Developing a new drug can take years and cost hundreds of millions of dollars, and even then, most candidates ultimately fail. Now, researchers at Stanford have built a virtual biotech company with 37,000 AI agents that work together to analyze drug targets and design therapies.

Roughly 90 percent of drugs that enter clinical trials never reach the market. That’s often because promising results in the lab don’t translate to patients, or the drug causes dangerous side-effects not caught earlier in the development process.

Part of the problem is the evidence that could help catch these issues earlier in the process is scattered across disciplines and formats, making it hard for any single team to weigh it all.

To get around this, a Stanford team created a system they call a virtual biotech, which consists of up to 37,000 AI agents built to mimic the divisions of a real drug-development company. In a paper published in Science, the system identified which types of drug targets are more likely to succeed in clinical trials and even proposed a lung cancer treatment that a major drugmaker later landed on too.

“Our idea was to see how far we could push this. Could we create a biotech company that takes on everything from looking for drug targets all the way to designing clinical trials?” senior author James Zou said in a press release.

The new system features a virtual chief scientific officer (CSO) that takes a query from a human user and then delegates tasks to an army of specialized “scientist” agents working on the problem.

These agents are armed with their own databases and tools and are split into one of four divisions that specialize in finding and validating drug targets, assessing safety risks, choosing how a drug should be delivered, and reviewing existing clinical trial data. The system has built-in access to the Open Targets database, a massive public repository of clinical trial data.

To test the system, the researchers gave it an existing study showing that genetic evidence can help predict which drugs succeed in trials and asked it how to build on that research. The CSO decided the first step was to improve the quality of the data it had access to because many trials in the Open Targets database don’t clearly record whether the drug actually worked.

So, it asked its researcher agents to dig through the outcomes of 37,075 individual Phase II and III trials, assigning one agent to each trial. The agents searched trial registries, published papers, and press releases for results. They crunched through the job in about six hours—a fraction of the time it would take a team of humans.

The CSO asked another agent to look for promising gene candidates by scouring a public database of human tissues showing which genes are switched on in which cell types. It came up with a two-part scoring system, which first measured whether a gene was active in just one type of cell or across many and then gauged whether its activity was controlled more like an on-off switch or could be dialed up and down like a dimmer switch.

Comparing those scores to the updated trial outcome data revealed a pattern. Drugs aimed at switch-like genes only found in a small number of cell types were 48 percent more likely to eventually reach the market, 40 percent more likely to advance from Phase 1 to Phase 2 trials, and had 32 percent fewer adverse events than drugs hitting more broadly active targets.

The researchers then pushed the system further, asking it to evaluate a protein called B7-H3 that’s associated with lung cancer. The agents discovered the protein was particularly common in connective-tissue cells called fibroblasts that are often found close to tumor cells.

The agents then discovered evidence those cells were suppressing the activity of nearby immune cells, preventing the body from detecting and reacting to the tumors. The system proposed a therapy that would tag cells expressing B7-H3 with an antibody to help direct a toxic chemotherapy drug to them.

The virtual biotech came up with its solution based solely on data available before January 2025, but in August of that year a major pharmaceutical company arrived at the same strategy independently, when its B7-H3-targeted therapy ifinatamab deruxtecan received FDA breakthrough therapy status. “This was really exciting as an independent, third-party validation that’s consistent with the effects and the design proposed by the virtual biotech,” Zou said.

However, coming up with drug targets is just one step in a long, expensive drug discovery process. While refining the candidate selection process could prevent drug companies from pursuing some obvious dead ends, it can’t speed up the rigorous lab testing and clinical trials required to get a drug to market.

Nonetheless, given the industry’s woeful record at translating promising science into finished products, an army of AI scientists that can significantly speed up a critical part of the drug discovery pipeline could be just what the doctor ordered.

The post Virtual Biotech Company Puts 37,000 AI Agents to Work on Drug Discovery appeared first on SingularityHub.

GPT-6 Astra helps Hex’s data agents turn answers into interactive visualizations that employees are proud to share.

Even if big AI companies agree to a pause, ensuring that nobody tries to sneak ahead could prove tricky.

You’re deploying a model on a system. It starts up, prompts are getting responses. Now the hard question: Is this fast? Your instincts might lead you to send...

You’re deploying a model on a system. It starts up, prompts are getting responses. Now the hard question: Is this fast? Your instincts might lead you to send curl commands, hand-roll an asyncio script, or vibe code yet another one-off load generator. All of these paths have the same problem: single-process performance limits, Python’s GIL capping concurrency, or numbers measured against a…

Source

Algorithms & Theory

Video Friday is your weekly selection of awesome robotics videos, collected by your friends at IEEE Spectrum robotics. We also post a weekly calendar of upcoming robotics events for the next few months. Please send us your events for inclusion.

