
Just as there are widely understood empirical laws of nature - for example, what goes up must come down, or every action has an equal and opposite reaction - the field of AI was long defined by a single idea: that more compute, more training data and more parameters makes a better AI model.
However, AI has since grown to need three distinct laws that describe how applying compute resources in different ways impacts model performance. Together, these AI scaling laws - pretraining scaling, post-training scaling and test-time scaling, also called long thinking - reflect how the field has evolved with techniques to use additional compute in a wide variety of increasingly complex AI use cases.
The recent rise of test-time scaling - applying more compute at inference time to improve accuracy - has enabled AI reasoning models, a new class of large language models (LLMs) that perform multiple inference passes to work through complex problems, while describing the steps required to solve a task. Test-time scaling requires intensive amounts of computational resources to support AI reasoning, which will drive further demand for accelerated computing.
What Is Pretraining Scaling? Pretraining scaling is the original law of AI development. It demonstrated that by increasing training dataset size, model parameter count and computational resources, developers could expect predictable improvements in model intelligence and accuracy.
Each of these three elements - data, model size, compute - is interrelated. Per the pretraining scaling law, outlined in this research paper, when larger models are fed with more data, the overall performance of the models improves. To make this feasible, developers must scale up their compute - creating the need for powerful accelerated computing resources to run those larger training workloads.
This principle of pretraining scaling led to large models that achieved groundbreaking capabilities. It also spurred major innovations in model architecture, including the rise of billion- and trillion-parameter transformer models, mixture of experts models and new distributed training techniques - all demanding significant compute.
And the relevance of the pretraining scaling law continues - as humans continue to produce growing amounts of multimodal data, this trove of text, images, audio, video and sensor information will be used to train powerful future AI models.
Pretraining scaling is the foundational principle of AI development, linking the size of models, datasets and compute to AI gains. Mixture of experts, depicted above, is a popular model architecture for AI training. What Is Post-Training Scaling? Pretraining a large foundation model isn't for everyone - it takes significant investment, skilled experts and datasets. But once an organization pretrains and releases a model, they lower the barrier to AI adoption by enabling others to use their pretrained model as a foundation to adapt for their own applications.
This post-training process drives additional cumulative demand for accelerated computing across enterprises and the broader developer community. Popular open-source models can have hundreds or thousands of derivative models, trained across numerous domains.
Developing this ecosystem of derivative models for a variety of use cases could take around 30x more compute than pretraining the original foundation model.
Developing this ecosystem of derivative models for a variety of use cases could take around 30x more compute than pretraining the original foundation model.
Post-training techniques can further improve a model's specificity and relevance for an organization's desired use case. While pretraining is like sending an AI model to school to learn foundational skills, post-training enhances the model with skills applicable to its intended job. An LLM, for example, could be post-trained to tackle a task like sentiment analysis or translation - or understand the jargon of a specific domain, like healthcare or law.
The post-training scaling law posits that a pretrained model's performance can further improve - in computational efficiency, accuracy or domain specificity - using techniques including fine-tuning, pruning, quantization, distillation, reinforcement learning and synthetic data augmentation.
Fine-tuning uses additional training data to tailor an AI model for specific domains and applications. This can be done using an organization's internal datasets, or with pairs of sample model input and outputs.
Distillation requires a pair of AI models: a large, complex teacher model and a lightweight student model. In the most common distillation technique, called offline distillation, the student model learns to mimic the outputs of a pretrained teacher model.
Reinforcement learning, or RL, is a machine learning technique that uses a reward model to train an agent to make decisions that align with a specific use case. The agent aims to make decisions that maximize cumulative rewards over time as it interacts with an environment - for example, a chatbot LLM that is positively reinforced by thumbs up reactions from users. This technique is known as reinforcement learning from human feedback (RLHF). Another, newer technique, reinforcement learning from AI feedback (RLAIF), instead uses feedback from AI models to guide the learning process, streamlining post-training efforts.
Best-of-n sampling generates multiple outputs from a language model and selects the one with the highest reward score based on a reward model. It's often used to improve an AI's outputs without modifying model parameters, offering an alternative to fine-tuning with reinforcement learning.
Search methods explore a range of potential decision paths before selecting a final output. This post-training technique can iteratively improve the model's responses
Most recent headlines
05/01/2027
Worlds first 802.15.4ab-UWB chip verified by Calterah and Rohde & Schwarz to be ...
07/10/2026
Dalet, a leading technology and service provider for media-rich organizations, today announced the latest Long-Term Supported (LTS) release of Dalet Flex. Build...
06/09/2026
June 9 2026, 23:00 (PDT) Dolby and MagentaTV Bring Fans Closer to the FIFA Worl...
04/08/2026
Dalet, a leading technology and service provider for media-rich organizations, t...
23/07/2026
Staines-upon-Thames, UK, 23rd, JULY, 2026 - Yospace, the trusted leader in Dynamic Ad Insertion (DAI), stitched 11.57 billion one-to-one addressable advertisem...
23/07/2026
This Spring 2026 graduate from Honolulu found a home at SoonerVision as a live-production star and is ready to enter the industry as a replay op on Prime Video&...
23/07/2026
As the AUSL postseason begins tonight, listening in to players during games enhances fan engagement for the young organization
Ballplayers chatting with announ...
