TPU

E97074

AI accelerator application-specific integrated circuit tensor processing unit

A TPU (Tensor Processing Unit) is a specialized hardware accelerator designed by Google to efficiently perform large-scale machine learning and deep learning computations.

Try in SPARQL Jump to: Surface forms Disambiguation Statements Elicitation Referenced by

All labels observed (10)

Label	Occurrences
TPU v5e	2
TPU v5p	2
TPU canonical	1
TPU VM	1
TPU pod	1
TPU v1	1
TPU v3	1
TPU v4	1
TPU v5lite	1
Tensor G3 TPU	1

How this entity was disambiguated

This entity first appeared as the object of triple T816538 — resolving that mention is where its identity was fixed. The disambiguator weighed these candidate entities and picked the highlighted one (or “None”, minting a new entity). This is how homonymy is resolved: the same surface form can point to different entities.

NED1 Entity disambiguation (via context triple) gpt-5-mini-2025-08-07

Target entity: TPU
Context triple: [TensorFlow, supportsHardware, TPU]

A. TPE
TPE is the three-letter IOC and international sporting code used to represent Chinese Taipei (Taiwan) in global competitions and events.
B. TPA
TPA is an abbreviation commonly used for a Tri-Party Agreement, a legal contract involving three separate parties that defines their respective rights and obligations.
C. PLA
The PLA is the unified military organization of the People's Republic of China, encompassing its ground, naval, air, rocket, and strategic support forces.
D. T-MEC
T-MEC is the Spanish-language name for the United States–Mexico–Canada Agreement, the trade pact that replaced NAFTA in North America.
E. TEC
TEC is the commonly used acronym for the Episcopal Church, a mainline Anglican Christian denomination based in the United States.
F. None of above. chosen
G. Unsure - the case is ambiguous/there is not enough information to decide.

NED2 Entity disambiguation (via description) gpt-5-mini-2025-08-07

Target entity: TPU
Target entity description: A TPU (Tensor Processing Unit) is a specialized hardware accelerator designed by Google to efficiently perform large-scale machine learning and deep learning computations.

A. TPE
TPE is the three-letter IOC and international sporting code used to represent Chinese Taipei (Taiwan) in global competitions and events.
B. TPA
TPA is an abbreviation commonly used for a Tri-Party Agreement, a legal contract involving three separate parties that defines their respective rights and obligations.
C. PLA
The PLA is the unified military organization of the People's Republic of China, encompassing its ground, naval, air, rocket, and strategic support forces.
D. T-MEC
T-MEC is the Spanish-language name for the United States–Mexico–Canada Agreement, the trade pact that replaced NAFTA in North America.
E. TEC
TEC is the commonly used acronym for the Episcopal Church, a mainline Anglican Christian denomination based in the United States.
F. None of above. chosen

Statements (51)

Predicate	Object
instanceOf	AI accelerator ⓘ application-specific integrated circuit ⓘ tensor processing unit ⓘ
architectureType	systolic array ⓘ
availableVia	Google Cloud ⓘ surface form: Google Cloud Platform
competesWith	FPGA-based accelerators ⓘ GPU ⓘ
componentOf	Google Cloud ⓘ surface form: Google Cloud AI infrastructure
deploymentModel	cloud service ⓘ on-premise appliance (TPU pods) ⓘ
designedBy	Google ⓘ
firstDeploymentContext	Google data centers ⓘ
firstIntroducedBy	Google ⓘ
hasGeneration	Tensor Processing Unit ⓘ surface form: TPU v1 Tensor Processing Unit ⓘ surface form: TPU v2 TPU self-linksurface differs ⓘ surface form: TPU v3 TPU self-linksurface differs ⓘ surface form: TPU v4 TPU self-linksurface differs ⓘ surface form: TPU v5e TPU self-linksurface differs ⓘ surface form: TPU v5lite TPU self-linksurface differs ⓘ surface form: TPU v5p
keyFeature	energy efficiency for ML workloads ⓘ high throughput for matrix operations ⓘ tight integration with Google data center network ⓘ
manufacturer	Google ⓘ
offeredAs	TPU self-linksurface differs ⓘ surface form: TPU VM Tensor Processing Unit ⓘ surface form: TPU node TPU self-linksurface differs ⓘ surface form: TPU pod
optimizedFor	large-scale linear algebra ⓘ matrix multiplication ⓘ tensor operations ⓘ
programmedWith	TensorFlow high-level APIs ⓘ TPUs (via XLA integrations) ⓘ surface form: XLA (Accelerated Linear Algebra)
supportsFramework	JAX ⓘ PyTorch ⓘ surface form: PyTorch (via XLA or specialized backends) TensorFlow ⓘ
supportsNumericFormat	bfloat16 ⓘ float32 ⓘ int8 ⓘ
targetWorkload	computer vision models ⓘ natural language processing models ⓘ recommendation systems ⓘ speech recognition models ⓘ training large language models ⓘ
usedByService	Google Assistant ⓘ Google Photos ⓘ Google Search ⓘ Google Translate ⓘ
usedFor	deep learning workloads ⓘ machine learning workloads ⓘ neural network inference ⓘ neural network training ⓘ

How these facts were elicited

Referenced by (12)

Full triples — surface form annotated when it differs from this entity's canonical label.

TensorFlow → supportsHardware → TPU ⓘ

Pixel 8 Pro → neuralProcessingUnit → TPU ⓘ

this entity surface form: Tensor G3 TPU