: weekly limits. The doubling is on the five-hour window. The weekly cap was not mentioned. If you were hitting weekly walls, you may still hit them — just later in the week.

"The salient exponent here would be two — the doubling of Claude's five-hour rate limits. Model performance, as measured by benchmarks, has been incremental." — Thomas Claburn, The Register

The right read: this is not a model upgrade. This is an operational upgrade. What you can already do with Claude, you can now do twice as much of, twice as fast, for the same money.

What This Unlocks for One Person

Forget the corporate framing. Think about what one person can now do that they could not do on May 5.

The marketing lead drafting a campaign brief who hit the Pro plan wall halfway through cross-referencing competitor messaging across nine industries. They lost two hours waiting for the reset. Now they don't.

The indie developer on Tier 1 API who wanted to build an agent that scrapes, summarizes, and posts daily, but the 30,000 ITPM ceiling on Opus made it stupid expensive to run on the cheap tier. Sixteen times more throughput at the same price. The agent is now buildable.

The small business owner who wanted to automate replies to customer questions across 200 product pages but couldn't justify a Max plan. Pro just doubled. The math works.

The consultant writing client deliverables who runs four projects in parallel, hitting peak-hour throttling at 2 PM right when client revisions land. The 2 PM cliff is gone. They can ship through the day.

The agency owner running content for five clients on a single Pro plan, sharding bulk work across odd hours to stay under the limits. The bulk work now fits inside one session, in business hours.

The Max 20x power user running multiple Claude instances in parallel against codebases, knowledge bases, and live agentic pipelines. The tier was already 20x Pro's allowance. It just doubled again. What was previously a "use carefully across the day" budget is now closer to "run it open."

Every one of these scenarios was a real wall last week. None of them are now. That is the actual story.

My Numbers: What One Solo Operator Is Going to Do With This

I run four Shopify stores from Santos, Brazil. Two production AI systems built from scratch — Nexus, the platform you are currently reading about, and The Magus, an autonomous publishing pipeline running on a separate site. Around 25,000 products optimized for SEO across multiple stores. Custom Liquid theme work. Daily autonomous blog generation pipelines. One person.

I run on Max 20x — Anthropic's heaviest consumer plan at $200 a month — and on Tier 1 API for the autonomous pipelines. That combination already does the work of a small team. After May 6, it does the work of a bigger one.

I am not a special case. I am a representative case. The reason a solo operator can run this much surface area is that Claude is doing a meaningful share of the work. Drafting product descriptions. Generating SEO copy. Writing code for Shopify themes. Producing first drafts of long-form content. Running 13-phase autonomous research-and-writing pipelines that publish blog content daily on schedule.

Before May 6, my pipelines were sharded. Big jobs split across sessions. Bulk catalog rewrites scheduled at off-peak hours to dodge the throttle. Token budgets watched like a hawk on the API side because Tier 1 was the cheap tier and the cheap tier had a low ceiling.

After May 6: a 4,000-product catalog rewrite fits in one session. The autonomous blog pipelines stop rationing tokens to stay under the ITPM ceiling. The research-gated drafts can go longer and deeper without aborting halfway. I can take Nexus from a daily blog cadence to twice-daily for clients who want it. The bulk image-alt-text rewrite I have been putting off across three stores finally has math that works at Tier 1.

None of this required a model upgrade. It required headroom. I just got it. So did you.

The SpaceX Colossus 1 Deal: Where the Compute Came From

Rate limits do not move because of generosity. They move because of compute. Anthropic signed an agreement to take all of the compute capacity at Colossus 1, the supercomputer originally built by xAI in 122 days inside a former Electrolux factory in Memphis, Tennessee. SpaceX acquired xAI in February 2026 in an all-stock deal valued at $1.25 trillion combined. In May 2026 Musk folded xAI into a new "SpaceXAI" sub-brand. So Colossus 1 is now SpaceX-owned but was built and originally operated by xAI to train Grok.

The numbers: 300 MW, 220,000 GPUs, half of xAI's fleet

The deal gives Anthropic more than 300 megawatts of new capacity and over 220,000 NVIDIA GPUs — including H100, H200, and GB200 accelerators — within the month. For context, 300 MW is enough to power roughly 240,000 average US homes. Per Data Center Dynamics, this hands Anthropic just under half of xAI's roughly 500,000-GPU total fleet. Musk's stated rationale: SpaceXAI had already moved Grok training to Colossus 2, so leasing Colossus 1 to a competitor became viable.

