LectureRechercheRadarCadre d'investissement
Connexion / Inscription
ToutTechnologieProduitsInvestissementBourseAutres

Lecture

2026-11-1448 publications

Actualiser

OpenAI Disbands Stand-Alone Robotics Team, Shifts to General-Purpose Model Control

Sujet · OpenAIMatériel
2026-11-14 01:05 GMT+8

OpenAI has disbanded its stand-alone robotics team, reassigning researchers to the company’s main research team and larger Applied organization. The move is described as an effort to accelerate progress in robotics rather than abandon it. A company spokesperson stated that general-purpose frontier models, including GPT-6 Astra, have demonstrated the capability to directly control robots, reducing the need for specialized AI models built solely as digital “brains” for robotic systems.

Previously, OpenAI rebuilt a standalone robotics team in 2024 and partnered with Figure, but that collaboration ended in February 2025. The company is currently developing its own humanoid robot, with CEO Sam Altman targeting a first public demonstration in 2027. This reorganization reflects a growing industry bet that models designed for language, code, and vision can also power “embodied AI,” eliminating the necessity for separate robotics divisions.

This strategic shift challenges the assumptions underpinning many robotics startups that raised billions on the premise that robots require their own specialized intelligence. Guanya Shi, an assistant professor at Carnegie Mellon University’s Robotics Institute, interpreted the move as a signal that OpenAI views robotics as a downstream application of general intelligence rather than a special problem. While general models show strong performance in simulated benchmarks—such as an Astra-powered agent achieving a 74.8% success rate in an Nvidia study—experts note limitations in real-time responsiveness and offline operation, suggesting a continued role for smaller, specialized onboard models.

Sources :Fortune

Recherche

Onsemi Showcases 20kW AI Data Center Power Prototype Using Vertical GaN

Sujet · 安森美垂直GaN电源技术Prise rapide
2026-11-13 18:30 GMT+8

At the electronica 2026 trade show in Munich, Germany, semiconductor company Onsemi showcased a prototype 20kW-class intermediate bus converter (IBC). The device uses Onsemi's proprietary 1200V-rated vertical gallium nitride (GaN) power devices to directly convert an 800V DC input to a 6V output, targeting the shift toward high-voltage DC power distribution in AI data centers.

As computing power increases in AI servers, traditional 48V or 52V distribution schemes require higher currents that increase transmission losses and cable size. Moving to 800V DC distribution reduces these losses but requires stepping down the voltage near the processors. By converting directly from 800V to 6V, the system eliminates intermediate conversion stages like 48V or 12V buses.

Onsemi's approach relies on its vertical GaN architecture, where current flows through the thickness of the chip rather than across its surface. This structure allows for higher voltage ratings and current density compared to conventional lateral GaN devices. The company claims this design can reduce chip size by approximately one-third compared to commercial lateral GaN products. The prototype utilizes an LLC resonant conversion topology with MHz-band switching frequencies to shrink magnetic components like transformers. According to the display panel at the event, the prototype achieved a conversion efficiency of 96.5%.

Onsemi representatives noted that the 1200V rating avoids the need to series-connect two 650V lateral GaN devices, simplifying the circuit. They also highlighted that GaN's lack of reverse recovery charge benefits electromagnetic interference (EMI) performance. However, they cautioned that GaN devices are sensitive to gate-source voltage control and require careful layout and thermal management due to their small chip size. This demonstration positions Onsemi against competitors such as Renesas Electronics, Texas Instruments, Navitas Semiconductor, STMicroelectronics, and Infineon Technologies, who have also announced developments in 800V-to-6V power systems for AI infrastructure.

Sources :EE Times Japan

Recherche

Google Cloud Launches AI Agent Observability and Governance in BigQuery and Looker

Sujet · 谷歌云AI代理治理Matériel
2026-11-14

As enterprises move AI agents from experimental prototypes into production, data and platform teams are discovering that deploying an agent is only half the battle—you also need to see how it reasons, which data it touches, and what it costs to run. On November 14, 2026, Google Cloud announced expanded AI agent observability and evaluation capabilities across its Google Data Cloud, providing unified visibility into first-party data agents in BigQuery and Looker, as well as custom third-party agents built with frameworks like Agent Development Kit (ADK) or LangGraph.

