diff --git a/spacy/kb/kb.pyx b/spacy/kb/kb.pyx
index 3ee434ab5..fa537edc9 100644
--- a/spacy/kb/kb.pyx
+++ b/spacy/kb/kb.pyx
@@ -18,6 +18,8 @@ cdef class KnowledgeBase:
     DOCS: https://spacy.io/api/kb
     """
 
+    _KBType = TypeVar("_KBType", bound=KnowledgeBase)
+
     def __init__(self, vocab: Vocab, entity_vector_length: int):
         """Create a KnowledgeBase."""
         # Make sure abstract KB is not instantiated.
@@ -107,16 +109,15 @@ cdef class KnowledgeBase:
             Errors.E1044.format(parent="KnowledgeBase", method="from_disk", name=self.__name__)
         )
 
-    KBType = TypeVar("KBType", bound=KnowledgeBase)
     @classmethod
     def generate_from_disk(
-        cls: Type[KBType], path: Union[str, Path], exclude: Iterable[str] = SimpleFrozenList()
-    ) -> KBType:
+        cls: Type[_KBType], path: Union[str, Path], exclude: Iterable[str] = SimpleFrozenList()
+    ) -> _KBType:
         """
-        Factory method for generating KnowledgeBase instance from file.
+        Factory method for generating KnowledgeBase subclass instance from file.
         path (Union[str, Path]): Target file path.
         exclude (Iterable[str]): List of components to exclude.
-        return (KBType): Instance of KnowledgeBase generated from file.
+        return (_KBType): Instance of KnowledgeBase subclass generated from file.
         """
         raise NotImplementedError(
             Errors.E1044.format(parent="KnowledgeBase", method="generate_from_disk", name=cls.__name__)
diff --git a/spacy/ml/models/entity_linker.py b/spacy/ml/models/entity_linker.py
index 2f8ab20a5..fd84981b6 100644
--- a/spacy/ml/models/entity_linker.py
+++ b/spacy/ml/models/entity_linker.py
@@ -99,6 +99,14 @@ def empty_kb(
     return empty_kb_factory
 
 
+@registry.misc("spacy.EmptyKB.v2")
+def empty_kb_for_config() -> Callable[[Vocab, int], KnowledgeBase]:
+    def empty_kb_factory(vocab: Vocab, entity_vector_length: int):
+        return InMemoryLookupKB(vocab=vocab, entity_vector_length=entity_vector_length)
+
+    return empty_kb_factory
+
+
 @registry.misc("spacy.CandidateGenerator.v1")
 def create_candidates() -> Callable[[KnowledgeBase, Span], Iterable[Candidate]]:
     return get_candidates
diff --git a/spacy/pipeline/entity_linker.py b/spacy/pipeline/entity_linker.py
index b2d8e7a13..55a04e7ca 100644
--- a/spacy/pipeline/entity_linker.py
+++ b/spacy/pipeline/entity_linker.py
@@ -18,7 +18,6 @@ from thinc.api import CosineDistance, Model, Optimizer, Config
 from thinc.api import set_dropout_rate
 