The strangeness of the deal is part of the story. Musk has called Anthropic "misanthropic and evil" in the past. After meeting the team in early May, he posted on X that he was impressed and that "no one set off my evil detector." The compute hunt is bigger than the rivalry.

The five-partnership compute map

SpaceX is not the whole story. Anthropic now holds five major compute agreements:

• Amazon — up to 5 gigawatts, with nearly 1 GW of new capacity online by end of 2026.
• Google and Broadcom — 5 GW agreement, beginning to come online in 2027.
• Microsoft and NVIDIA — strategic partnership including $30 billion of Azure capacity.
• Fluidstack — $50 billion American AI infrastructure investment, disclosed alongside the SpaceX deal.
• SpaceX — Colossus 1, 300+ MW, 220,000+ NVIDIA GPUs, Memphis, Tennessee.

Tens of gigawatts of contracted future capacity. The SpaceX deal landed first because Colossus 1 was already built and online. The rest comes through 2026 and 2027.

The orbital footnote

Anthropic also said it has expressed interest in partnering with SpaceX on multiple gigawatts of orbital AI compute capacity. No agreement has been signed there. SpaceX filed paperwork with the FCC in January 2026 for a million-satellite orbital data center constellation. File this under "future, not present." But the fact that it is in the announcement at all tells you where the compute hunt is going.

For Shopify Operators Specifically

If you run a Shopify store, the numbers above translate directly. The doubled five-hour limit and the 16x Opus API token-per-minute increase mean that bulk catalog jobs which previously had to be sharded across multiple sessions now complete in a single session.

Concrete example. 4,000 SKUs. You want to rewrite product descriptions, update SEO titles, refresh metafields. Roughly 3,000 input tokens per product, 800 output. At the old Tier 1 ceiling of 30,000 ITPM, you processed about 10 products per minute on the API side. A 4,000-SKU job took 6.6 hours of sustained throughput, busted past the five-hour session limit, and got hit by the peak-hours penalty. You sharded. You scheduled. You babysat.

At 500,000 ITPM, the same job is now bottlenecked by Shopify's GraphQL cost throttle, not by Claude. The doubled session limit comfortably contains a full pass. Peak-hours throttling is gone. The job runs once, to completion, during business hours. This is the workflow Nexus SEO automates end-to-end — feeding the catalog, drafting copy with brand voice constraints, writing back via GraphQL with retry logic baked in.

If you are setting this up yourself rather than buying it, the playbooks in Shopify Mastery cover the GraphQL plumbing and theme integration patterns Claude Code agents need to run safely against a live store.

One ceiling did not change: Shopify's own GraphQL leaky-bucket cost system. If your agent fires too many productUpdate mutations in parallel, Shopify will throttle you regardless of how generous Claude is being. Build with exponential backoff, idempotency keys, and checkpointing. Use prompt caching — Anthropic's ITPM limits are cache-aware, so cached input tokens do not count toward the per-minute ceiling. With an 80% cache hit rate, your effective throughput is roughly 5x the published number.

The Window Is Six to Twelve Months

Capital raised today takes 12 to 24 months to translate into deployed compute. Hardware orders. Data center construction. Power provisioning. Demand is growing faster than capital can become silicon.

Anthropic just bought itself maybe six to twelve months of headroom. Then demand catches up. The window opened by the Claude Code rate limits doubled SpaceX deal is real but finite. The Amazon and Google/Broadcom capacity comes online through late 2026 and 2027. Between those drops, expect quiet adjustments. Peak-hour penalties returning. Weekly cap reductions. Tier reshuffles. The March 2026 peak-hours cut was that pattern. The May reversal is the relief. The next squeeze is somewhere on the calendar.

Three rules for builders:

1. Build for the squeeze, not the drop. Assume the next throttle is six months out. Design retry logic, token budgeting, and graceful degradation now.
2. Keep the LLM swappable. Your agent should treat Claude as a provider, not a dependency. A thin abstraction layer protects you from any single platform's pricing or limit decisions. The broader industry is shifting in the same direction — on April 27, 2026, GitHub announced Copilot is moving from premium request units to GitHub AI Credits on June 1, 2026, with billing calculated by token consumption. Pricing models are converging on tokens consumed.
3. Watch the changelog, not the marketing. The March 2026 peak-hours cut was buried. The May 2026 reversal made headlines. Real operators read both.

While you're auditing your stack, also check whether your Shopify products are visible to AI shopping agents. The same agents getting more compute are also doing the buying.

Frequently Asked Questions

How many people use Claude?

Roughly 30 million monthly active users across web and mobile as of early 2026. claude.ai received around 287 million visits in February 2026 and 613 million in March. Anthropic also reports more than 300,000 business customers, with 70% of the Fortune 100 paying for Claude. Annualized revenue passed $30 billion in April 2026.

