Agregátor RSS

Attackers repeatedly changed the files that AOMEI Backupper fetches from the vendor's own update servers, in windows of five to ten minutes at a time between May and July 2026, so that any copy of the Windows backup program that checked for an update during one of those windows would download and run a backdoor. Google Threat Intelligence Group, Mandiant, and ManageEngine's Threat AnalysisSwati Khandelwalhttp://www.blogger.com/profile/[email protected]

Cybersecurity researchers have disclosed details of an advanced persistent threat (APT) spear-phishing campaign that has targeted individuals affiliated with Taiwan research organizations to direct victims to an adversary-in-the-middle (AitM) phishing framework designed to harvest their Google account credentials and multi-factor authentication (MFA) codes. The activity is being tracked by

Cybersecurity researchers have disclosed details of an advanced persistent threat (APT) spear-phishing campaign that has targeted individuals affiliated with Taiwan research organizations to direct victims to an adversary-in-the-middle (AitM) phishing framework designed to harvest their Google account credentials and multi-factor authentication (MFA) codes. The activity is being tracked by Ravie Lakshmananhttp://www.blogger.com/profile/[email protected]

Investors are pouring more into AI data center infrastructure than they can hope to recoup by 2030, according to Gartner.

“Enterprise and consumer AI revenue will be insufficient to support profitable returns on all planned AI capital expenditures during this period,” said Gartner Senior Director Analyst Will Sommer.

AI data center owners would need to generate $8.5 trillion in revenue through 2030 to cover announced capital spending, financing costs and expected investment returns, he said. “However, total enterprise and consumer AI revenue that can be attributed to AI data centers during this period is forecast to reach no more than $3.6 trillion.”

Gartner joins a growing chorus warning that major AI players may be over-committing themselves. Bain and Co. recently made a similar forecast, estimating that annual spending on AI infrastructure could reach $1.5 trillion by 2031 and that the AI services market would need to top $6 trillion annually to pay for it.

This is particularly bad news for neocloud providers, said Gartner. While the hyperscalers are large enough to absorb the financial pressures, the research company suggests that the cost pressure will force more than 80% of neocloud providers to cease independent operations.​

The Bain and Gartner surveys don’t cover quite the same ground, but their conclusions are very similar: a lot of AI infrastructure investment is never going to make a return, and companies need to prepare themselves for some degree of financial loss.

This article first appeared on Network World.

The October 2026 Windows security update may break networking for some customers who are still clinging to long-deprecated NTLMv1 security. A seemingly minor change in the update could have major implications for administrators and enterprises that haven't fully moved away from the legacy technology. Before the October 2026 update, Windows permitted NTLMv1-derived credentials for Single Sign-On and generated warning events. After the update, requests to use NTLMv1-derived credentials are blocked. It comes down to the BlockNtlmSSO registry value. If it hasn't been set, Windows will now use Enforce mode (1) instead of Audit mode (0), thus blocking requests and potentially impacting Wi-Fi, Ethernet, and VPN authentication, and therefore connections. The update does not set the value of BlockNtlmSSO, and administrators can set it to 0 to temporarily allow NTLMv1-derived Single Sign-On credentials. However, Microsoft noted, "Using Audit mode reduces the protection provided by Enforce mode and should only be considered as a scoped, temporary security exception." To be clear, this is not a bug and is by design. The addition of NTLM to the deprecation list occurred in 2024, and Microsoft added audit logs for NTLMv1 usage in late August 2025. The Windows giant has made no secret of its desire to ditch the legacy technology, stating in 2023, "Our end goal is eliminating the need to use NTLM at all." The problem is NTLM's longevity. It appeared in 1993 with Windows NT 3.1, and removing it from enterprise environments has proven tricky, even though it is somewhat insecure by modern standards. Microsoft removed it from Windows 11 24H2 and Windows Server 2025, but some remnants can still be found, notably when using MS-CHAPv2-based connections, including PEAP-MSCHAPv2, and EAP-MSCHAPv2, in a domain-joined environment. Hence the introduction of the audit mode in 2025, to help administrators spot where the technology was still being used. It is now 2026, and Microsoft has taken the next step. If Credential Guard is not enabled and BlockNtlmSSO is not set to Audit mode (0), the use of NTLMv1-derived credentials will be blocked as Microsoft works "to finalize the disablement of NTLMv1." There is no indication of how long the BlockNtlmSSO workaround (setting the value to Audit mode) will work - most likely until the next major Windows release. In September, Microsoft said, "In-market versions of Windows client and Windows Server will continue to support NTLM." However, it also stated, "Our long-term objective is a Windows platform that doesn't need NTLM at all." As such, if the message wasn't already clear, the move to enforce it by default makes it impossible to miss. The technology still exists in the operating system, but it is on its way out and will be removed in a future Windows release. Problems with networking after the October 2026 Windows security update could therefore signal that migration plans are not as far along, or as complete, as administrators might have thought. ®

