On October 1, WhatsApp starts charging the answer.

Dear Mark,

WhatsApp became the global standard by making messaging free, accessible, and reliable anywhere under the same conditions. That foundation replaced the friction of paid SMS and turned WhatsApp into everyday infrastructure, reaching more than nine in ten internet users in countries such as Brazil.¹

In 2024, you brought that same principle to business messaging by making customer service replies free, while retiring the self-hosted gateways and moving every business onto Meta's Cloud API. Today, many businesses rely on WhatsApp to deliver support, assistance, and intelligence.

From October 1, 2026, you will charge those businesses for each service reply through the Business Platform, whether a person or an independent AI writes the answer. The 1,000 free messages per phone number each month are spent by the second business day for an operation answering 20,000 customers a month.²

The published rates run from 0.14¢ to 5.5¢ per billable service message, set market by market. A business does not choose its rate: it inherits the geography of its customers. At those endpoints, one million billable replies means $1,400 to $55,000 in delivery fees alone, before software, before inference, before the people who make the service work.³ And you can revise those rates every quarter.

Your own Business Agent pays no delivery fee: you bill it by the token, about four to five cents a reply, and nothing for delivery.⁴ Every independent assistant pays for its intelligence, then pays you to deliver the answer. In the most expensive markets, that delivery fee alone matches what your agent bills for the whole answer. A reply is charged either as a Business Agent message or as a service message, never both: the delivery fee applies only to replies you did not write.⁴

For any customer service, replies are the work itself. While AI inference is becoming cheaper every month, these delivery fees set an artificial floor. In those markets, delivering a reply costs 5.5¢ while generating it on an efficient model costs a fraction of a cent: the fee can be ten times the intelligence it carries. No amount of technical efficiency can optimize away a delivery fee set by the platform.

This leaves businesses with a bad choice:

They can pay delivery charges on every reply, absorb them, or pass the cost on to their users.

Or they can switch to your Business Agent to escape a delivery fee that applies only to independent replies, while giving up sovereignty over their AI, their data, and where their models run.⁵

No business should have to trade control over its processes and data just to get fair delivery terms.

You make the case for that freedom yourself. In The Future is for Everyone, you argue for competing agents and open models, and say agents should share “a person's goals and values, not our company's.”⁶ Those principles must hold when an independent service wants to reach a customer on your platform. Give them room to serve their customers on fair terms.

Businesses should choose who supplies their intelligence and where it runs.

The ask is simple:

The signatories of this letter are ready to have that conversation.

Keep WhatsApp Open For Business.

Sincerely,
the undersigned