Will DBT read from materialised table or rerun SQL?

My new company use DBT and I'm struggling to find the answer to this question.

In my example, each model is set to materialise as a table.
If I have models that reference a previously run model and produce a DAG like this:
Query1 > Query2 > Query3 > Query 4.

Will this cause DBT to:
read from the materialised table of Query1 when it hits {{ref ('Query1')}} in Query2?
OR.
run all the SQL from Query1 again, and keep it in memory, when it hits {{ref ('Query1')}} in Query2?

Solution

The former. dbt will replace {{ ref('Query1') }} with the fully-qualified name of the database relation that is materialized from the model whose name is Query1.

You can see the SQL that dbt actually executes in your target directory (inside your project directory), under target/compiled/path/to/my/model and/or target/run/path/to/my/model. If you do that for Query2, you'll see {{ ref('Query1') }} has been replaced with something like "my_database"."my_schema"."Query1".

This is true only because you said that all of these models are materialized as tables. There are other materializations supported by dbt:

Views: In this case, dbt will create a view for each model. Each downstream model will query the view, so the compiled SQL will similarly reference "my_database"."my_schema"."Query1" when you use {{ ref('Query1') }}. But under the hood, in order to execute that query, your database will be re-executing the SQL defined in Query1, since that is how views work
Ephemeral: An "ephemeral" materialization isn't really a materialization! If you materialize Query1 as ephemeral, and then {{ ref('Query1) }} in Query2, dbt will inject a CTE with the code from Query1 into the compiled Query2 query. That will look like:

-- Query1 is materialized as ephemeral
with "__dbt_cte_Query1" as (
    -- the code from the Query1 model file
)
-- your code from the Query2 model
select
    ...
from
    -- in your Query2 file, you used {{ ref('Query1') }} here
    "__dbt_cte_Query1"
...