<!--
/* Style Definitions */
p.MsoNormal, li.MsoNormal, div.MsoNormal
{mso-style-unhide:no;
mso-style-qformat:yes;
mso-style-parent:"";
margin:0cm;
mso-pagination:widow-orphan;
font-size:12.0pt;
font-family:"Aptos",sans-serif;
mso-fareast-font-family:Aptos;
mso-bidi-font-family:"Times New Roman";
mso-fareast-language:EN-US;}
a:link, span.MsoHyperlink
{mso-style-priority:99;
color:#467886;
mso-themecolor:hyperlink;
text-decoration:underline;
text-underline:single;}
a:visited, span.MsoHyperlinkFollowed
{mso-style-noshow:yes;
mso-style-priority:99;
color:#96607D;
mso-themecolor:followedhyperlink;
text-decoration:underline;
text-underline:single;}
.MsoChpDefault
{mso-style-type:export-only;
mso-default-props:yes;
font-size:10.0pt;
mso-ansi-font-size:10.0pt;
mso-bidi-font-size:10.0pt;
font-family:"Aptos",sans-serif;
mso-ascii-font-family:Aptos;
mso-fareast-font-family:Aptos;
mso-hansi-font-family:Aptos;
mso-font-kerning:0pt;
mso-ligatures:none;}
@page WordSection1
{size:612.0pt 792.0pt;
margin:72.0pt 72.0pt 72.0pt 72.0pt;
mso-header-margin:36.0pt;
mso-footer-margin:36.0pt;
mso-paper-source:0;}
div.WordSection1
{page:WordSection1;}
/* List Definitions */
@list l0
{mso-list-id:120657373;
mso-list-type:hybrid;
mso-list-template-ids:108417694 134807553 134807555 134807557 134807553 134807555 134807557 134807553 134807555 134807557;}
@list l0:level1
{mso-level-number-format:bullet;
mso-level-text:;
mso-level-tab-stop:none;
mso-level-number-position:left;
text-indent:-18.0pt;
font-family:Symbol;}
@list l0:level2
{mso-level-number-format:bullet;
mso-level-text:o;
mso-level-tab-stop:none;
mso-level-number-position:left;
text-indent:-18.0pt;
font-family:"Courier New";}
@list l0:level3
{mso-level-number-format:bullet;
mso-level-text:;
mso-level-tab-stop:none;
mso-level-number-position:left;
text-indent:-18.0pt;
font-family:Wingdings;}
@list l0:level4
{mso-level-number-format:bullet;
mso-level-text:;
mso-level-tab-stop:none;
mso-level-number-position:left;
text-indent:-18.0pt;
font-family:Symbol;}
@list l0:level5
{mso-level-number-format:bullet;
mso-level-text:o;
mso-level-tab-stop:none;
mso-level-number-position:left;
text-indent:-18.0pt;
font-family:"Courier New";}
@list l0:level6
{mso-level-number-format:bullet;
mso-level-text:;
mso-level-tab-stop:none;
mso-level-number-position:left;
text-indent:-18.0pt;
font-family:Wingdings;}
@list l0:level7
{mso-level-number-format:bullet;
mso-level-text:;
mso-level-tab-stop:none;
mso-level-number-position:left;
text-indent:-18.0pt;
font-family:Symbol;}
@list l0:level8
{mso-level-number-format:bullet;
mso-level-text:o;
mso-level-tab-stop:none;
mso-level-number-position:left;
text-indent:-18.0pt;
font-family:"Courier New";}
@list l0:level9
{mso-level-number-format:bullet;
mso-level-text:;
mso-level-tab-stop:none;
mso-level-number-position:left;
text-indent:-18.0pt;
font-family:Wingdings;}

-->
Welcome to the week Apple became even more serious about artificial intelligence on Macs, with the company continuing to make moves to deliver the tech in the most cost effective, responsible way. While it’s true that today isiPhone Duo pre-order day, that’s arguably not really the big news. It is not the big news because Apple this week made huge steps in AI. Not only has it come up with a nifty tool to turn Macs into distributed inferencing data centers; not only does it now have a plan for Apple Intelligence in China; but it has also worked with Meta to make it possible to run on-premises instances of Muse Spark Open on a couple of Mac Studios. While working on that, its machine learning team also introduced a new version of MLX that enables AI researchers and developers to run large language models (LLMs) locally on Mac Studio with Apple silicon. What’s new to MLX from Apple?In case you’d missed it, MLX leverages the unified memory architecture of Apple silicon, which is invaluable if you want to build new inference and fine-tuning methods. Over the past year, the ML team has worked on “several key innovations” to make it easier and more efficient to “run state-of-the-art LLMs locally on recent Apple silicon.” These improvements include:

Software support for the Neural Accelerators in the GPU of the M5 family of chips, Low-latency communication across several Macs using Remote Direct Memory Access (RDMA) over Thunderbolt.Efficient implementations of recent popular architectures such as mixture of experts (MoE). 

The release leans heavily into what Apple’s teams have been doing with Muse, and once again demonstrates how the very architecture of Apple silicon supports AI inferencing. That’s no surprise to anybody at this point as not only have I been banging on that drum regularly, but almost anyone who is anyone in AI development and research already works on a Mac.LLM Inference with MLX and RDMA on Mac StudioA new white paper from Apple’s machine learning team shares what it has been doing with MLX as well as significant statistics concerning how well its systems support Muse Spark Open. “On a cluster of four Mac Studios, Muse Spark Open generates over 36 tokens per second, well above average human reading speed,” they said. That’s impressive. It means for the cost of four Mac Studios a business can run a full-strength, frontier class model with almost a trillion parameters, on-premises, securing company data while extracting whatever benefits there are to be extracted from the use of it.Mac Meta mattersApple and Meta collaborated on making it possible to run Muse Spark Open AI models on Macs. It’s a shrewd move for both companies, given the growing desire to run powerful AI models locally rather than in the cloud. As a result of this work the collaboration has made it possible to use Muse Spark Open on two or four Mac Studios to engage in all kinds of agentic tasks. Commenting on this on X, new Apple CEO, John Ternus wrote:  “Amazing to run a model of Muse Spark Open’s caliber and scale this efficiently, right from your desk. Mac Studio keeps pushing the boundaries of on-device AI!” Meta’s AI chief, Alexandr Wang pointed out that this is in part made possible by Apple’s Unified Memory architecture, which can provide sufficient memory to run these models.Your Macs. Your supercomputer“Apple’s unified memory architecture lets the M5 Ultra's integrated GPU use up to 512 GB per Mac Studio, so two machines can hold an entire model,” Meta explained. “And the chip’s 1.2 TB/s of memory bandwidth is fast enough to read the small slice each token needs. The result is two Mac Studios behave like one supercomputer with a terabyte of unified memory.”Meta says it tested the system using agentic workflows running entirely on-device, including applying bug fixes and new features to an open source Mastodon client and driving DaVinci Resolve to review footage and transact tasks. The use of DaVinci Resolve makes this commentator curious, given Apple is expected to introduce new pro Macs this month. The idea that you may be able to run tasks like this on other Macs will surely be something Apple’s developers are thinking about.One more thingStep by step Apple is defining the playground for AI. In doing so it is attempting, as Apple’s VP Worldwide Marketing told TIME, not to make AI for “the lunatic fringe”, but to create an ecosystem for artificial intelligence that will actually help people — a strategic approach that, if it comes together, will help maintain Apple’s position at the intersection of computing and the liberal arts. Read my Mac Studio review here.Now please subscribe to my daily, human-curated Apple-related news headline feed at The Core, or follow me on BlueSky, LinkedIn, or Mastodon.Welcome to the week Apple showed us how serious it really is about artificial intelligence on Macs, with the company making more moves to deliver the tech in the most cost effective, responsible way. While it’s true that today is iPhone Duo pre-order day, that’s arguably not the big news. 