23/07/2026
Snap Inc. and the Esports Foundation have announced Snap's activations at the Esports World Cup 2026 in Paris, running July 6 through August 23. For the fir...
23/07/2026
Sony has announced the FX5, a full-frame Cinema Line camera built around a newly developed fully stacked Exmor RS CMOS image sensor and BIONZ XR2 image processi...
23/07/2026
TNDV (a division of Live Media Group) provided technical production support for the Pickleball Slam 4 live broadcast from Seminole Hard Rock Live in Hollywood, ...
23/07/2026
LTN has announced it has completed satellite-to-IP migrations for more than 3,00...
23/07/2026
Daktronics has announced the winners of the 2026 DakClassroom Student Showcase, a national competition recognizing student work in media production, storytellin...
23/07/2026
The French Tennis Federation (FFT) and beIN Media Group have announced a five-ye...
23/07/2026
Clear-Com provided the communications infrastructure for the Fryderyk Awards Gal...
23/07/2026
The 2026 FIFA World Cup is, on paper, an unparalleled financial success. Total t...
23/07/2026
Standing inside HQ3 at MediaCity Studios during the FIFA World Cup 2026, it quic...
23/07/2026
Having attracted upwards of 17 billion views throughout the tourney, the broadca...
23/07/2026
Quickly creates reverse reverb effects
Low End Candy have recently released a new plug-in that offers a quick and simple way to create reverse reverb effect...
23/07/2026
New engines, resonators, effects & more
Expressive E's versatile soft synth has just been treated to an update that greatly expands its sonic range. Noi...
23/07/2026
SGL Carbon has further expanded its photovoltaic capacity at its Ort im Innkreis site in Austria. With the commissioning of new PV carports, the sites solar sys...
23/07/2026
NITV launches new By Mob For Mob series in partnership with NRMA Insurance
23 July, 2026
Media releases
National Indigenous Television (NITV) has launched ...
23/07/2026
Rohde & Schwarz presents new software for three-phase power analysis on MXO seri...
23/07/2026
In our latest blog, Laura Rognoni explores how the launch of Play Snooker reflects a wider shift in sports economics - from monetising peak broadcast audiences ...
23/07/2026
Download Now!
To receive a short case study showcasing Vivid Broadcast's audio setup, please enter the following details:
--
First Name* First Name ...
23/07/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
23/07/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
23/07/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
23/07/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
23/07/2026
Witbe brings its AI-native testing and monitoring infrastructure to Amsterdam: demonstrations on real devices, a peer-reviewed technical paper in the conference...
23/07/2026
Quickplay, Caretta Research, Evergent, and Yospace have announced that the renowned IBC Pride Happy Hour will once again bring the vibe and fun back to IBC on F...
23/07/2026
At IBC2026, Interra Systems will present this year's focus: operational AI at scale for QC, monitoring, and captioning workflows. The industry has moved pas...
23/07/2026
Illusionfx is finally coming!
Alex Perea July 23, 2026
0 Comments
The new website is now live to mark the occasion.
Since the last update, feedback h...
23/07/2026
Evergent (5.A86) will showcase its Agentic Revenue Orchestration Platform for streaming services, sports leagues, pay-TV operators at IBC2026 (RAI Amsterdam, 11...
23/07/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
23/07/2026
Share
Copy link
Facebook
X
Linkedin
Bluesky
Email...
23/07/2026
David Dvir, 2D founder (left), Sundeep Reddy (center), and Jean-Marc Bouchut (right), both of Zeiss...
23/07/2026
Screen Australia announces new national documentary accelerator program 23 July 2026
Screen Australia has today announced the Above the Line Documentary Accel...
23/07/2026
X-Rite Pantone and Fashion Institute of Technology (FIT) Partner to Advance Digi...
23/07/2026
Thursday 23 July 2026
New Samsung Galaxy Foldable range available now to pre-order on Sky Mobile
Sky Mobile, the award-winning network voted this year by Expe...
23/07/2026
Thursday 23 July 2026
Jason Statham leads Sky Original action thriller MUTINY in explosive new trailerTurn on cookies to view this content. Go to Privacy optio...
23/07/2026
Wuppertal July 23, 2026
Riedel Implements RefCam and Easy5G for the First Time...
23/07/2026
RT Commercial has today announced Centra as the official food partner once again of The Traitors Ireland, coming to RT One and RT Player later this year.
Ce...
23/07/2026
RT Commercial today announced Honda as sponsor of Today with David McCullagh RT Radio 1, weekdays 9am 11am.
Broadcasting directly after Morning Ireland in ...
23/07/2026
RT News is pleased to announce Barry Gallagher has been assigned as the new journalist covering news from Cork, Kerry and the wider Southern region
Working al...
23/07/2026
Lock in and load up the cloud. GFN Thursday brings fresh updates and new adventu...
22/07/2026
Among other tweaks to coverage, the broadcaster added announce positions in the ...
22/07/2026
The Australian Sports Technologies Network (ASTN) has reported a sharp acceleration in Australia's sportstech sector, with revenue growth rising from 10% in...
22/07/2026
From a retrofitted broadcast bus to packed Match Day Live! events, Bennett detai...
22/07/2026
TMRW Sports has named Brendan Donohue as President, Flag Football, where he will lead the company's professional flag football business in partnership with ...
22/07/2026
Stephen Arnold Music (SAM) designed and produced soundtracks for nine projection...