What does Claude actually cost in 2026?

Pro is $20/month ($17 if billed annually). Max 5x is $100/month for five times Pro's usage. Max 20x is $200/month for twenty times Pro's usage. Team Premium (with Claude Code included) is $100/seat/month on annual billing. API is pay-as-you-go billed per token, separate from subscription plans.

Did Anthropic double Claude Code rate limits for all plans including free?

No. The doubling applies specifically to Pro, Max, Team, and seat-based Enterprise plans. The peak-hours throttle was removed for Pro and Max. Free plan changes were not part of this announcement.

What is the SpaceX Colossus 1 deal and how does it affect my Claude access?

Anthropic signed an agreement to use all the compute capacity at Colossus 1, originally built by xAI and now owned by SpaceX following the February 2026 acquisition. The data center, located in Memphis, Tennessee, contributes more than 300 megawatts and over 220,000 NVIDIA GPUs within the month. That capacity directly funds the rate limit increases on Claude Code and the Opus API.

What are the new Claude Opus API rate limits after the May 2026 update?

Anthropic raised Opus API rate limits by more than an order of magnitude. Tier 1 users went from 30,000 maximum input tokens per minute to 500,000. Output tokens per minute climbed roughly 900%. Higher tiers received similar proportional increases.

I'm a solopreneur, not a developer. Does this affect me?

Yes. If you use Claude Pro at $20 a month for any meaningful work — drafting, research, learning, coding, customer support — your sessions just got twice as long. The peak-hours penalty that capped your midday throughput is gone. Same money, more capability. If you've been hitting Pro's walls daily, Max 5x at $100/month or Max 20x at $200/month also doubled their (already much larger) windows.

How do the new limits affect Shopify development workflows?

Shopify developers running Claude Code workflows against the GraphQL Admin API can now complete much larger jobs in a single session. Bulk product description rewrites, metafield updates, and inventory operations that previously had to be sharded across sessions now fit inside the doubled budget. Tier 1 API users running Opus see a 16x throughput increase per minute.

Will Claude rate limits go up again in 2026 or 2027?

Likely yes, in waves. Anthropic's compute agreements with Amazon, Google/Broadcom, Microsoft/NVIDIA, and Fluidstack bring additional capacity online through late 2026 and 2027. However, the 12-to-24-month lag between capital raised and deployed infrastructure means quiet squeezes between capacity drops are expected.

Are the weekly Claude Code limits also doubled?

Anthropic's announcement specified the five-hour limits and the Opus API limits. Weekly limits were not addressed, which leaves them as the open question for heavy users planning multi-day agent workloads.

Sources

• Anthropic — Higher usage limits for Claude and a compute deal with SpaceX: anthropic.com/news/higher-limits-spacex
• Anthropic — Claude Plans & Pricing: claude.com/pricing
• The Register (Thomas Claburn) — Claude hitches a ride on SpaceX's datacenter capacity: theregister.com
• IT Pro — Anthropic is increasing Claude Code usage limits: itpro.com
• Engadget — Anthropic is doubling Claude Code rate limits after deal with SpaceX: engadget.com
• CNBC — Anthropic, SpaceX announce compute deal that includes space development: cnbc.com
• Data Center Dynamics — Anthropic to use all of SpaceX-xAI's Colossus 1 data center compute: datacenterdynamics.com
• The Next Web — Anthropic raises Claude Code and Opus API rate limits, citing SpaceX Colossus 1 deal: thenextweb.com
• Business Standard — Anthropic hikes Claude usage limits for paid users post SpaceX compute deal: business-standard.com
• PCWorld — Anthropic doubles Claude Code limits, thanks to a deal with SpaceX: pcworld.com
• Wikipedia — Colossus (supercomputer): en.wikipedia.org/wiki/Colossus_(supercomputer)
• Wikipedia — xAI (company): en.wikipedia.org/wiki/XAI_(company)
• GitHub Blog — GitHub Copilot is moving to usage-based billing: github.blog
• Backlinko / Demandsage — Claude Statistics 2026: demandsage.com
• Get Panto — Anthropic AI Statistics 2026: getpanto.ai
• Second Talent — Claude AI Statistics and User Trends for 2026: secondtalent.com

Related Posts

• Every AI Agent on the Market Is a Chatbot With Better Marketing — Except This One
• AI Commerce Is Live on Shopify — 33% of Stores Haven't Even Started
• How to Optimize Thousands of Shopify Product Listings Without Losing Your Mind