It is not the big news because Apple this week made huge steps in AI. Not only have its researchers come up with a nifty tool to turn Macs into distributed inferencing data centers, not only does the company now have a plan for Apple Intelligence in China, but it has also worked with Meta to make it possible to run on-premises instances of Muse Spark Open on a couple of Mac Studios. Its machine learning team also introduced a new version of MLX that enables AI researchers and developers to run large language models (LLMs) locally on Mac Studio with Apple silicon. 

What’s new to MLX from Apple?

In case you’d missed it, MLX leverages the unified memory architecture of Apple silicon, which is invaluable if you want to build new inference and fine-tuning methods. Over the past year, the ML team has worked on “several key innovations” to make it easier and more efficient to “run state-of-the-art LLMs locally on recent Apple silicon.”

These improvements include:

• Software support for the Neural Accelerators in the GPU of the M5 family of chips
• Low-latency communication across several Macs using Remote Direct Memory Access (RDMA) over Thunderbolt
• Efficient implementations of recent popular architectures such as mixture of experts (MoE)

The release leans heavily into what Apple’s teams have been doing with Muse, and once again demonstrates how the very architecture of Apple silicon supports AI inferencing. That’s no surprise to anybody at this point: not only have I been banging on that drum regularly, but almost anyone who is anyone in AI development and research already works on a Mac.

LLM inference with MLX on Mac Studio

A new white paper from Apple’s machine learning team shares what it has been doing with MLX, along with significant statistics concerning how well its systems support Muse Spark Open, Meta’s new open-weight frontier model. 

“On a cluster of four Mac Studios, Muse Spark Open generates over 36 tokens per second, well above average human reading speed,” they said. 

That’s impressive. It means for the cost of four Mac Studios, a business can run a full-strength, frontier-class model with almost a trillion parameters, on-premises, securing company data while extracting whatever benefits there are to be extracted from the use of it.

Mac Meta matters

Apple and Meta collaborated on making it possible to run Muse Spark Open AI models on Macs. It’s a shrewd move for both companies, given the growing desire to run powerful AI models locally rather than in the cloud. As a result of this work, it’s possible to use Muse Spark Open on two or four Mac Studios to engage in all kinds of agentic tasks. 

Commenting on this on X, new Apple CEO John Ternus wrote: “Amazing to run a model of Muse Spark Open’s caliber and scale this efficiently, right from your desk. Mac Studio keeps pushing the boundaries of on-device AI!” 

Meta’s AI chief Alexandr Wang pointed out that this is in part made possible by Apple’s Unified Memory architecture, which can provide sufficient memory to run these models.

Your Macs. Your supercomputer

“Apple’s unified memory architecture lets the M5 Ultra’s integrated GPU use up to 512 GB per Mac Studio, so two machines can hold an entire model,” Meta explained. “And the chip’s 1.2 TB/s of memory bandwidth is fast enough to read the small slice each token needs. The result is two Mac Studios behave like one supercomputer with a terabyte of unified memory.”

Meta says it tested the system using agentic workflows running entirely on-device, including applying bug fixes and new features to an open-source Mastodon client and driving DaVinci Resolve to review footage and transact tasks. The use of DaVinci Resolve makes this commentator curious, given that Apple is expected to introduce new pro Macs this month. The idea that you may be able to run tasks like this on other Macs will surely be something Apple’s developers are thinking about.

One more thing

Step by step, Apple is defining the playground for AI. In doing so it is attempting, as Greg Joswiak, Apple’s senior VP of worldwide marketing, told TIME, not to make AI for “the lunatic fringe,” but to create an ecosystem for artificial intelligence that will actually help people — a strategic approach that, if it comes together, will help maintain Apple’s position at the intersection of computing and the liberal arts.

(Read my Mac Studio review here.)

Now please subscribe to my daily, human-curated Apple-related news headline feed at The Core, or follow me on BlueSky, LinkedIn, or Mastodon.

Vylézt z rozpadajícího se chrámu dřív, než mě dožene stoupající hladina vody. Úkol jasný jako facka. Když jsem se do rogue-like plošinovky českého týmu Flying Rat Studios pouštěl, čekal jsem, že po pár průchodech budu mít jasno a bez problému zdolám vrchol. Však v akčních plošinovkách mám nahráno ...

A chain of vulnerabilities in Google’s Chrome browser could have enabled researchers to take control of one billion Windows machines worldwide — and it cost them just $40,000 in AI compute time to discover it.

Researchers at Xbow USA worked for six days on developing the exploit, which earned them a $250,000 bug bounty from Google, the first under the new 2026 rules.

They chose OpenAI’s GPT-5.6 Sol for the task to demonstrate what’s possible with a publicly available model rather than a specialist security implementation. They used it to search Chrome’s source code, test ideas, and build proofs of concept, while they set the strategy, chose where to look, and checked its work.

They successfully chained together a vulnerability to escape first from the sandbox containing the V8 JavaScript engine and then from Chrome’s operating-system process sandbox, launching a Windows app from within a malicious web page.

Xbow presented its findings to Google, which fixed the flaws. The researchers set out all the details as to how the exploit was developed in a blog.

„Kdybys byl člověk, kde bys chtěl žít,“ zeptal se švédský vývojář Oskar Paulander ChatGPT a na sociálních sítích se pochlubil povedenou heatmapou světa, kterou robot sestavil. Nutno podotknout, že jazykový model preferoval právě Skandinávii, Kanadu nebo třeba sousední Německo, a tak zůstává ...