 from ..kb import KnowledgeBase, Candidate
-from ..ml import empty_kb
 from ..tokens import Doc, Span
 from .pipe import deserialize_config
 from .legacy.entity_linker import EntityLinker_v1
@@ -64,6 +63,7 @@ DEFAULT_NEL_MODEL = Config().from_str(default_model_config)["model"]
         "entity_vector_length": 64,
         "get_candidates": {"@misc": "spacy.CandidateGenerator.v1"},
         "get_candidates_all": {"@misc": "spacy.CandidateAllGenerator.v1"},
+        "generate_empty_kb": {"@misc": "spacy.EmptyKB.v2"},
         "overwrite": True,
         "scorer": {"@scorers": "spacy.entity_linker_scorer.v1"},
         "use_gold_ents": True,
@@ -91,6 +91,7 @@ def make_entity_linker(
         [KnowledgeBase, Generator[Iterable[Span], None, None]],
         Iterator[Iterable[Iterable[Candidate]]],
     ],
+    generate_empty_kb: Callable[[Vocab, int], KnowledgeBase],
     overwrite: bool,
     scorer: Optional[Callable],
     use_gold_ents: bool,
@@ -115,6 +116,7 @@ def make_entity_linker(
             Iterator[Iterable[Iterable[Candidate]]]
         ]): Function that produces a list of candidates per document, given a certain knowledge base and several textual
         documents with textual mentions.
+    generate_empty_kb (Callable[[Vocab, int], KnowledgeBase]): Callable returning empty KnowledgeBase.
     scorer (Optional[Callable]): The scoring method.
     use_gold_ents (bool): Whether to copy entities from gold docs or not. If false, another
         component must provide entity annotations.
@@ -151,6 +153,7 @@ def make_entity_linker(
         entity_vector_length=entity_vector_length,
         get_candidates=get_candidates,
         get_candidates_all=get_candidates_all,
+        generate_empty_kb=generate_empty_kb,
         overwrite=overwrite,
         scorer=scorer,
         use_gold_ents=use_gold_ents,
@@ -192,6 +195,7 @@ class EntityLinker(TrainablePipe):
             [KnowledgeBase, Generator[Iterable[Span], None, None]],
             Iterator[Iterable[Iterable[Candidate]]],
         ],
+        generate_empty_kb: Callable[[Vocab, int], KnowledgeBase],
         overwrite: bool = BACKWARD_OVERWRITE,
         scorer: Optional[Callable] = entity_linker_score,
         use_gold_ents: bool,
@@ -215,14 +219,15 @@ class EntityLinker(TrainablePipe):
             Callable[
                 [KnowledgeBase, Generator[Iterable[Span], None, None]],
                 Iterator[Iterable[Iterable[Candidate]]]
-            ]): Function that produces a list of candidates per document, given a certain knowledge base and several textual
-            documents with textual mentions.
+            ]): Function that produces a list of candidates per document, given a certain knowledge base and several
+            textual documents with textual mentions.
+        generate_empty_kb (Callable[[Vocab, int], KnowledgeBase]): Callable returning empty KnowledgeBase.
         scorer (Optional[Callable]): The scoring method. Defaults to Scorer.score_links.
         use_gold_ents (bool): Whether to copy entities from gold docs or not. If false, another
             component must provide entity annotations.
-        candidates_doc_mode (bool): Whether or not to operate candidate generation in doc mode, i.e. to provide a generator
-            yielding entities per document (candidate generator callable is called only once in this case). If False,
-            the candidate generator is called once per entity.
+        candidates_doc_mode (bool): Whether or not to operate candidate generation in doc mode, i.e. to provide a
+            generator yielding entities per document (candidate generator callable is called only once in this case). If
+            False, the candidate generator is called once per entity.
         threshold (Optional[float]): Confidence threshold for entity predictions. If confidence is below the
             threshold, prediction is discarded. If None, predictions are not filtered by any threshold.
         DOCS: https://spacy.io/api/entitylinker#init
@@ -241,16 +246,16 @@ class EntityLinker(TrainablePipe):
         self.model = model
         self.name = name
         self.labels_discard = list(labels_discard)
+        # how many neighbour sentences to take into account
         self.n_sents = n_sents
         self.incl_prior = incl_prior
         self.incl_context = incl_context
         self.get_candidates = get_candidates
         self.get_candidates_all = get_candidates_all
+        self.generate_empty_kb = generate_empty_kb
         self.cfg: Dict[str, Any] = {"overwrite": overwrite}
         self.distance = CosineDistance(normalize=False)
-        # how many neighbour sentences to take into account
-        # create an empty KB by default
-        self.kb = empty_kb(entity_vector_length)(self.vocab)
+        self.kb = generate_empty_kb(self.vocab, entity_vector_length)
         self.scorer = scorer
         self.use_gold_ents = use_gold_ents
         self.candidates_doc_mode = candidates_doc_mode