Krotite is a cosmic relic, one of the oldest minerals in the Solar System.
© The New York Times
Google has released new tools that allow developers to run agentic AI workflows locally using Gemma 4 12B, a 12-billion-parameter model from Google DeepMind.
In a blog post, the company said the model, combined with the Google AI Edge stack, can be used to build and test applications on everyday machines. The model-runtime combination supports capabilities such as autonomous data processing, visual insight generation, webpage creation, and tool use.
The release includes Google AI Edge Gallery for macOS, where developers can use Gemma 4 12B to generate and run scripts for tasks such as data analysis. Google also said its Eloquent voice dictation and editing app now runs fully on-device on macOS, with support for local transcription and voice-driven text editing.
Google has also expanded LiteRT-LM, its lightweight command-line tool for running language models locally, with a new serve command. The company said this allows the CLI to act as a local LLM server and lets developers connect Gemma 4 12B to standard tools, SDKs, and frameworks through a local endpoint.
“Your data stays on your device while maintaining reliable responsiveness, utility, and cost efficiency,” the company said in the blog post.
The announcement comes as enterprises are looking beyond large, general-purpose models for some AI workloads. Gartner predicted that by 2027, organizations will use small, task-specific AI models at least three times more than general-purpose large language models, citing demand for more contextualized and cost-effective AI systems.
But running agents on employee devices brings a number of problems. Companies must work within the limits of endpoint hardware, which can restrict the size of models that run effectively and the number of model instances that can operate at one time.
“While the AI can now fit on a laptop, enterprise IT infrastructure is largely unprepared to manage it,” said Rishi Padhi, principal analyst at Gartner. “Even highly optimized models like the Gemma 4 12B require around 16GB of unified memory or VRAM to run alongside standard applications. Many standard-issue enterprise laptops lack the memory bandwidth and NPUs/GPUs required for fluid, multi-turn agentic execution.”
Anand Joshi, AI analyst at TechInsights, said local deployment also changes the nature of the workloads. On a PC, search may mean finding information across internal folders and files. In a data center, the same function could involve searching the internet or querying a large database such as SQL.
“The framework for local deployment of agentic AI is different from that of a data center,” Joshi said. “The models are smaller; you can run only one instance of a large model at a time. You are limited by memory, CPU, and so on.”
Security and governance are also likely to become bigger concerns as AI agents move closer to enterprise endpoints. Agentic AI is designed to take actions, creating new security risks when local models are given access to employee files or allowed to interact directly with applications and scripts.
“Sandboxing these agents without breaking their utility is still a major operational challenge,” Padhi added. “And all this while enterprises need to audit AI usage for compliance and security. When inference happens entirely offline, capturing logs, tracking model drift, and ensuring employees are using the approved, compliant ways for a model becomes incredibly difficult.”
Running AI agents locally could reduce some cloud inference costs, but the savings may be offset in the near term by higher spending on endpoint hardware and management.
“First and foremost, it is an OpEx-to-CapEx shift, as it shifts that financial burden by forcing accelerated hardware refresh cycles for premium PCs or edge devices,” Padhi said. “It would require buying expensive, high-memory laptops for employees at a time when memflation in the hardware industry is already driving up end-user average selling prices for laptops.”
Many enterprises refreshed PCs in 2025 to support Windows 11, but at that point, most AI inference still ran in the cloud, and the case for on-device AI remained unclear, Padhi said.
Enterprises may therefore move cautiously, buying AI-capable PCs only where local inference has a clear business case.
Over time, however, on-device AI could make enterprise AI spending more predictable by reducing exposure to variable cloud inference bills. The tradeoff is that companies may face a higher baseline cost for equipping and managing employees’ devices.
For enterprises, local AI is unlikely to replace cloud-based AI outright. Analysts said local AI is more likely to be used for workloads that benefit from endpoint processing, especially when applications must operate offline or when privacy and response times are critical.
“For local agentic AI to proliferate, the use cases on edge will have to complement data center/cloud use cases,” Joshi said. “I don’t expect local agentic AI to replace cloud AI, but it has potential to take a slice away from the cloud, and models like Gemma are significant steps towards enabling that.”
The market, Joshi added, is still determining where local AI fits best. “I estimate that use cases that require privacy or have strict latency needs will move to local node first, with further migration of others in the next 2-3 years,” he said.
Padhi said model placement will depend on the privacy requirements of a workload, the computing power it needs, and where the relevant data resides. Tasks such as code generation or analysis of local files could increasingly run on employee devices, while enterprise-wide RAG systems and more complex AI workflows are likely to remain cloud-based.
The article originally appeared on InfoWorld.
A rescued sea lion is shaking up what scientists thought they knew about rhythm and the brainA nanoscale fabric coating repels stains and microbes using only tap water.
Built to track enemy submarines, the Navy’s underwater listening network inadvertently revealed that whales may be singing across entire oceans.A new evolutionary analysis suggests the behavior appears across many bird groupsThe New York Times described the stunning behavior over 100 years ago.A barn owl’s ability to hunt by hearing alone relies on exquisite variations in the structure of its feathers.Apple’s MacBook Neo appears to be a triumph of strategic disruption that has already cast shock waves across the industry — and that energy is still playing out.
Approximately 55,000 MacBook Neo computers have been sold every day since it was introduced in March, according to IDC data (as first noted by TechCrunch). In fact, it looks as if Apple sold 1.1 million of these Macs in the first 20 days of sale, the analysts said.
There’s no real reason to imagine that level of demand has declined very much.
After all, not only do these Macs continue to dominate Amazon’s US laptop charts, but supply chain rumors claim Apple has doubled its manufacturing orders. “MacBook Neo shipments have come in better than expected, with the 2026 shipment forecast raised from 5 million to 10 million units,” Apple analyst Ming-Chi Kuo said recently.
IDC’s March data may not capture the larger extent of the demand, as IDC analyst Navkendar Singh pointed out that MacBook Neo shipments “began to spike from early April”, which suggests demand has accelerated since then.
MacBook Neo demand exceeded expectations across multiple nations, including in India, where the company shifted 18,000 of them in the opening weeks.
Apple has also instructed processor maker TSMC to manufacture additional A18 processors specifically for its affordable laptop, while earlier speculation has claimed the company has been using ongoing memory price increases as a strategic competitive tool. (The Neo starts at $599, with a pricier model set at $699.)
By expanding the potential customer base for Macs with a lower cost Neo, Apple is aiming a claim at the biggest-selling part of the PC market. And it is doing so even as rapidly increasing component prices force others to choose between higher product prices and profitability, or much-reduced margins in to compete at the same price. levels
That’s a losing battle; competitors for the most part can’t hope to match Apple’s bargaining position when it comes to the cost of components like memory because they don’t have the same scale. That means that even when component costs increase for everybody, Apple pays less, because it orders more.
That scale means that for many component suppliers, it’s Apple’s business that keeps the meat on the table while other customers merely contribute the gravy. So, suppliers are happy to make deals with Apple to secure that main course — to continue the analogy — but are less likely to match those deals for dessert. As such, Apple is expected to be the only laptop vendor to see growth this year.
IDC’s figures confirm Apple’s strategy is working, with strong demand for the Neo, and, indeed, all Apple’s new laptops. At the same time, the researcher predicts overall global PC shipments will decline 11.3% this year, with a painful 20% sales drop envisioned for Q4.
“We’re not seeing any relief to the memory shortage situation before the end of 2027, which means prices will continue to rise and PC manufacturers will struggle to maintain full product portfolios for the foreseeable future,” Jean Philippe Bouchard, vice president of devices and consumers at IDC said in a statement.
“The introduction of the MacBook Neo is putting real pressure on the entire PC ecosystem,” added Jitesh Ubrani, research manager for IDC’s Consumer Devices Trackers.
Competitors are already responding with new devices equipped with ARM-based processors and aggressive promotional pricing. But none truly match what Apple has with MacBook Neo, and all must reach profitable scale to compete long-term.
None have yet done so.
“The MacBook Neo launch stands out as one of Apple’s most strategically important recent Mac releases,” Counterpoint analyst David Naranjo said.
Apple is directly targeting customers that previously saw its products as too expensive. That allows it is also to aggressively build business in parts of the market such as education that tend to be more resilient to economic headwinds. MacBook Neo is also enjoying strong demand across the enterprise.
Both these parts of the market give Apple’s competitors their lunch. “The competitive pressure from the Neo is providing a partial offset to broader price increases, keeping some low-cost notebook options alive,” Ubrani said. “But the overall trajectory for average selling prices (ASPs) is firmly upward. IDC forecasts ASP growth of 17% in 2026, and even as memory capacity expands over the next two years, pricing is unlikely to return to 2025 levels.”
Apple’s control over its processors, along with its strategic approach to component purchasing, means it should be able to maintain its existing Mac price points for a while. “Apple’s vertical integration (own silicon, own OS) gives it more levers than competitors reliant on third-party chips and Microsoft licensing,” Hexnode CEO Apu Pavithran told me recently.
So, while PC makers either exit the market or raise prices in pursuit of profits, MacBook Neo will continue racing off the shelves, particularly to large enterprise and education customers.
The Neo is more than a lower-cost Apple notebook. It’s a hugely disruptive product that is already driving noteworthy change across the PC industry; it’s forcing competitors to make difficult choices between cost and price — even as they grapple with the existential challenges of memory shortages, component price hikes, and raw materials costs.
That’s not bad for a product that costs your local school just $499.
Just a reminder: the original $399 iPod cost only slightly less when it was first introduced, before subsequently disrupting the music industry.
You can follow me on social media! Join me on BlueSky, LinkedIn, Mastodon, and follow The Core.
One way archaeologists learn how ancient people, including Neanderthals, did things is to attempt to do those things themselves, a process called experimental archaeology. Normally, that involves making stone tools, butchering deer, or distilling birch tar. But in a new study, it meant doing very destructive things to teeth from one of the world’s most carefully protected animals.
That's because the archeologists suspected that Neanderthals once used rhino teeth as tools. By using the teeth to make stone tools, the researchers demonstrated that Neanderthals probably did the same thing, adding to what we know about the wide range of items in their toolkits.
Some Neanderthal archaeological sites in Europe and Asia seem to have many more rhinoceros teeth lying around than you’d expect. We know Neanderthals hunted a now-extinct species of rhinoceros in Europe and eastern Asia, but the people who had inhabited these sites looked like they had been collecting rhino teeth for some reason.
© By Bernard DUPONT from FRANCE - White Rhino Skull, CC BY-SA 2.0, https://commons.wikimedia.org/w/index.php?curid=55453022
Login