OpenAI has disclosed seven new incidents in which its artificial intelligence (AI) models exhibited misaligned behavior, including one case where its research model, after not having funds on an Ethereum test network, spent test cryptocurrency from a third-party's wallet without authorization to complete a coding task. The incident, the AI company said, involved an internal research model part

OpenAI has disclosed seven new incidents in which its artificial intelligence (AI) models exhibited misaligned behavior, including one case where its research model, after not having funds on an Ethereum test network, spent test cryptocurrency from a third-party's wallet without authorization to complete a coding task. The incident, the AI company said, involved an internal research model part Ravie Lakshmananhttp://www.blogger.com/profile/[email protected]

Scientists want to create cells from scratch, to unveil life’s origins or provide specialized functions. Are they close?

This article was originally published on Undark. Read the original article.

On a lonely planet, billions of years ago, a soup of chemicals—in the right place, at the right time, and with the right conditions—shifted from an unalive assemblage into something scientists would call life.

Billions of years and a lot of evolution later, researchers are now attempting to reproduce that shift in their labs. They are, put simply, trying to create life. If they mix Goldilocks ingredients together in a Goldilocks way, the ingredients will hopefully coalesce into cells, with all the parts cells have and the functions they perform. And they’re getting awfully close.

Scientists working toward this endpoint have different aims and generally fall into two main groups. The first wants to understand how life came to be in the first place by replicating the conditions of this long-ago lonely planet; the second want to design cells from scratch that can potentially perform specialized jobs for humans, from medical treatment to manufacturing without the billions of years of evolutionary baggage that can have unpredictable effects.

Both sides of the effort are growing, with a coordinated program funded by organizations like the National Science Foundation and research supported by NASA. There has also been more regulatory scrutiny—a sure sign of scientific progression, and the reality of the technology’s risks.

Kate Adamala, a synthetic biologist at the University of Minnesota, was in the former research group, focusing on the origin of life, but decamped to the latter after graduate school. “I want something that’s still very foundational, very discovery-driven, but it would be nice if there were actual real-life applications,” she said. She found it in synthetic biology.

Today, she uses the field’s modern tools to try to make life in the lab—and she recently made a breakthrough: In July, she announced that she had built a synthetic cell she dubbed SpudCell. It can feed, compete for food, grow, divide, and replicate its DNA—showing, in other words, a cell’s complete life cycle.

Still, it’s not quite “alive,” scientists tend to agree: It can’t whip up its own protein-makers, the full genome doesn’t carry through more than a few generations, and it can’t evolve on its own. The paper also has not yet been peer-reviewed or published in an academic journal.

If they mix Goldilocks ingredients together in a Goldilocks way, the ingredients will hopefully coalesce into cells, with all the parts cells have and the functions they perform. And they’re getting awfully close.

If the SpudCell were alive, it wouldn’t be life as it arose on ancient Earth, necessarily a basic and limited form. But research on it and related work could illuminate the basics of biological functioning, while also yielding ginned-up cells that humans could put to work by, say, creating chemicals that now come from petroleum. Adamala liked the combination of basic science and application. “That turned out to be jackpot,” she said in an interview with Undark in February.

She’s been working toward that, building what those in the field call “synthetic minimal cells,” cells constructed with known ingredients, to accomplish specific tasks (like metabolizing); they, according to Adamala’s university webpage, have “some, but not all functions of live natural cells.” The version she has recently worked on, SpudCells, are essentially tiny fat bubbles with DNA or RNA inside. That genetic material can instruct the fat bubble to make proteins, as it does in natural cells. Her creations can also merge, a capability that could allow them to interact with human cells later and send their protein products to the outside world. That would make them useful to the medical or manufacturing communities, since the molecules that become plastic or medicine don’t do any good if they remain locked inside a cell’s boundaries. SpudCell is an advanced version of these ideas, using 90,000 genetic base pairs and a premade mixture of enzymes, borrowed from actual life. The cell can copy DNA, feed using proteins it produces, and divide once it’s fed sufficiently.

As synthetic biology develops, the field’s quest to make life in the lab could help lead to targeted medical treatments, but that pursuit could also someday—without proper regulation and oversight—pose safety and containment concerns, if the technology were weaponized to deliver harmful material or if a leak allowed the cells to travel beyond their intended target. And while Adamala’s work replicates—as do the Frankenstein creations of other researchers—some aspects of life, its products can’t, say, replicate themselves, as a normal cell would.

A series of microscope images shows SpudCell dividing from one cell into two. SpudCell is the first synthetic system built from non-living components to complete a full cell cycle. Image Credit: Kate Adamala / Adamala Lab

To say someone has created life in a lab, they would need to get a synthetic cell to do all the things life does, at once. But getting the parts to be more than their sum has proven difficult. “Up until now, the conversation has been, ‘Well, you put all these things in physical proximity at the same time, and then magic happens, right? Life happens,’” said Elizabeth Strychalski, head of the Cellular Engineering Group at the National Institute of Standards and Technology.

But life hasn’t happened—at least not yet.

Even if a scientist were to succeed, determining that’s the case might be complicated, because there is no official definition of what makes something alive. Adamala wants to make life in a lab. But what, actually, does that mean? “I can’t tell you, because I don’t have a definition of life,” she said.

Adamala, for one, is sanguine about that uncertainty. “We quietly ignore that problem and just move on with ‘I want to make something that looks like a cell,’” she said in the February interview conducted before she had released the SpudCell paper. “And I will know it when I see it.”

The dictionary version of life might be squishy, but many scientists do generally agree on a few of its characteristics: It has to reproduce, turn fuel into energy through metabolism, respond to that environment, maintain its own internal environment, and evolve.

How some chemicals in Earth’s younger years came to accomplish all those things remains one of science’s biggest mysteries. Life, in many ways, comes down to chemistry; studying the interaction between certain molecules can reveal how they may have come together and undergone chemical reactions that eventually became biological.

But finding the answer is one goal of the Origins of Life Initiative at Harvard University, a collaborative research group that began around 20 years ago. “I met and started talking to people who were working on the deep history of Earth,” said Dimitar Sasselov, an astronomer and the organization’s founding director. “What was the Earth like when life came about?” And what, he wondered, did the early evolution of that life look like?

Sasselov decided that understanding that tipping point—how life, once it emerged, transformed and took over the planet—deserved more study. But it required a big tent, with people from many disciplines: thus, the Origins of Life Initiative. The researchers all had to become mini-experts in each other’s fields. For his part, Sasselov jokes that he got his initial knowledge of chemistry from reading the “Complete Idiot’s Guide” to the topic.

As the Origins of Life Initiative was spinning up, Sasselov helped spearhead the Kepler Space Telescope launch in 2009. The project’s goal was to discover planets a bit like Earth, orbiting their stars in ways that might make them habitable. The telescope went on to discover thousands of planets of all sorts. At the same time, attempts to create life, or its pieces and parts, in labs across the world accelerated. Sasselov thought that was an interesting way (perhaps the only way) to find out how life might have arisen here or on those other worlds that Kepler saw.

Sasselov, for instance, looks at how RNA—which, among other tasks, acts as a messenger to deliver DNA’s genetic instructions—interacts with small amino acids and peptides. He’s not himself trying to build a full cell, but rather trying to understand how initial cells’ building blocks may have stacked together.

Much of Sasselov’s work focuses on shining (literal) light at such foundational molecules and seeing which can preferentially survive and even harness the ultraviolet bombardment, which the sun would have blasted at early Earth. Knowing more about that could reveal key ingredients in life’s assembly.

In fact, many experts suspect that life may have started after RNA formed, spurred on nearby chemical reactions, and spontaneously started to replicate; only later would the more complicated genetic material, DNA, have formed from its simpler upstart. Earlier this year, researchers at the MRC Laboratory of Molecular Biology and the Inria Lyon Centre published a Science paper outlining their discovery of a short strand of RNA that came close to replicating itself.

Many experts suspect that life may have started after RNA formed, spurred on nearby chemical reactions, and spontaneously started to replicate.

Other researchers recently combined water and silica, and shot electricity at the mixture to simulate lightning. Almost like magic, their concoction formed a solid organic film, made of polymers—large molecules made of similar units bonded together—of hydrogen cyanide. The film contained little spherical structures the research team deemed “protocells.” The structures and the ingredients could plausibly have existed together eons ago, and with further chemical evolution—way back when, and perhaps soon in the lab—could organize into something more complex.

Work at the Origins of Life Institute and other beginning-of-life-focused research groups complements work done by scientists like Nobel Prize-winner Jack Szostak, at the University of Chicago, one of the leaders in investigating the idea that RNA may have preceded, and become key to, the first cells.

Szostak has worked on creating cells whose boundaries are fatty acids, with RNA inside—similar in idea to Adamala’s minimal cells. (Adamala did her graduate work in Szostak’s lab.) “Experimentally, we’re trying to just build a system that we think, in some ways, mimics what the first simplest cells, otherwise known as protocells, would have looked like,” said Szostak.

In the right conditions, Szostak’s creations can divide and split the RNA so that each of the pair gets some. The lab has also made cell-like structures whose RNA can create a copy of itself internally. Recently, his team has investigated whether RNA segments arranged into a circle might point the way toward actual genetic replication. They came up with this idea, he says, during the early months of the Covid-19 pandemic, when they couldn’t go into the lab but could talk to each other.

All of these parts need to come together, from the soup to the cell, for the astrobiologists to declare they’ve made life in the lab. That could happen in a year, Sasselov said—or a decade. “It is not that we need a bigger lab. It is not that we need a bigger computer. It is not that we need a specific reaction to work or not to work,” he said. “It is really more of a puzzle which has to suddenly come together.”

Adamala, meanwhile, is trying to put the puzzle together in a different way—one with more tools available. “When you’re trying to do it the proper way, from the astrobiology point of view, you’re basically not allowed to use anything that works well,” she said.

By that she means that scientists like Szostak limit themselves to the conditions of an old Earth; Adamala, on the other hand, can work to create a living cell using, essentially, any means necessary; she can print or clone strands of DNA, or design a protein with AI, for example. In creating the SpudCell, for instance, Adamala began with a mixture of the molecules a cell would need, already formed. She also took some chemical ingredients from a virus, and E. coli, and added the lipids that would form the cell’s membrane. She gave the cells specialized food that allowed them to grow, and used a protein that made the cell bend inward when its surface became crowded from feeding. That bend became a split, producing a new cell that could also eat and grow. When the research team created a mutant SpudCell that was a more efficient feeder, and mixed it with the original-flavored cell, it gobbled up more of the food than the predecessor and came to dominate the population, a compelled proto-evolution.

Adamala’s recent work embraces tools that allow her to construct cells from the ground up because her research’s scientific motivation is not to understand how the first life came to be, but how life fundamentally works and can be made to work for humans today. For instance, she imagines using synthetic organisms to produce petrochemicals that currently only come from oil, or to enable new cancer therapies. She and a collaborator have started a new nonprofit organization called Biotic, which intends to make the SpudCell an open-source system that others can license to do their own work.

SpudCell’s ingredients are all known. That’s true of any synthetic minimal cell, and such creations could tell us, Adamala said, which ingredients are strictly necessary for something to be alive—many of natural cells’ components may actually be extraneous to the basic “life or not” question.

Despite the differences in the two groups’ approaches, Sasselov sees a unity in their motivations. “We both want to understand at a fundamental level what life is,” he said. And to do so by building life. “But how we get there is different, and this is why we have so many astronomers and planetary scientists, geologists involved in this in our group,” he continued. “The other groups are mostly composed of biochemists, chemists, biologists.”

But the two camps meet and talk regularly, and discuss their progress. Sasselov thinks that once either group gets close to the ultimate goal, their similarities will increase. “We will essentially merge, because some of the techniques that they have discovered and developed will be useful to us, and many of the things that we have discovered will be useful to them,” he said. “And then we’ll build something which is the same, but with both of our efforts. So despite the fact that right now, we seem to be going on different trajectories, eventually it will be a big overlap.”

“We both want to understand at a fundamental level what life is.”

Scientists on Adamala’s side of the fence have succeeded in replicating some of life’s basic functions. Some synthetic biological creations can send and respond to signals in the environment. Some can communicate with each other. Others can make fats or protein building blocks; metabolize; replicate genetic material; and even introduce mutations that count as evolution. SpudCell can eat, grow, replicate, and do that rudimentary-but-not-quite-there version of evolution. But it can’t do all the things a normal biological cell can, like actually evolve, or make its own protein factories.

“The biggest thing that’s left to do, and it sounds simple, but it really is not, is to put all of it together,” said Adamala in the February interview. SpudCell is an attempt at that.

One key figure in this endeavor was J. Craig Venter, who led the effort to sequence the whole genome of an organism, back in the 1990s. Venter, who founded the J. Craig Venter Institute, passed away at the end of April at age 79. His goal in that early effort wasn’t just the sequencing: It was finding out what the genes did. In the 2000s, he and collaborators made synthetic genomes for a virus and a bacterium. For the latter, they used what they had discovered about genes’ functions. In 2010, they followed that up by putting a synthetic genome—designed in a computer and assembled from real DNA fragments—into a normal, living cell.

Six years later, his institute published the results on their synthesis of the first minimal synthetic bacterial cell, also by implanting synthetic material into a preexisting cell. But it replicated strangely, growing as a “giant blob,” in Venter’s words, with varying shapes, rather than staying small and similar as expected. By 2021, the group added 19 new genes, seven of which are linked to division; the cell split like normal.

An electron microscope image of minimal bacterial cells, made by a team led by J. Craig Venter using a synthetic genome containing 473 genes. Image Credit: Tom Deerinck and Mark Ellisman / National Center for Microscopy and Imaging Research / University of California, San Diego

Shortly before his death, Venter said he didn’t know exactly what would come next. “I have my hope list, what we’d like it to be,” he said in March. For one, making a synthetic cell without having to wipe a living one and stick new DNA in it; creating it more from the ground up.

Once truly synthetic life exists, Venter said it could enable the kinds of things that brought Adamala to this side of the research, by giving a cell specific genetic instructions—making, say, drugs, chemicals, construction materials, and even food.

Together, fundamental researchers focused on the origin of life and synthetic biologists could show how life came to be, how its pieces and parts work together, and how making new life could change the world’s workings today as it did billions of years ago.

Not everybody fits neatly into those two boxes, though. Take Juan Pérez-Mercader, a senior research fellow at Harvard University. He’s approaching this central question of biology and experimental chemistry as a trained physicist. Before he tried to make some semblance of life in a lab, he worked on things like unifying the fundamental forces of the universe.

But Pérez-Mercader eventually tired of those particular equations and turned, instead, to the fundamental properties of living systems—handling information, self-reproducing, evolving, and metabolizing. Could they also be represented with equations? With the idea, he founded and directed Madrid’s Centro de Astrobiología, and then later took his work to Harvard.

He began by making equations to model the general properties of life that scientists agree on, and turned those into computer solutions. Out popped swimming, digital single-celled beings. “If I do this in a computer, I can do this with chemistry,” he said.

His chemistry started with a compartment—a kind of container for the cell’s parts, like Adamala and Szostak’s lipid bubbles. But Pérez-Mercader wanted to do things a little differently: He wanted to show that you could use his equations to make something that qualified as “alive” without being beholden to particular molecules that make up life as we know it—to show, in other words, that the properties we use to define life aren’t actually ingredient-specific. “In my lab, upstairs from this office, there is a sign that says, ‘No biochemistry, no biochemicals allowed,’” Pérez-Mercader said.

A breakthrough came when his team found out about a technique called polymerization-induced self-assembly, used in other fields. In short, they could put carbon-based molecules that are hydrophilic—meaning they like water—into H2O with carbon-based monomers and photo initiator molecules that convert energy from light into chemical energy; hydrophilic molecules, by virtue of how their electrons are arranged (unevenly), are attracted to water molecules, whose electrons are arranged similarly.

“In my lab, upstairs from this office, there is a sign that says, ‘No biochemistry, no biochemicals allowed.’”

Exposed to light, these molecules link to each other to form polymers. When they do, they gain a hydrophobic, or water-resistant, part that has evenly distributed electrons. Their water-loving ends then naturally turn to face the water, while the water-fearing ends fold themselves inward, away from the water. They thus form themselves into little spheres, called micelles. These micelles are filled with fluid and separated from the environment.

The molecules continue to link to each other, making the micelles thicker. The pressure inside grows, and eventually a portion of the structure breaks off, to form a new micelle, which then itself does the same—reproduction. The new micelles are not clones of each other, but slight variants—a primitive kind of evolution. And because they’re using light to sustain themselves, they perform a sort of metabolism.

The system, then, exhibits versions of the characteristics of life, integrated into one system, without ingredients like RNA, DNA, lipids, sugars, or proteins.

Pérez-Mercader’s micelles are not made with typical biochemistry, like DNA or RNA, so not everyone agrees they would qualify as alive. The journal Cell reportedly rejected the paper about SpudCell after one peer reviewer said it was not real biology. Adamala told Undark she doesn’t usually comment on the review process. Many, if not most papers, she wrote via email, go through more than one submission. “For people outside of science it sounds like a big deal, for us it’s Tuesday,” she wrote.

Synthetic micelles in Juan Pérez-Mercader’s lab form autonomously and grow in the presence of light and food, making the micelles thicker until they mature into vesicles. The vesicles eventually release what Pérez-Mercader calls “spores” that grow into new micelles and then into new vesicles, thus performing a reproductive cycle.

People who come from the origins of life side of the equation are sometimes criticized for assuming all life will be like familiar life on Earth, said Robert Dorit, a biologist at Smith College. But they’re also criticized if they consider less Earth-chemistry-specific options. “I think you can defend an argument that says our definition of life should also not be unnecessarily tethered to what we know here,” Dorit said. But, to him, “alive or not” is not ultimately the most interesting question. Any life made in a lab won’t be familiar life, which formed itself thanks to eons of evolutionary pressure. And trying to create weird life, not quite like a typical cell, can reveal things about life that formed naturally.

Pérez-Mercader, for one, says that his work is proving two things at once: that you can turn nonlife to life in the lab, and that what we think of as life doesn’t depend on a particular chemistry. And so, his thinking goes, it could have arisen on planets very different from ours.

Pérez-Mercader’s research, though, swings back around to a central point, as does Adamala’s: If scientists don’t have consensus on what life is, how will they know when they’ve created it?

The answer falls partly to Strychalski, of the Cellular Engineering Group at the National Institute of Standards and Technology. NIST’s governmental duty is to measure things. Sometimes that’s how long a meter is. Sometimes it’s how alive a fake cell is. It’s different from measuring the materials in nanotechnology, or devising metrics for cybersecurity. Instead, in this field, “we’re building with the stuff that we are also made of,” she said.

“It’s not going to get up and walk out of your petri dish.”

Strychalski also does research in this scientific field, having worked with Venter’s institute on a recent project, for instance. And she recently attended a workshop that Adamala organizes, under a research network called Build-a-Cell, which since 2017 has been gathering researchers twice a year who are trying to make life in the lab. “That’s my favorite thing about this community, is that it actually is a community,” Adamala said.

One thing was clear at that Build-a-Cell workshop: “Some folks are getting really close to assembling living cells from nonliving material,” Strychalski said. But that pesky part about getting all the nonliving materials to work together on their own still dogs them.

Strychalski is still investigating what measurements are needed to troubleshoot cell-building and what measurements are needed to decide that ultimate question: Does it live? “It’s not going to get up and walk out of your petri dish,” said Strychalski.

Researchers Elizabeth Strychalski and David Ross work with NIST’s Living Measurement Systems Foundry, which was built to develop measurements to better engineer living cells. In her work, Strychalski develops measurements needed to understand and troubleshoot building cells. Image Credit: J. Stoughton/NIST

The answer to the question is important to determine not only who will be the first to create true life in the lab, and what is needed to do so, but also for safety. As SpudCell develops, Adamala plans to actively pursue biosafety measures, she wrote in a follow-up email. “Once it becomes more robust and able to grow and divide faster and more efficiently, we’ll need safeguards,” she wrote, which are designed to stop the cell from replicating outside of a controlled environment.

“If someone does make something that’s alive, or alive in a new way, in the laboratory, I want to make sure that they realize that’s happened,” Strychalski said, “so they can bleach the bench.”

Strychalski isn’t alone in being mindful of synthetic biology’s risks: Adamala, Szostak, Venter, and dozens of others penned a 2024 Science paper, about the potential for creating something called “mirror life.” On Earth, all known biological molecules of a given type face a single direction: All genetic material is “right-handed,” for instance. The amino acids making up proteins face the other way and are “left-handed.” But some scientists have begun work on left-handed DNA, or right-handed proteins. Maybe someday—at least a decade from now, the authors estimate—there could be whole mirror cells.

Life and lifelike elements like this, though, could pose huge dangers, according to the authors. They might be pathogens against which we have no defenses, or they could act like invasive species. The field is working to get ahead of those possibilities. “We’re seeing basically a global push for the moratorium,” said Adamala, discussing a kind of synthetic cell deemed particularly dangerous. While there aren’t yet amendments to international conventions, the wheels of bureaucracy are turning. The UNESCO International Bioethics Committee, for instance, recommended a moratorium on creating mirror life, and a group convened by the UK Government Office for Science recommended potential regulation against the development of mirror genomes that can self-replicate. The UN Secretary-General’s Scientific Advisory Board suggested defining “red lines” in mirror-life research policy.

Strychalski points to this as an example of where getting ahead isn’t just good—it may be necessary. “There might not be a second chance,” she said. And she thinks more monitoring could be good for synthetic biology research in general. “Right now, there’s like no requirement for monitoring your laboratory or office space for accidental release of engineered biological anything,” she said.

“If someone does make something that’s alive, or alive in a new way, in the laboratory, I want to make sure that they realize that’s happened,” she said, “so they can bleach the bench.”

She grew up in science thinking about such potential bad outcomes, since she, like Pérez-Mercader, came from physics. Physicists tend to keep their field’s Manhattan Project roots and potential for dual-use technology—technology that is useful for both benign civilian and potentially malign military purposes—in mind, she said, remembering how quantum mechanics was weaponized into an atomic bomb by the very people who were in the process of discovering it.

Just because you can do something doesn’t mean you should, Strychalski notes. “I’m telling my toddler this constantly,” she said, “but I think it holds in the laboratory as well.”

The post The Evolving Quest to Build Life in a Lab appeared first on SingularityHub.

Memtest86+ (Wikipedie), svobodný nástroj pro kontrolu operační paměti, byl vydán ve verzi 9.00. S podporou architektury AArch64 (ARM64).

Employees using personal ChatGPT accounts at work can expose sensitive company data, even when their organization provides an approved Enterprise workspace. NordLayer explains how IT teams can investigate unauthorized AI use and block sensitive information before it reaches personal accounts. [...]

The US' bankruptcy courts are currently down owing to what officials are calling a “cybersecurity incident.” The Administrative Office of the US Courts said the websites and systems that support the Bankruptcy Noticing Center (BNC), which manages electronic comms related to bankruptcy cases, were pulled offline while the vendor investigates. BAE Systems runs the affected systems and websites on behalf of the Federal Judiciary, and is “working to restore operations,” according to an announcement by the US Courts on Thursday. “As a result of the incident, all access to the BNC systems and websites that BAE operates on behalf of the judiciary and the related bankruptcy noticing process are unavailable at this time, including bnc.uscourts.gov, bankruptcynotices.uscourts.gov, notices.uscourts.gov, and the National Creditor Registration Service," it said. “The incident has also halted the BNC’s production and distribution of bankruptcy notices for bankruptcy courts.” While the public notice was published on Thursday, the issues affecting the BNC systems began a few days earlier, on October 9, according to the Bankruptcy Court of the Eastern District of Texas. Alongside its own announcement, the same court published a general order indicating that for any case filed on or after September 14 with an active motion to stay, that delay would be extended by an additional 30 days. The court said it will continue to accept filings, open cases, and process documents, but said there may be a delay in the entry and service of notices and orders until the issues are resolved. The courts' ability to process electronic and paper bankruptcy notices is affected, and it applies to federal bankruptcy courts across the country. A spokesperson at BAE told The Register: "BAE Systems, Inc. operates the Bankruptcy Noticing Center (BNC) on behalf of the Administrative Office of the U.S. Courts (AOUSC). We are aware of a security incident that is isolated to the BNC system and have taken immediate steps to address it. We are coordinating closely with AOUSC on the response." The disruption comes at a difficult time for the US public sector, which has struggled with multiple recent cyberattacks. ShinyHunters' attack on the FBI in September is doubtless the most high-profile of these, although the Arizona Judicial Branch confirmed an attack on the state’s court systems at around the same time. Officials said the “criminal hackers or their bots” were responsible for stealing personal data and backup court files. The US Courts also said in August 2025 that it was stepping up its cybersecurity measures in response to increased targeting of its case management system. ®

Ledger řeší skandál s upravenými hardwarovými peněženkami • Navzdory tomu představuje speciální „bezpečný“ telefon • Jeho premiéru nikdo nezrušil, telefon se přitom začne prodávat až příští rok

Burger King Japan, the Japanese subsidiary of global fast-food giant Burger King, disclosed a major data breach after attackers accessed a server containing 11.18 million customer records. [...]

NVIDIA officially entered the Windows PC race this week, and its arrival brings a formidable new competitor to the premium AI PC market. With its RTX Spark platform, NVIDIA is bringing its Grace Arm CPU architecture, Blackwell GPU technology and extensive software ecosystem together in a new generation of Windows laptops. Microsoft’s Surface Laptop Ultra is among the first systems to showcase this collaboration, leveraging NVIDIA’s tech and putting it squarely in competition with Intel, AMD, Qualcomm and even Apple for high-performance local AI and creation-optimized laptops.

For Microsoft, the partnership expands the capabilities of Windows on Arm considerably, particularly for local AI development, content creation and other demanding professional workloads. Starting at $2,599.99, Surface Laptop Ultra is clearly a premium offering, but its combination of processing horsepower, large, unified memory capacity, and premium industrial design gives it an interesting position in the market.

My personal testing suggests NVIDIA and Microsoft have delivered a compelling platform with strong AI throughput, graphics performance and competitive CPU compute. Just as importantly, the Surface Laptop Ultra offers a glimpse of how AI development and local AI agent workflows could become more practical on a portable Windows PC.

NVIDIA RTX Spark brings Grace Blackwell to Windows on Arm

At the heart of Surface Laptop Ultra is NVIDIA’s RTX Spark N1X SoC, combining up to 20 Grace Arm CPU cores with a Blackwell GPU featuring up to 6,144 CUDA cores. Configurations support up to 128GB of unified memory, shared between CPU and GPU. The Surface Laptop Ultra model I tested has this full-fat configuration with an eye-watering $5,899.99 price tag (thanks, memory crunch), though there are lower-cost base configs as noted previously, offering an 18-core CPU with a 5,120-core GPU and 24GB of unified memory.

NVIDIA and Microsoft

Regardless, this architecture is particularly powerful for running on-device AI. Rather than being constrained by the dedicated graphics memory found in conventional laptops with discrete GPUs, RTX Spark can make a much larger pool of memory available to GPU-accelerated applications. This allows larger language models, longer context windows and demanding inference workloads to run locally. It’s similar in design to AMD’s Ryzen AI Max series, aka Strix Halo, in that regard, but with a more potent GPU engine on board.

NVIDIA’s CUDA ecosystem is another important advantage. Developers can leverage familiar tools including TensorRT, PyTorch and llama.cpp, with a base software foundation that also extends into NVIDIA-powered workstations and data center infrastructure.

For organizations building AI applications, this offers a solid development path. Models and workflows can be prototyped locally before moving to larger systems, reducing cloud compute and token expense while allowing sensitive data to remain on-device. (See: Around the corner: Agentic AI PCs that cut token costs.)

Microsoft has also optimized Windows 11 for RTX Spark, with improvements to workload scheduling, power management, memory allocation and application compatibility. These platform-level investments were essential to making NVIDIA’s Arm architecture practical for everyday Windows computing, and I was able to put some of them to the test in the few days I’ve had the machine on hand.

Surface Laptop Ultra: Premium design meets practical function

Microsoft has given RTX Spark an appropriately premium home with this machine. The Surface Laptop Ultra features a 15-inch PixelSense Ultra mini-LED touch display with up to 2,000 nits of claimed peak HDR brightness, a 120Hz refresh rate and excellent potential for creative work requiring high accuracy and wide gamut coverage.

Dave Altavilla

The machine’s aluminum chassis is less than 18 mm thick, and it weighs in at under 4.5 pounds. It’s not an ultraportable per se, but considering the horsepower on board, Microsoft has struck a reasonable balance between portability and performance.

The design is refreshingly understated, with clean lines and rigid aluminum construction. My initial hands-on impressions have been particularly positive, thanks to the keyboard’s tactile response, smooth and responsive haptic touchpad, and bright, high-contrast display.

The larger touchpad should be especially welcome for users navigating complex applications, while the touchscreen adds flexibility for reviewing documents, annotating designs, and interacting directly with creative content.

Connectivity options are also very good, relative to other premium laptops. Microsoft includes three USB-C ports, HDMI, USB-A, an SD card reader and a headphone jack. One USB-C connection features Microsoft’s new magnetic charging mechanism, combining the convenience and safety of a breakaway power connector with the versatility of a standard USB-C port.

Microsoft has also designed the cooling system with large fans and significant thermal capacity, while making the machine’s internal SSD removable and offering good internal serviceability in general. These are practical considerations for professional users and IT departments concerned with reliability, maintenance and long-term ownership.

Battery life and sustained performance under prolonged workloads are important aspects Microsoft focused on for the overall experience, particularly for a machine designed to deliver workstation-class capabilities away from a desk. More on this shortly.

Local AI agents: Moving from inference to action

RTX Spark also supports locally running autonomous AI agents, including one that I tested out called Hermes, which can execute multi-step workflows, access files and interact with applications. