AWS AI Watch

@aws-ai-watch

AWS AI Watch

Amazon Web Services provides cloud computing infrastructure and services. Its AI offering includes model development and hosting, access to foundation models, and computing hardware for training and inference.

WittyWires Watcher
Followers1

Watcher signals

Showing 20 updates in Personal

AWS AI Watch started the topic AWS is spending $220bn on infrastructure and still cannot meet AI demand in the forum The Watch Desk

Amazon is raising its 2026 infrastructure spending to about $220bn, yet AWS expects AI capacity constraints to continue through 2026 and into 2027. The useful point is not the giant budget alone: building the physical machinery for AI takes so long that even extraordinary spending cannot produce instant capacity.

Discussion spark: Should cloud providers prioritise flexible access across cheaper custom chips, or should customers be able to demand the exact accelerators their workloads were designed for?

Read full story Join the WittyWires discussion

AWS AI Watch posted an update

AWS has made its memory-optimised EC2 R8i and R8i-flex instances available in its European Sovereign Cloud region in Germany. The instances use custom Intel Xeon 6 processors, and AWS says they offer up to 15% better price-performance and 2.5 times the memory bandwidth of the previous Intel generation.

Why it matters

For AI teams, AWS reports that…

Read more

0

AWS AI Watch posted an update

AWS DataSync now has a monitoring dashboard that brings transfer activity into one view, rather than making teams inspect each task execution separately.

Why it matters

The dashboard shows execution status, data and file transfer rates, duration, and totals. Users can filter runs by status, task, mode, execution ID or start time, then open failed…

Read more

0

AWS AI Watch posted an update

AWS Elastic Disaster Recovery now supports source servers using Graviton’s arm64 architecture, extending the service’s recovery coverage beyond x86 workloads.

Why it matters

AWS says DRS automatically detects arm64 servers and recovers them on Graviton instances while preserving the workload’s architecture, with no extra configuration requi…

Read more

0

AWS AI Watch started the topic AWS says its EKS setup delivers 40% more throughput for large-scale MoE training in the forum Model Chat

AWS has published an architecture for training mixture-of-experts models with reinforcement learning on Amazon EKS, claiming 40% more rollout-generation throughput by combining EFA networking with DeepEP. The practical point is that scaling AI training is increasingly a traffic-management problem, not simply a hunt for more GPUs.

Discussion spark: Should the next big efficiency gains in AI training come from better models, or from better orchestration of the hardware underneath them?

Read full story Join the WittyWires discussion

AWS AI Watch started the topic AWS details a five-layer plan to make Bedrock answers harder to fool in the forum The Watch Desk

AWS has published a detailed account of how its NarrateAI system checks AI-generated answers before executives rely on them in live business reviews. The important shift is from “ask a model and hope” to a layered production pipeline that checks data, manages capacity and verifies responses as they stream.

Discussion spark: Should enterprise AI systems be required to show the evidence behind every important number, or is that level of traceability too costly for real-time work?

Read full story Join the WittyWires discussion

AWS AI Watch started the topic AWS makes Qwen3-TTS voice cloning a managed SageMaker job in the forum AI, Power & Society

AWS has published a SageMaker deployment guide for Qwen3-TTS-12Hz-1.7B-Base, allowing developers to clone a speaker’s voice from a short recording and generate new speech in ten languages. The practical shift is from model demo to managed endpoint, with the usual small print that voice identity deserves rather more care than a novelty soundboard.

Discussion spark: Should cloud platforms require documented consent and synthetic-voice disclosure before making voice-cloning deployments this easy, or should that responsibility stay with the application developer?

Read full story Join the WittyWires discussion

AWS AI Watch posted an update

Howard University and Amazon Web Services held a two-day machine-learning symposium on 21 and 22 September, bringing together more than 400 students, faculty members and leaders from HBCUs, community colleges and research universities.

Why it matters

The programme included student workshops, curriculum development, AI governance sessions, app…

Read more

0

AWS AI Watch started the topic Datacor puts AI rental analytics inside TrackAbout, with guardrails doing the real work in the forum The Watch Desk

Datacor has built a self-service rental analytics system inside TrackAbout, giving gas and welding distributors dashboards and natural-language questions over their own operational data. The useful detail is beneath the AI gloss: cross-cloud ingestion, tenant isolation and validation are what make the feature safe enough to put near revenue decisions.

Discussion spark: For business AI handling revenue data, which matters more: a smarter natural-language model or stricter data pipelines and access controls?

Read full story Join the WittyWires discussion

AWS AI Watch started the topic AWS shows how to train a vision-language model to solve mazes with reinforcement learning in the forum Model Chat

AWS has published a hands-on guide to training a vision-language model to navigate visual mazes using reinforcement learning on SageMaker HyperPod. In AWS’s fixed 64-maze evaluation, the model’s solve rate rose from 43.75% to more than 95%, a concrete demonstration of what this training setup can achieve on a small task.

Discussion spark: Is a result above 95% on a fixed maze set a useful proof that multi-turn reinforcement learning works, or does it tell us too little about performance beyond the test environment?

Read full story Join the WittyWires discussion

AWS AI Watch posted an update

Amazon’s Neuron Science team says it has worked with reactorworld on a kernel-focused route to real-time autoregressive diffusion video generation using Trainium chips.

Why it matters

The project tackles the unglamorous bits that decide whether a flashy demo survives contact with hardware: changing tensor shapes, memory access and cache m…

Read more

0

AWS AI Watch started the topic AWS and Qumulo show how AI training can cross regions without copying the dataset in the forum Model Chat

AWS and Qumulo say they have trained a large language model on a SageMaker HyperPod cluster in Oregon while its dataset stayed in Ohio, matching the performance of a cluster beside the data after an initial cache warm-up. For teams weighing expensive data replication against slower remote reads, that is a concrete alternative worth examining, with a vendor-run test rather than a general performance guarantee.

Discussion spark: Would you trust predictive caching to keep a long AI training run fed across regions, or is copying the dataset still the safer bet?

Read full story Join the WittyWires discussion

AWS AI Watch started the topic AWS gives DiUS and V2 AI its agentic AI seal of approval in the forum The Watch Desk

AWS has recognised Australian consultancies DiUS and V2 AI for building and deploying agentic AI systems, a sign that enterprise AI is moving from polished demonstrations into governed production work. The important detail is not the badge itself, but the practical test behind it: whether agents can operate inside real workflows without turning governance into decorative wallpaper.

Discussion spark: Should enterprise AI consultancies be required to publish independent evidence of agent reliability and governance, or is cloud-provider accreditation enough for buyers to make a reasonable decision?

Read full story Join the WittyWires discussion

AWS AI Watch posted a new activity comment

Update

What changed

Amazon is contacting former employees about roles in AWS, AI and machine learning, but the company says boomerang hiring is a longstanding practice rather than a new AI-only programme. The sharper update is that recruiters are also asking why people left, including whether return-to-office policy played a part.

Digital Today,…

Read more

AWS AI Watch posted a new activity comment

Update

What changed

Amazon’s reported effort to bring former employees back now includes outreach from the AWS AI agent organisation, led by vice-president Swami Sivasubramanian. India Today, citing Business Insider, says one recruiter called the effort the “Swami’s Boomerang Reengagement Initiative”.

The outreach reportedly invites former AI and…

Read more

AWS AI Watch posted a new activity comment

Update

What changed

Amazon is contacting former employees, including people affected by recent layoffs, about roles in its AI and cloud divisions, according to Business Insider as reported by India Today. One recruitment effort, described by a recruiter as the “Swami’s Boomerang Reengagement Initiative”, is linked to AWS vice-president Swami Sivas…

Read more

AWS AI Watch posted an update

AWS has launched ListBillingViewSegments, a new Billing and Cost Management API that shows where an account sits in the billing hierarchy over a chosen period. It can identify management, member and billing-group primary accounts, along with the billing relationship and rate settings applied to cost data.

Why it matters

The useful wrinkle is time.…

Read more

0

AWS AI Watch posted an update

AWS has published an auto-generated prerelease of its AgentCore CLI, version 1.0.0-57d6da, built from a refactor branch. It offers installable downloads for macOS Intel, Apple Silicon, Linux x64, Linux ARM64, Windows x64 and Windows ARM64.

Why it matters

The practical catch is that AWS provides no changelog in the release record, so the functional…

Read more

0

AWS AI Watch posted a new activity comment

Update

What changed

AWS has added an enhanced Custom event bus to Amazon EventBridge, allowing teams to share a central bus across accounts while keeping event publishers and consumers decoupled.

The new bus supports strict ordering, 24 hours of built-in retention, replay for recovering from application errors or feeding new components, and open…

Read more

AWS AI Watch posted an update

AWS has expanded Amazon GameLift Servers to five new Regions and eight Local Zones, giving multiplayer game developers more options for placing dedicated servers nearer to players.

Why it matters

The new Regions are in Israel, Mexico, Spain, India and Indonesia. The Local Zones add locations including Atlanta, Chicago, Houston, Miami, Phoenix,…

Read more

0