Humanoids Summit Seoul: 22–23 September 2026, SEOUL

IROS 2026: 27 September–1 October 2026, PITTSBURGH

CoRL 2026: 9–12 November 2026, AUSTIN

Enjoy today’s videos!

A quadrotor is really just two birotors that love each other very much.

[ CLIMB Lab ]

The last 30 seconds of this video are fantastic.

[ ETH Zurich Robotic Systems Lab ]

This is great! But if you really want to impress me, convince me that it’s a) safe to be in my home and 2) a realistic alternative to a human.

[ Figure ]

I doubt anyone would call Digit 5 pretty, but here is what serious humanoid safety looks like.

[ Agility ]

NYU Tandon researchers built an undulatory robot that pumps water between its head and tail to climb slopes, clear steps, and swim, all while keeping the same basic gait.

On an incline, shifting water toward the head increases the force pressing the robot’s front contact points against the surface, giving them more traction. Head-biased placement was the only configuration that successfully carried WorMa up the steepest incline tested, 19.5 degrees; every other configuration slid back down. It also reduced the robot’s cost of transport by at least 33 percent compared with the other configurations.
Steps required something different. Neither a fixed head-heavy nor tail-heavy robot could clear a step on its own. The researchers instead had WorMa approach the step with water in its head for traction, then shifted the water to the tail and raised the head and neck so the tail could push the robot closer. Once the head anchored on the step’s edge, the water shifted back to the head, and the robot continued. That sequence got WorMa over steps as high as 15 centimeters.

[ NYU Tandon School of Engineering ]

Kubi 2.0 is a desktop telepresence AI robot that allows remote users to look around and interact as if they were physically present. It is also a Physical AI device enabling AI agents to interact with people in the physical environment. For over the past 10 years in Japan, we have successfully implemented our telepresence solutions to help hospitalized and homebound students attend classes, interact with friends, and participate in school events. Now, we are expanding this initiative globally through Kickstarter.

[ Kubi 2.0 Telepresence Kickstarter ]

I am oddly impressed by this use case for a wheeled quadruped.

[ DEEP Robotics ]

Legged robots are versatile on land, but their use in underwater environments remains limited. Extending quadruped locomotion to water enables amphibious mobility with applications in inspection, environmental monitoring, and disaster response. This paper presents the design, modeling, and experimental validation of a reproducible underwater quadruped robot. The robot is built around custom waterproof motor housings machined from polyoxymethylene plastic, which use off-the-shelf O-rings and dynamic shaft seals. A simplified model is derived to describe the dynamics of this underwater legged system, capturing how drag forces on spherical end effectors transmit torque to the floating base. Building on this model, a closed-loop attitude controller is developed using an error formulation defined on the special orthogonal group SO(3). The controller is evaluated both in simulation and experimentally in a water tank, where the robot tracks desired orientation setpoints in roll, pitch, and yaw.

[ Norwegian University of Science & Technology Autonomous Robots Lab ]

Thanks, Kostas!

Ugo Nova, a domestically produced [in Japan] semi-humanoid robot, accelerates the social implementation of physical AI, from research and development to operation in industrial settings.

[ Ugo ]

I appreciate how much manipulation you can actually do with a good two-finger gripper and a clever robot.

[ Daimon Robotics ]

One humanoid robot rolls off the line every 10 minutes. UBTECH’s benchmark humanoid robot smart factory just goes live—from 1,000 to 10,000 units, we made the leap in less than 9 months.

This is great! But what are they all going to do?

[ UBTECH ]

MIT engineers have built a robotic optics lab that builds, runs, and dismantles an optics experiment on demand. The setup centers on a robotic arm that physically arranges mirrors, lenses, and other optical components that are each set in custom-designed, QR-encoded housings. The robot arranges each component in precise alignments to direct and focus a beam of light (red laser). The autonomous system could speed up testing of high-tech materials, from solar cells and sensors to video displays and quantum technologies. This video is sped up 2x.

[ MIT ]

Student-built rovers. Lunar terrain. High-stakes competition at NASA’s Kennedy Space Center. See the energy, engineering, and excitement of Lunabotics as university teams put their robotic systems to the test in a simulated lunar environment. These are highlights from the previous competition, as well as a look ahead as teams get ready for Lunabotics 2027.

[ Lunabotics ]

This CMU Robotics Institute Seminar is by Erol Şahin from METU, on “Augmenting Bee Colonies with Robotics and AI Technologies for Ecosystem Support.”

Earth’s ecosystems are facing a rapid decline in biodiversity, with honeybees—keystone pollinators critical to ecosystem stability—being among the most affected. The EU-funded RoboRoyale project addresses this crisis by integrating advanced robotics and AI to augment the beehive, enabling observation at unprecedented resolutions and scales. Featured on the cover of Science Robotics and receiving the 6th Edge of Government Award at the World Government Summit in 2024, our system tracks the Queen’s behaviors, colony efficiency, comb states, and long-term foraging activities, while advancing microrobotic intervention capabilities to support hive health. In this talk, I will discuss the challenges of developing this system, share key findings regarding complex social interactions, and explore the future potential of bio-hybrid research.

[ Carnegie Mellon University Robotics Institute ]

Welcome to another edition of Road to KubeCon, where we’re tracking the Kubernetes and cloud-native ecosystem on the way into KubeCon + Cloud Native Con NA 2026, to be held in Salt Lake City, Utah, November 9-12.

This week, we look back at the past week of significant movements in the Kubernetes space. Most notably, we see interesting advances in cloud-native architectures for AI inference. We take a look at that, plus a new Gartner quadrant, new Kubernetes hardening updates, and important CNCF project updates.

HPE challenges server virtualization platforms

On Monday, Gartner published its Magic Quadrant for Server Virtualization Platforms, a guide comparing solution providers in the server virtualization market. The quadrant names HPE as a Challenger based on Ability to Execute and Completeness of Vision.

Hewlett Packard Enterprise (HPE) is a presenting sponsor of Road to KubeCon. HPE Software helps IT organizations modernize infrastructure, streamline operations, and accelerate AI initiatives across hybrid, multi-vendor environments.

According to the HPE newsroom, the recognition reflects ongoing momentum behind HPE Morpheus Software, its virtualization and cloud operations portfolio. HPE was positioned in the Challengers quadrant alongside Canonical and Oracle.

The announcement comes as enterprises rethink their virtualization strategies. Rather than simply swapping in another hypervisor, HPE argues that organizations increasingly need unified governance and ways to provision, orchestrate, observe and secure workloads — including VMs, containers and AI workloads — across clouds.

Kubernetes hardens container storage

On Wednesday, Red Hat’s Nispriha Jagan and Neeraj Krishna wrote on the Kubernetes project blog about two new storage security features shipped as Alpha in Kubernetes v1.37, which included 67 enhancements.

The notable security features are new bind mount options and emptyDir permissions. The enhancements come as multiple security findings have surfaced regarding emptyDir volumes, one of the most common writable volume types. The additions are made possible by low-level Linux security mechanisms.

According to the authors, these enhancements give users native controls to harden Kubernetes workload storage better. “Supporting noexec, nodev, and nosuid gives users a native way to harden volume mounts to match security benchmarks and policy,” the authors write.

KubeCon adds AI Inference + Agentic track

Last month, CNCF announced it will feature an AI Inference + Agentic track at KubeCon + CloudNativeCon North America 2026, exploring the intersection of generative AI and cloud native infrastructure. Attendees can explore the sessions here.

The added track underscores the growing use of Kubernetes for production AI workloads, particularly as the focus shifts from training models to serving them in production. It also reflects emerging practices for building agentic systems around protocols like MCP and A2A, as well as infrastructure such as AI gateways.

China Merchants Bank unifies AI inference on Kubernetes

China Merchants Bank, a leading Chinese commercial bank, recently showcased its cloud-native AI infrastructure at a CNCF event in China. Its infrastructure team won the CNCF End User Case Study Contest with an architecture combining Kubernetes with several cloud native projects:

• Kueue, for job queueing and quotas,
• KEDA, for event-based auto-scaling,
• Prometheus, for systems monitoring and metrics,
• HAMi, for sharing accelerator capacity across Kubernetes workloads,
• and Fluid, for accelerating access to datasets.

The bank has a large pool of nearly 10,000 accelerator cards used for AI computation. These are heterogeneous, meaning they are not all the same type or configuration.

According to the CNCF announcement, the architecture unified management of 99% of its AI compute resources, while increasing average utilization from 35% to more than 60%. It also cut the cost of processing 1 million tokens by 60% under comparable conditions.

The case study shows how cloud-native infrastructure can improve utilization and efficiency for AI training and inference, even in regulated areas like financial services.

Industry take: Can AI inference on Kubernetes handle token cost issues?

Interest in AI inference on cloud native infrastructure is palpable. However, this week Val Bercovici, chief AI officer at WEKA, an AI-native data platform, questions whether Kubernetes’ existing resource model fits the changing economics of large-scale AI inference.

Bercovici tells The New Stack: “With AI inference, it’s cost per token, and that cost depends on state Kubernetes was never designed to manage: request mix, KV cache occupancy, the balance of prefill and decode, and how memory and bandwidth are consumed inside the accelerator after a pod is already running.”

“My view is that Kubernetes doesn’t go away,” Bercovici says. “But unless its resource model evolves, it becomes a tax on inference economics.” 

He foresees a new scheduling and memory layer to emerge around Kubernetes that can compute what a token actually costs to serve. Then platforms could make more informed, cost-based decisions about how inference workloads are scheduled and served.

As Kubernetes evolves, so do the demands on the teams running it. Presenting sponsor HPE helps teams address that complexity with software spanning virtualization, cloud management, observability, and automation.

Move over, platform engineering. Hey, agentic engineering.

A new Weave Intelligence report, State of AI in Platform Engineering Volume 2, authored by Sam Barlien, Luca Galante, and Florian Lipp, surveyed 242 platform engineering leaders on the before-and-after effects of introducing agentic AI into platform engineering.

38% of teams are shipping at least twice as much as before AI. When assessing ROI across the software delivery life cycle, 20% report efficiency gains and 11% report operational savings. Yet only 8% report a transformative, structural shift. Meanwhile, 29% are still prototyping without realized gains, with some outliers reporting negative results.

The biggest roadblock to scaling AI usage? A lack of platform readiness, including APIs, deterministic pathways, and standardization. Weave’s takeaway is that platform engineering must increasingly account for AI readiness and agentic experience as agents become another key platform consumer.

OpenTelemetry Kubernetes attributes processor reaches v1.0.0

On Wednesday, OpenTelemetry, the graduated CNCF project and open standard for telemetry, announced the v1.0.0 release and distribution of its Kubernetes attributes processor. It’s a helpful feature that uses the Kubernetes API to add Kubernetes metadata, such as stability, distributions, warnings, issues, and other metrics, to resource attributes.

According to the release notes, written by Elastic’s Christos Markou and Datadog’s Pablo Baeyens, the feature has been in progress in the OpenTelemetry Collector SIG since late 2025, based on a roadmap of users’ most-requested features. Existing attribute processors should review the breaking changes and migration guide.

DigitalOcean opens Spot GPU node pools

Technically, this occurred the week before last, but didn’t make the digest. As of September 9, DigitalOcean Kubernetes’ (DOKS) Spot GPU Node Pools entered public preview. According to the release notes, the feature runs worker nodes on interruptible GPU capacity at a “lower, variable rate than on-demand GPU nodes.” This could offer a cost-effective option for fault-tolerant workloads.

Other updates from the K8s universe

More updates from the infrastructure-heads, platform engineers, and multi-cloud operators working in the Kubernetes ecosystem: 

• KubeCon registration applications are open until October 4, 11:59 p.m. MDT. Travel-scholarship applications have closed.
• Cilium 1.20.2 release includes several bug fixes and CI changes.
• The latest Talos Linux 1.14 update benefits multi-tenant services. “Welcome departure from LUKS-only,” comments software engineer Charles Deane on LinkedIn.
• The Apache Flink community announces Flink Kubernetes Operator 1.16.0, centered on “extensibility and documentation.”
• Arm introduces Neoverse CSS N4 alongside its AGI CPU to improve throughput efficiency and support the variable compute demands of agentic AI.

Follow the Road to KubeCon

Road to KubeCon is an eight-part series presented by HPE, which will be at KubeCon + CloudNativeCon North America in Salt Lake City. Before you go, explore how HPE Software helps IT teams do more with less complexity.

We’ll be here every Friday until KubeCon.

If you’d like to participate, Bill Doerrfeld, the writer of this series, is open to pitches — you can send release notes, quotes, reports, videos, case studies, or hot takes through his contact page.

If you didn’t catch the inaugural edition covering the Kubernetes v1.37 release, check it out here. You can also visit the Road to KubeCon page for the complete archive.

The post Kubernetes can run AI inference. But can it count the real cost? appeared first on The New Stack.

Nature Machine Intelligence, Published online: 18 September 2026; doi:10.1038/s42256-026-01312-x

Living organisms continuously monitor their internal physiological states while interacting with complex environments and adapt their actions in response to changes in these internal signals. New work combines such biological principles with cybernetics, reinforcement learning and neuroscience to develop a framework for autonomous and adaptive artificial embodied agents.

How to verify your app aligns with your intent without ever reading a line of generated code.

The post Coding Agents Keep Shipping Silent Failures — Here Is How to Catch Them appeared first on Towards Data Science.

The reliably Republican South Dakota senate seat has an incumbent on the ballot. But PACs associated with AI labs and investors have already spent more money on the race than actual residents have.

Anthropic’s CEO says that safety hinges on understanding how AI “thinks.” So far the evidence is disturbing.

The Trump administration says it wants AI models free from ideological bias, as it pressures their developers to reflect the president’s worldview.

The recurring cost of production AI is not inference. It is re-qualification: the eval reruns, prompt retuning, and regression testing you owe every time a model changes under you. Here is what that tax actually covers, and how to budget for it before it surprises you.

The post We Pinned Our Model Version to Stay Safe. The Provider Deprecated It Anyway. appeared first on Towards Data Science.

The satirical collaboration between a creative studio and the music producer behind Big Data, “Bezzy” is a cute little doll that spews the sounds of real data centers.

You built a great agent, but something happened when it moved into production.

In testing, your agent reviewed pull requests efficiently on its own. It read the diff, grepped the codebase for related usages, ran the test suite, checked whether CI was still red from an earlier commit, and drafted a comment—all before you’d finished reading the diff yourself.

In production, however, imagine the same five steps ran behind every other PR review your team’s agents performed that hour. Some reviews landed in seconds; others sat for minutes because the test-suite step landed on a node that was mid-burst from someone else’s agent. 

The agent didn’t change. The execution environment did, and that’s what decided whether review time held steady or crept up.

Agentic applications introduce a different execution pattern than traditional chat applications. As those workflows become longer and more dynamic, infrastructure has a much larger influence on latency, reliability, and cost than it does for a simple chatbot.

That’s why your agent is only as good as your infrastructure.

Agents aren’t chatbots with more steps

The difference between serving inference for a chatbot vs. an AI agent isn’t simply that one is “more capable.” They execute work differently:

• A chatbot usually makes one inference call per user message. The model receives a prompt, generates a response, and waits for the next user input before proceeding.
• An agent executes the entire workflow, turning one user message into a chain of inference calls. It might decide to search documentation, retrieve data from a database, call an API, execute code, evaluate the result, and then repeat that process before producing an answer. Each of those decisions may trigger another inference call, and every result becomes additional context for the next step.

“Infrastructure has a much larger influence on latency, reliability, and cost than it does for a simple chatbot.”

That execution model changes the infrastructure requirements for AI agents. 

One question, many steps behind it

Instead of optimizing for individual inference requests, the system has to support long-running workflows whose latency and reliability depend on every component in the chain.

A single user request often expands into a sequence of inference and tool execution steps, sometimes called multi-turn tool calls or the agentic loop. Rather than generating one response, the model alternates between reasoning and interacting with external systems.

For example, you ask an agent why checkout latency spiked overnight. The agent pulls the deploy log, queries the monitoring system, runs a diagnostic against the connection pool, weighs whether the culprit is a bad deploy or a capacity issue, and then folds that into another inference call before finally producing a full-fledged response.

Each reasoning step becomes another inference request, and every tool result is added to the model’s context before the next step.

This workflow changes what reliability means

Multi-turn workflows are inherently sequential, which is why even low latency can quickly add up to a significant amount. Every inference step waits for the previous one to finish. If a database query takes two seconds, the model can’t begin the next reasoning step until that result returns. The model may generate tokens quickly, but the other steps slow it down.

“In this agentic workflow, every step in the chain has to hold, because the chain is only as strong as its slowest link.”

In this agentic workflow, every step in the chain has to hold, because the chain is only as strong as its slowest link. Instead of processing isolated inference requests, the inference stack has to orchestrate a chain of dependent model invocations and external tool calls. As those workflows become longer, the stack increasingly determines how quickly, reliably, and cost-effectively the application performs.

But the user doesn’t see an orchestration hiccup. They see an agent that hung or gave up.

That’s why end-to-end agent reliability depends on much more than model quality. Infrastructure determines whether each step has the resources it needs to execute predictably under load.

Why the bill and the performance both feel unpredictable

A second difference in agentic workflows catches teams off guard: demand patterns and their impact on your inference bill. 

Most inference services, and the pricing built on top of them, assume traffic arrives at a predictable pace. A typical inference solution knows the predictable demand pattern: User traffic increases, request volume increases, and capacity scales accordingly. Cloud infrastructure is typically optimized for these steady request patterns using mechanisms such as autoscaling, load balancing, and capacity planning.

Agent workloads don’t behave that way. Individual workflows pause while waiting on external systems, then resume as soon as new information becomes available.

• The pause: The agent waits on an external API or database, so the GPU serving that workflow has no inference work to perform, and its accumulated context may be evicted from GPU memory while it waits
• The burst: As soon as external systems return results, inference resumes simultaneously across many workflows, creating short and sharp spikes in GPU demand, each re-processing its full accumulated context

If you’re watching GPU utilization and it looks less like steady load and more like a heartbeat — flat, then a spike every time tool results come back — that’s the signature. It means you’re provisioning for the average when you should be provisioning for the peak, and it’s usually the first place p99 latency quietly blows out.

“If you’re watching GPU utilization and it looks less like steady load and more like a heartbeat, that’s the signature.”

Inference services designed around steady or predictable request streams can struggle to allocate resources efficiently under these conditions. This leads to inconsistent latency, GPU underutilization, or higher operating costs.

When performance becomes unpredictable, or a bill doesn’t match what you expected, it’s evidence your infrastructure was built for a different workload than the one you’re actually running.

What infrastructure built for agents actually looks like

Agentic applications place different demands on infrastructure than traditional AI workloads: long dependency chains, bursty demand, and continuous evolution. Because of this, the infrastructure is deciding whether the chain holds, and whether the bill holds too.

For an agent to run, it needs infrastructure purpose-built to support:

Performance that holds across the whole chain. The infrastructure must keep latency consistent across multi-step and multi-tool workflows.

Scalability that responds to bursty demand. Infrastructure should scale quickly as inference demand fluctuates, without requiring capacity to remain provisioned during idle periods.

• Predictable economics even for dynamic workloads. The infrastructure bill should reflect actual usage.
• None of this makes bursty demand disappear, but it changes how the system absorbs it. A large enough simultaneous burst, or a workflow that accumulates enough context before pausing, still costs something. The goal isn’t zero cost or zero limit; it’s making both predictable.

Your agent is only as good as your infrastructure. Get it right, and your agent’s responsiveness, reliability, and cost-effectiveness will improve your work. 

Learn more about how infrastructure can be purpose-built for agentic workflows: Check out the documentation to get started.

The post Your agent is only as good as your infrastructure appeared first on The New Stack.

The trend is clear: open-weight models are taking an increasingly large bite out of production AI usage.

On Monday, The New Stack reported that open-weight models accounted for 60% of OpenRouter’s US token consumption in August, with Chinese-developed models making up the majority of that volume. The latest data point hails from Vercel, whose AI Gateway routes tens of trillions of tokens each month across the applications running on its infrastructure.

As per Vercel’s September report, published on Thursday and covering activity through August, open-weight models handled 56% of all tokens routed through the gateway, the first time they have accounted for a majority of monthly token volume. In December 2025, their share was just 7%; by April it had reached 13%, and it rose every month thereafter. Vercel’s previous report, published in August, put July’s open-weight share at 36%.

So the pattern was already clear. But last month, Vercel CEO Guillermo Rauch took to social media to declare that August 22 had been a “record day for open weight share of tokens on Vercel AI Gateway,” accounting for 62% of traffic.

Rauch saw the milestone as just an early indication of where usage is heading, with enterprises still early on the adoption front.

“This is very likely just the start, because enterprise adoption is still early.”

“This is very likely just the start, because enterprise adoption is still early, and harnesses, CLIs, IDEs, SDKs, etc need to be adapted to be model agnostic,” Rauch wrote at the time.

Tokens and dollars: Anthropic dominates spend

For context, Vercel launched AI Gateway last year as a way for developers to access models from multiple providers through a single interface, saving them from having to manage separate API keys, accounts and rate limits. The service sits between applications and the underlying model providers, routing requests while tracking usage and costs — giving Vercel a useful vantage point into which models its customers are actually running in production.

Token volume, in this context, is essentially a measure of how much model inference is flowing through the gateway. Vercel counts input and output tokens, along with reasoning, cached-input and cache-creation tokens.

While it’s a good proxy for the amount of work being handed to different models, it shouldn’t be confused with the amount of dollars being spent. Open-weight models from the likes of DeepSeek, Moonshot AI and Z.ai are generally cheaper to run than the proprietary models offered by US frontier labs — and so handling 56% of Vercel’s token volume doesn’t mean open-weight models are taking 56% of the money passing through its gateway.

Indeed, Vercel’s data shows that open-weight models accounted for just 14 cents of every estimated dollar spent through AI Gateway in August, despite processing 56% of its tokens. Their share of spending remains far behind their share of usage, although Vercel says the open-weight share of gateway spending is on the rise.

Across Vercel’s AI Gateway, the average price per token fell 23.2% in August, marking a third consecutive monthly decline. Among teams that processed more than 10 million tokens in both July and August, the median cost per token fell 7.6%.

Anthropic, meanwhile, has remained remarkably consistent at the spendy end of the market. Its models accounted for 64 cents of every dollar spent through the gateway in August. Vercel says the Claude-creator’s share has never fallen below 61% in any month since December 2025, with its models occupying the top two positions by spend throughout that period — often taking third spot, too.

Loyalty lies in the model

There has been plenty of movement within that Anthropic share, however. Fable 5 fell from 13.2% of total gateway spend in July to 4.9% in August, while the cheaper Opus 5 climbed to 22.5%. More broadly, Vercel’s data suggests that 90% of teams using Fable reduced their usage, with more moving those workloads to Opus 5 than to any other model.

Opus ultimately gained almost twice as much usage as Fable lost, which Vercel attributes to the newer model handling similar workloads at roughly half the price. Or, in other words, Anthropic kept the dollars even as customers shifted toward a cheaper model within its own lineup.

“Lab loyalty doesn’t follow brand, it follows model profile, and consistency wins.”

“Lab loyalty doesn’t follow brand, it follows model profile, and consistency wins,” Vercel’s report authors note.

This trend was evidenced elsewhere, too. Within five days of Z.ai launching GLM-5.3-Flash, the new model was processing three times the daily volume of GLM-5.2.

But Vercel’s data also suggests customers are more than prepared to cross lab boundaries when a replacement fails to meet the same needs on capability and price: more than three-quarters of the volume lost by Google’s Gemini 3 Flash moved to models from other providers, including OpenAI and Anthropic. And the consequence for Google wasn’t insignificant: its overall share of token volume on the gateway fell from 30% to 5%, with the decline in Gemini 3 Flash alone accounting for 22 of those 25 percentage points.

“When a new model preserves what users valued in its predecessor, the lab retains its customers,” the authors note. “When it doesn’t, those customers fill the need through other providers.”

The post Open-weight models now handle a majority of tokens on Vercel’s AI Gateway. But Anthropic still takes 64% of the spend. appeared first on The New Stack.

How do you set yourself up for a career that will stand the test of time when things are changing so fast?

The post Starting a Career in Data Science in the Age of AI appeared first on Towards Data Science.

On Wednesday, MIT Technology Review hosted a live Roundtables event for subscribers that asked the question everyone’s asking right now: Could AI really kill us all?

But attendees had so many more questions than we had time to answer in the 30 minute session. So we asked our senior AI editor Will Douglas Heaven and AI reporter Grace Huckins to round up some of the best questions attendees submitted and try their best to answer them.

Thanks to all who submitted questions!

Am I gonna die?

Yes, eventually. Unfortunately, my journalistic powers of prognostication aren’t powerful enough for me to tell you how. But it certainly could be because of AI. AI-powered drones have already killed people in Ukraine, and AI-driven cyberattacks on hospitals will surely claim victims before long. 

Could AI go even further, and kill all of us? Less likely. But some people—quirky people, but undeniably knowledgeable about AI—have been warning for years that this could happen. And while I’m not yet stockpiling canned food or trying to get in good with a bunker-owning megabillionaire, I have noticed that the doomers’ predictions about AI capabilities and alignment have, over the past couple of years, proved disconcertingly accurate. That certainly doesn’t mean that their more dire forecasts will come true, but it’s enough for me to sit up and take notice.

— Grace Huckins

Are you going to die because of AI? I’d say there’s a non-zero chance. Let’s say you’re unlucky enough to be the victim of a freakish near-future event or accident. Maybe it’s a cyberattack carried out by a swarm of AI agents on critical infrastructure. Sadly, a scenario like that now no longer feels as far-fetched as it once did. Or maybe a novel AI-designed pathogen cuts through the population. Or the world economy crashes, causing conflicts and famine. Both plausible, but I think less likely. 

Are we all going to die because of AI? Nope. There are no circumstances outside of apocalyptic science fiction in which AI could kill us all. You can spin up any number of scare stories, but they’re not grounded in present-day realities about what the tech can do or where it’s headed. 

Some people argue that there’s no harm in preparing for the worst, however wacky it might seem. Maybe. But I think such catastrophizing can make people excuse or overlook many of the more immediate problems with the existing technology and the companies building it. 

— Will Douglas Heaven

Why would AI kill us?

Someone might tell it to, and it might listen. That’s part of the reason researchers are so concerned about AI’s biological capabilities—imagine what Aum Shinrikyo, the doomsday cult behind the Tokyo subway sarin attack of 1995, would have done with a tool that could design a pathogen deadlier than Ebola and more transmissible than measles. Those of us who don’t want to die have to figure out how to defend against all plausible biological weapons, but our would-be attackers only have to manufacture one effective pathogen.

Then there’s the more exotic-sounding possibility that an AI could decide to kill us itself. There are various stories about how this might happen out there, but the most widespread involve AI systems that don’t hate people, necessarily—we are just an obstacle between them and the goals that we gave them.

Much as the OpenAI agents behind the Hugging Face hack compromised another site’s infrastructure to get a good score on a test, the idea is that some future, more powerful AI might get rid of us to prevent us from shutting it down—all in pursuit of some goal that we instructed it to go after. 

— Grace Huckins

How can we best ensure alignment so the worst doesn’t happen, and who is doing the best work to achieve it? 

Alignment is a huge area of research. In simple terms, it involves building models that behave in ways we want them to and not in ways we don’t. We need to trust agents better before handing over more autonomy. Alignment is supposed to establish that trust. But it’s hard. 

LLMs aren’t designed in the way other software is, where dos and don’ts can be hard-coded in. Instead, aligned behavior needs to be instilled when models are trained. One approach is to reward them for doing things you want them to (a little like raising a toddler, perhaps). Another approach involves giving an LLM a written list of rules it is supposed to follow (kind of like a constitution). 

Anthropic and OpenAI are both leaders in this field—and yet neither has been able to develop models that are fully aligned. A big problem is that LLMs are far more inconsistent and far less predictable than people. They can behave in one way in one situation and another way in a situation that to us seems very similar. They can also be swayed by unexpected constraints. For example, faced with an impossible task (as many of the agents involved in the Hugging Face hack were), models may try to do whatever it takes to achieve their goal. As Grace mentions above, that could be an issue.

The main reason top AI firms now say they want a slowdown is that they want to focus on cracking alignment. Alignment isn’t necessarily a pipe dream. But the jury’s out on whether full alignment will ever be feasible. 

— Will Douglas Heaven

Is AI really dangerous, or is this the tech companies drumming up PR?

This is always a reasonable thought when it comes to tech companies heading for an IPO—CEOs have an obvious incentive to make their products seem radical and transformative. But I’m not so sure it makes sense here. Telling the public that an already unpopular product could kill them and everyone they love is horrible corporate image management.

There are other stories you can tell about the CEOs’ motivations—maybe they want to cool down the public furor over data centers by portraying themselves as responsible stewards of a world-changing technology, or maybe they want to buy time to get their ducks in a row and prevent the next PR catastrophe. 

But there’s also a simpler explanation. Thinking that AI could bring about human extinction has been pretty common in San Francisco for a while, and these men are steeped in that milieu—as are their employees, many of whom signed an open letter in July urging their companies to work to make an AI slowdown possible.

— Grace Huckins

Part of the concern occurs when AI agents are allowed to act autonomously and with no supervision. What’s the issue preventing more control over these agents? 

This question goes to the heart of what we want this technology to be able to do. The trade-off between autonomy and control is tricky to get right because, on the one hand, a lot of the power of AI agents is that they can carry out tasks and solve problems without a human having to micromanage them. On the other hand, that requires you to trust that the unsupervised agents won’t run amok. 

What we’re seeing is that AI labs haven’t yet got this trade-off quite right. Their models are not trustworthy, they are not properly monitored, and they are not always under control. Figuring out how to fix that while still allowing for useful autonomous activity is one of the big research challenges of the moment.  

— Will Douglas Heaven

What steps can be taken now and in the near future to ensure that AI is controlled, monitored, and regulated effectively? 

That’s the million-dollar question. Whether or not you think AI could kill us, you can’t deny that it could do some real damage, because it already has—by driving people toward psychosis and by hacking websites, for example. Preventing that damage, or at least mitigating it, is hard for two reasons. 

The first is that we barely understand how AI works, and it’s quickly growing more powerful. There is lots of ongoing research about how to monitor and control misbehaving agents, but the current approaches are fragile. You can see if an agent discusses misbehaving in its “chain of thought,” the workspace where it plans its actions—but OpenAI’s newest agents don’t show their work in the same way as previous ones. And you can try to monitor agents with other agents, but that requires you to trust the monitor.

The other obstacle is more familiar. There’s a huge conflict of interest when AI companies regulate themselves, but the US government has thus far failed to step in, despite some bipartisan support in Congress for efforts to do so. The executive branch, for its part, seems stringently opposed for the time being. But if the winds do shift, I for one would appreciate some strong transparency regulations, so that we can get a fuller story the next time an unreleased frontier model mounts a cyberattack.

— Grace Huckins

If this dialogue makes it into web discourse, will it become a self-fulfilling prediction?  

That’s a real concern. LLMs are influenced by what they read. One theory for why chatbots so often talk about (and role-play) apocalyptic scenarios is that they have been trained on millions of pages of science fiction stories and doomer internet forums. All the text being produced right now, including this article, could in turn influence the behavior of future models. Extremely meta.

In fact, the team at METR, a third-party organization that OpenAI called in to help understand what happened in the lead-up to the Hugging Face hack, raised a related possibility in its report on the incident. METR used OpenAI’s new model Astra to help analyze the vast numbers of agent transcripts and behavior logs.

But feeding all that material to the model could have unintended consequences. There’s a good chance that the agents doing the analyzing were biased by the text produced by the agents they were analyzing. There’s no such thing as a clean slate anymore. 

— Will Douglas Heaven

With thanks to Eric, Pranab, Rafael, Kenneth, George, Chris, Yoon Jae, James, Carl, Nicole (and more!) for the fantastic questions.

Multi-agent coding systems don't necessarily fail because agents can't communicate. They can fail because important commitments made in conversation have nowhere to live afterward.

The post Multi-Agent Coding Isn’t Enough — Agents Need a Commitment Layer appeared first on Towards Data Science.

The left wants AI regulation. They just can’t agree on what it should look like or how concerned they should be.