The core of this update is a unified telemetry framework designed to address three operational challenges: tracing multi-step reasoning while protecting sensitive data, controlling token consumption and query spend, and evaluating answer accuracy before prompt or schema changes reach users. For first-party agents, Conversational Analytics agents emit telemetry automatically; for custom agents, developers can stream structured telemetry into BigQuery via single-line SDK instrumentation. All data lands in customer-owned BigQuery tables, ensuring data sovereignty.

In terms of privacy protection, Google Cloud has introduced a 'fail-closed' mechanism. When a knowledge source uses row-level security, column-level policy tags, or dynamic data masking, the system automatically skips logging to prevent sensitive content leakage. Meanwhile, administrators can enable Cloud Monitoring and Cloud Trace to aggregate metrics and trace steps, but these operations exclude core content such as user prompts and model responses by default. If deeper auditing is required, agent owners can opt in to conversation logging on a per-agent basis, inheriting dataset IAM permissions and Customer-Managed Encryption Keys (CMEK).

To control costs, the new tools offer granular token and latency tracking and allow configuration of per-query BigQuery byte scan caps (big_query_max_billed_bytes), using dry-run estimates to block oversized queries and avoid on-demand scan charges. Additionally, daily input/output token quotas at the project, user, or agent level are in preview. These FinOps guardrails enable platform teams to proactively prevent runaway spend as agent adoption grows.

Beyond runtime observability, Google Cloud launched Prism, an open-source evaluation tool that allows testing agents grounded in BigQuery tables or Looker Explores and streaming evaluation results directly into BigQuery for side-by-side analysis with production logs. This integration enables enterprises to verify whether changes to prompts, models, or data schemas improve SQL accuracy and answer quality before deployment, thereby reducing regression risks.

Sources :Google Cloud Blog

Recherche

SemiAnalysis's Dylan Patel: Blocking Lithography Equipment Slows China's AI More Than Banning Chip Exports

Sujet · 中美AI出口管制Matériel
2026-11-14 07:13 GMT+8

On November 13, 2026, three members of the semiconductor research firm SemiAnalysis—Jordan Schneider, Dylan Patel, and Jordan Nanos—recorded an emergency podcast regarding Jensen Huang's resignation as CEO of Nvidia. During their discussion on policy directions, Dylan Patel proposed a strategic framework that contradicted the prevailing mainstream thinking in Washington at the time: rather than banning Chinese access to remote US computing power, the focus should be on blocking the equipment required to manufacture chips.

Patel's core logic is to 'flip the table.' He uses a jigsaw puzzle analogy to explain: if you remove one piece from the top of the supply chain (such as banning a specific chip export), China will specifically target and overcome that single piece; but if you flip over the entire puzzle board, the difficulty level changes completely. Specifically, instead of banning the chips themselves or restricting model access, it is more effective to prevent companies like ASML from selling DUV lithography equipment to China. Without this equipment, China cannot mass-produce advanced chips, regardless of how it designs architectures or optimizes algorithms.

The second pillar of this strategy is to permit, rather than prohibit, remote computing access. Patel's game-theory reasoning suggests that if companies such as Alibaba, ByteDance, and Moonshot must rent computing power from US cloud service providers, the United States gains three distinct advantages. First, Chinese data may be processed within the US, allowing intelligence agencies access. Second, if tensions escalate, the US can cut off access at any moment—a 'kill switch.' Third, every megawatt of computing deployed outside China generates massive profits for the US ecosystem: Nvidia earns approximately $7 million per deployment, with cloud service providers adding another $7 million to $10 million.

Patel argues that cutting off remote access would instead incentivize smuggling, accelerate the development of China's domestic chip industry (which was going to happen anyway), and eliminate China's economic incentive to maintain the status quo of manufacturing electronics for American companies. He cites Nvidia's largest PCB supplier as a Chinese company as an example. Maintaining the current situation gives China a financial reason to continue cooperation.

However, this argument faces pushback from within the same podcast. Jordan Schneider points out that parts of the Chinese decision-making apparatus (potentially including Xi Jinping himself) are deeply embarrassed by how desperately domestic enterprises want to import GPUs. Even when Trump offered to sell them, the Chinese government hesitated to allow these companies to make purchases. This implies that the RASA Act might provide China with a political excuse to 'force self-reliance,' rather than triggering the comprehensive economic war Patel fears. Schneider also questions whether China has the capacity or willingness to launch a full-scale retaliation—given its slowing economic growth, unemployment rates reaching 20%, and the fact that Trump continues to push forward despite the high costs of the war in Iran, suggesting the US government's pain tolerance may be underestimated.

Sources :chinatalk.media

Recherche

Aviation Journalist Jeff Wise: The 'More Efficient' the AI Air Traffic Control System, the Thinner the Safety Redundancy May Become

Sujet · FAA AI空管裁员Structurel
2026-11-14 19:00 GMT+8

Air traffic control is undergoing a paradox: using AI to address staffing shortages may create new risks that require even more personnel to mitigate. Aviation journalist Jeff Wise traces this logical chain in a long-form investigation for WIRED. In May 2026, the US Federal Aviation Administration (FAA) lowered its target for active controllers from 14,633—set in 2024—to 12,563, citing "modern staffing models and scheduling tools." In September of the same year, the FAA deployed the SMART system, developed by Boston-based company Air Space Intelligence—a cloud platform that uses AI to assess flight schedules, weather, airport capacity, and airspace conditions to predict traffic flows and identify conflicts in advance. The contract totals $875 million over 12 years.

Wise points out that this logic is particularly dangerous in aviation. The industry has long been technologically conservative, not because it is backward, but because when human lives are at stake, time-tested reliability is preferable to cutting-edge unpredictability. Mary Schiavo, former Inspector General of the US Department of Transportation and an aviation lawyer, states plainly: certification standards and airworthiness requirements prove and demonstrate safety, which usually means incremental rather than radical change. Yet AI is penetrating the sector at an unprecedented pace. In September 2026, Joby Aviation completed a 3,200-mile cross-state autonomous flight—from Concord, California, to Dare County, North Carolina—navigating busy airspace and dodging storms without any human touching the controls. A safety pilot remained on board throughout, but never needed to intervene.

The technical risks are more subtle than they appear. Current research focuses on deterministic neural networks—models with frozen weights after training, which do not produce "hallucinations" like large language models. Wise cites a warning from Darren Cofer, a researcher at Collins Aerospace, in a 2021 paper: when such systems encounter situations not present in their training data, they generate "unexpected (and potentially dangerous) outputs." This is not a hallucination problem, but an out-of-distribution generalization issue—the model does not admit ignorance in scenarios it has "never seen," but instead provides uncalibrated outputs.

Political pressure is accelerating this process. A recently retired FAA employee, speaking anonymously to Wise, said: "We're introducing all this new technology to improve the system, and it's going to be a disaster. It's literally a Pandora's box we will come to regret." He attributes this to the "Trump/DOGE/Musk effect"—pushing through new technologies without fully understanding the complexity of the national airspace system. On October 13, nearly two dozen Democratic members of Congress wrote to the US Government Accountability Office (GAO), requesting a review of SMART due to concerns that it might "hallucinate safety-critical information."

On November 13 (Friday), the FAA issued a nationwide ground stop order, instantly throwing global airline schedules into chaos. The US government quickly stated there was no evidence of a cyberattack and no indication that SMART was linked to the communication failure. However, when asked if he could rule out an AI failure, US Secretary of Homeland Security Markwayne Mullin replied only: "There’s nothing certain right now."

Sources :WIRED

Recherche

CrowdStrike Explains Centralized Control Principles for AI Traffic Gateways: What It Inspects and What It Cannot Block

Prise rapide
2026-11-13 14:00 GMT+8

As AI agents shift from answering questions to executing actions, security teams face not just a visibility problem, but a control problem. In an explanatory article accompanying the general availability of its Falcon AI Gateway, CrowdStrike proposes a specific principle: make the approved path the only path. This means that teams use existing network controls to block applications and workloads from connecting directly to model providers, allowing only traffic routed through the gateway to reach the model endpoints.

The architecture operates in a straightforward manner: the gateway authenticates each request, routes it to the correct provider and model, applies usage limits, and enforces Falcon Guardian policies both before the request reaches the provider and when the response returns. Administrators can switch between Anthropic, Amazon Bedrock, Microsoft Foundry, and Google Vertex AI without reconfiguring every workload. Applications simply change their base URL to point to the gateway and use a gateway key, allowing them to continue using their original API formats.

The most valuable part of the article is the clear delineation of the gateway's capability boundaries. Falcon AI Gateway and the Falcon sensor share the same Falcon Guardian policy model, but enforce it at different layers: the sensor acts on the endpoint, including local processes, files, and commands run by the agent; the gateway acts on traffic routed through it and cannot stop local processes. This means that if an AI agent performs dangerous operations locally, the gateway will not see it; conversely, if the agent sends prompts containing sensitive data to the model, the sensor might not see it, but the gateway can intercept it.

Because traffic flows through a single point, the AI Kill Switch in Falcon Guardian can cut off agents' access to the model via traffic routed through Falcon AI Gateway. However, this capability is limited to traffic routed through the gateway—if an agent bypasses the gateway to call the model directly, the Kill Switch is ineffective. CrowdStrike states that this architecture stems from its experience running an internal AI Hub gateway, which has served tens of thousands of employees globally for years. A key lesson learned during this process was to keep the request path as short as possible, because every additional hop adds another failure point and latency.

For teams evaluating AI agent security architectures, the practical contribution of this article is providing a clear layered model: the traffic layer gateway handles prompt inspection, response filtering, usage control, and the Kill Switch; the endpoint layer sensor handles local processes, file operations, and command execution. The two cover different attack surfaces and are indispensable to each other. However, it must be remembered that this is entirely a vendor's description of its own product design, and its actual interception effectiveness and policy coverage have not been independently verified.

Sources :CrowdStrike.com

Recherche

Analysis Claims Chinese AI Models Kimi K3 and DeepSeek V4 Trained on Banned Nvidia Chips

Sujet · 中美AI出口管制Matériel
2026-11-14 17:54 GMT+8

According to a report by German tech outlet t3n, Will DePue, head of the non-profit research group Unemployment Labs, claims to have found "convincing evidence" that Moonshot AI's Kimi K3 and DeepSeek's V4 series models were trained on Nvidia Blackwell GPUs. The analysis relies on examining connection strengths within model weights, which are said to reflect the configuration of the underlying training hardware. These high-performance chips are subject to strict US export controls prohibiting direct sales to China.

DePue, a former OpenAI employee who worked on post-training for o3 and 4o models, stated that large language model weights contain signatures of the compute infrastructure used during training. This finding offers new technical clues regarding how Chinese AI laboratories may be circumventing restrictions on advanced compute access. Previously, White House technology advisor Michael Kratsios cited intelligence sources alleging illegal use of Nvidia chips by Moonshot AI, while Epoch AI estimated that approximately two million advanced chips flowed into China through indirect channels between 2024 and 2025.

TechCrunch notes that this analysis does not prove the models relied exclusively on Nvidia hardware nor necessarily indicate legal violations, as the chips could have been located in offshore data centers. Regulatory consequences from these findings remain unclear.

Sources :t3n.de

Recherche

Runway Releases Open-Source Praxis-1: A Video-Trained Robot Control Model Claiming Top Performance in Its Category

Sujet · Praxis-1Matériel
2026-11-14 17:45 GMT+8

AI company Runway released the open weights and source code for its robot control model, Praxis-1, on November 14, 2026, under the commercially permissive Apache 2.0 license. The model's core innovation lies in its "video-to-robot" transfer learning path: it first learns how objects move and interact from ordinary footage, then fine-tunes on videos of human actions and limited robot data. This allows it to predict environmental changes and calculate corresponding robot movements based on camera images, text instructions, and position data.

Runway calls Praxis-1 a "World Action Model," aiming to reduce reliance on expensive and time-consuming task-specific robot datasets. According to the company, adapting the model to a Franka robot arm required only about 277 hours of recordings from the public DROID dataset, while a dual-arm robot used between 1,400 and 2,600 hours of training material. Whether this approach actually needs less robot data overall than comparable models without general video pretraining remains an open question.

In terms of performance, Runway evaluated Praxis-1 itself on Nvidia's RoboLab-120 simulation benchmark, which tests sorting, grasping, or stacking across 120 tasks. The results show a 46.2 percent success rate, higher than FLUX 3 Action (42.9 percent), also built as a World Action Model, placing it at the top of that category. However, on the overall leaderboard, it ranks fifth behind two AI agents and two robot models with different architectures. It must be emphasized that all test results were run by Runway and have not been independently verified.

Beyond simulations, Runway conducted comparative experiments on a real Franka robot arm. When completing ten identical tasks (such as placing a banana on a plate or closing a drawer) with ten attempts each, Praxis-1 succeeded 88 times, while FLUX 3 Action managed 80. Additionally, in transfer tests on a mobile robot platform never seen before, after fine-tuning on about 95 hours of data, Praxis-1 hit an average success rate of 80 percent in simulated tests, significantly higher than Google's original RT-1 model's 52.4 percent.

Despite showing potential, Praxis-1 has clear limitations. According to Runway's research report, the model sometimes confuses similar-looking objects, struggles with vague instructions, and cannot reliably plan complex long-sequence actions autonomously, still requiring manual step decomposition. Currently, the standard model version requires a GPU with 80 GB of memory, while a compressed version is estimated to fit on an RTX 5090 (32 GB VRAM), though actual peak memory usage hasn't been measured. Developers can download the model from Hugging Face, with code hosted on GitHub.

Sources :The Decoder

Recherche

Jensen Huang's Departure Disrupts AI Startup Funding: Nvidia's 'Valuation Anchor' Effect Weakens Under Interim CEO

Sujet · Jensen Huang ResignationMatériel
2026-11-14 06:43 GMT+8

The resignation of Nvidia founder and former CEO Jensen Huang is sending shockwaves through the artificial intelligence startup funding ecosystem. According to TechCrunch, multiple venture capitalists and founders note that Huang's departure removes one of the most significant sources of capital and valuation benchmarks in the AI sector. Over the past few years, Nvidia has poured tens of billions of dollars into startups annually, participating in 105 rounds this year alone, often as the largest investor. "If that direction changes, it'll be a big blow to so many young companies," said Maria Davidson, founder of Relational Capital.

Huang's investment strategy was distinctly personal, favoring sectors like neoclouds, inference providers, and open-weight models that require massive upfront capital. Umesh Padval, managing partner at Seligman Ventures, stated that most major investment decisions came directly from Huang himself. For instance, Nvidia backed CoreWeave while it was still private and participated in two of Crusoe's rounds, including its $1.4 billion Series E. It also acquired Hugging Face for approximately $13 billion and holds stakes in Mistral AI, Lambda, and Black Forest Labs. Last year, Nvidia wrote an $800 million check into Reflection AI's $2 billion Series B round before the startup generated meaningful revenue.

With Huang stepping down, Chief Financial Officer Colette Kress has been named interim CEO. Investors widely expect Kress to adopt a more conservative approach, unlikely to maintain the same level of spending on the early-stage ecosystem. "She's much more conservative," one investor told TechCrunch. "I don't think she's going to commit the same amount of money to keep the early-stage ecosystem going." This shift could lead to lower valuations for startups, as Nvidia's participation previously provided comfort to other investors, enabling them to follow on at lofty prices.

While some hope that large tech rivals like Google, Amazon, Microsoft, and AMD will fill the void, Dharmesh Thakker, a general partner at Battery Ventures, doubts they will pour hundreds of millions into pre-revenue infrastructure startups the way Nvidia did. He suggests these competitors are more likely to back mature startups to drive adoption of their own chips and clouds. If Nvidia becomes more selective, it may leave room for smaller competitors but also signals tighter capital conditions for specific AI subsectors reliant on its strategic investments.

Sources :TechCrunch

Recherche

Meta Muse AI Calling Now Available to US Paying Subscribers, Some Businesses Hang Up on AI

Sujet · Meta MusePrise rapide
2026-11-13 23:12 GMT+8

Meta's AI agent Muse has opened its phone-calling feature to all paying US subscribers after testing in September 2026. According to Meta AI chief Alexandr Wang, Muse has more than 10 million users and handles calls to cable companies, the DMV, and restaurants. The Decoder reports that many businesses hang up immediately when they detect an AI on the line; Meta temporarily tested human operators for Muse calls in September.

Sources :The Decoder

Recherche

Grace Huckins Tests Humanoid Cleaning Robot Service: Currently Relies on Human Teleoperation, Each Cleaning Session Trains the Autonomous Model

Sujet · Tau RoboticsMatériel
2026-11-14 15:45 GMT+8

San Francisco-based startup Tau Robotics advertised a new service on social media: for $30, a humanoid robot will come to your apartment to clean. Free science journalist Grace Huckins saw this post and immediately joined the waitlist. She is not someone who typically hires cleaners, but driven by journalistic curiosity, she wanted to see what this service was really like. To get a genuine experience, she did not disclose to Tau Robotics that she was a journalist, registering only with her real name.

Tau Robotics' long-term goal is to make autonomous cleaning robots a reality, but Huckins points out directly in her article: they are far from achieving this at present. Currently, these robots are teleoperated by humans who observe the scene through cameras attached to the robot's head and wrists. As the robots clean, they continuously collect data, which Tau Robotics plans to use to train more autonomous follow-up models.

This means that every successful wipe of a countertop and every successfully folded T-shirt helps improve the model controlling the robot. The $30 paid by consumers buys a human-teleoperated cleaning session combined with data collection—the robot itself acts as a data acquisition terminal, the human operator performs the actual cleaning work, and the collected operational data is used to improve future autonomous systems.

Specific failures observed by Huckins include the robot overheating and tipping over, issues directly highlighted in the article's title. These robots are clearly unsuited for climbing stairs and require significant human guidance. However, Huckins also noted that the robot is indeed learning and is capable of cleaning toilets.

At the end of the article, Huckins disclosed her identity: she is a free science journalist pursuing a PhD in neuroscience at Stanford University and a contributor to the US edition of MIT Technology Review. The main body of the article was cut off by a paywall; the loaded text is sufficient to confirm the core mechanisms—teleoperation, camera placement, data collection purpose, and current limitations—but full testing details and further failure descriptions are not visible.

For readers interested in AI and robotics, this test provides specific firsthand observation: current consumer-facing humanoid robot services, while advertising autonomous capabilities, actually rely on human teleoperation for operation, with the cleaning task itself serving as a source of training data. Whether Tau Robotics represents a broader industry pattern cannot be determined from this single report alone, but it does showcase one concrete approach to combining consumer services with model training.

Sources :t3n.de

Recherche

Speech-to-Text Company AssemblyAI: Distributing Across GPU Instance Families with Queue Retries Reduces Spot Inference Costs by 70%

Sujet · AssemblyAIMatériel
2026-11-14 03:35 GMT+8

Speech-to-text inference is a GPU-intensive workload. Each audio file triggers a deep learning pipeline comprising acoustic models, language models, and post-processing stages, running on NVIDIA GPUs to meet latency requirements. At scale, this means thousands of GPU instances are running continuously. AssemblyAI needed a way to run these workloads at scale while maintaining availability and latency commitments; using On-Demand Instances exclusively was not cost-effective.

Spot Instances offer significant discounts, but Amazon Elastic Compute Cloud (Amazon EC2) can reclaim instances with a two-minute notice when capacity is needed elsewhere. The challenge was: how to absorb Spot interruptions gracefully while maintaining service level agreement (SLA) commitments? AssemblyAI’s answer involves four pillars working in concert: ECS capacity providers mixing Spot and On-Demand instances, diversification of Spot across multiple GPU instance families, a queue-based retry architecture, and multi-region deployment to satisfy data residency requirements.

The most transferable insight is instance family diversification. AssemblyAI’s infrastructure spans three EC2 GPU instance families: g4dn, equipped with NVIDIA T4 GPUs, provides a cost-effective baseline for standard speech recognition models; g5, equipped with NVIDIA A10G GPUs, offers higher throughput for more demanding model architectures; and g6, featuring NVIDIA L4 GPUs, delivers improved performance-per-dollar for larger models, including large language model inference. By spreading demand across instance families and sizes, AssemblyAI draws from a significantly larger pool of available Spot capacity. These three families tap into independent capacity pools, so when one family experiences capacity pressure, the others remain available. The article explicitly states that operating at this scale, diversifying across instance families and sizes is the most effective lever for minimizing Spot interruption rates.

The recovery path during interruptions is equally critical. When a Spot instance receives a two-minute interruption notice, tasks that can complete within the window finish normally; tasks requiring more time are returned to the SQS queue, becoming immediately available to other workers. Another healthy GPU instance may pick up the message and complete the inference on a different instance type or in a different Availability Zone. To shorten recovery time, AssemblyAI creates custom AMIs containing model artifacts and container images, which take approximately two minutes to build, allowing new instances to start inference directly without pulling images or downloading models. For end customers, the impact of a Spot interruption is a slight increase in processing time, but the transcription ultimately completes successfully and is delivered as expected.

At the capacity scheduling level, AssemblyAI configures multiple capacity providers within a single capacity provider strategy in each ECS cluster. An On-Demand capacity provider serves as an always-on baseline, while a Spot capacity provider spans different GPU instance families and Availability Zones to handle demand spikes. Each capacity provider is associated with an EC2 Auto Scaling group, and ECS managed scaling automatically adjusts the number of GPU instances based on capacity provider weights and pending inference tasks. When the task queue grows, the Spot capacity provider requests additional instances from the diversified pool; when demand drops, instances scale in gracefully.

Sources :Amazon Web Services

Recherche
Suivant →
Connexion / Inscription
LectureRechercheRadarCadre d'investissement

Patel responds that every escalation-retaliation cycle in past US-China trade frictions ultimately led to compromise rather than total decoupling, because neither side could afford the cost of broken supply chains. He acknowledges he is arguing from the position of a 'devil's advocate' and concludes that one can be a hawk and still oppose the RASA Act, but it is difficult to oppose the MATCH Act.

From an industrial investment perspective, Patel's framework implies that if the MATCH Act passes, lithography equipment vendors like ASML will face structural risks to their revenue from China; meanwhile, Nvidia, cloud service providers, and compute rental platforms could continue to profit from permitted remote access. This judgment assumes that the US government indeed chooses the equipment blockade path over total decoupling, and that China does not retaliate with equivalent leverage using rare earths, market access, or other tools.

Two parallel systems will emerge in the future: one driven by human decision-making, the other by AI—in selected corridors near drone hubs and eVTOL vertiports. Michael McCormick, a professor of air traffic management at Embry-Riddle Aeronautical University, explains why traditional ATC cannot cope: "If you try to handle that volume and complexity of aircraft movements, the National Airspace System would be overwhelmed." But Wise’s conclusion is cautious: any computer system can be hacked; any AI system creative enough to devise novel solutions could also hallucinate crucial details. Without a human in the loop, seemingly obvious errors might go unnoticed until it is too late.

Wise documents a unfolding causal chain: the promise of AI efficiency is used to justify workforce reductions, which in turn weaken the oversight capacity needed to guard against AI failure modes. He quotes the retired FAA employee as the underlying tone of the entire piece—"We're introducing all this new technology to improve the system, and it's going to be a disaster"—and notes that in aviation, "efficiency" calculations can never look solely at labor costs: when a single error can endanger dozens of lives, redundancy is not waste, but safety itself.

Boundaries to emphasize: all performance data—including the 70% cost savings, approximately fourfold year-over-year growth in compute demand, and the approximately two-minute AMI creation time—are self-reported by AssemblyAI and AWS. The 70% figure compares against On-Demand pricing; the article notes that running entirely on On-Demand Instances would be cost-prohibitive but does not disclose the mixed configuration AssemblyAI previously operated. This pattern applies to short-lived, independently retriable GPU computing tasks; training jobs requiring long-term continuous state maintenance or inference scenarios maintaining context across requests do not fit the queue-retry assumption. Furthermore, Spot availability for g4dn, g5, and g6 in specific European regions is subject to AWS capacity fluctuations, meaning the effectiveness of the diversification strategy depends on the actual independence of these capacity pools.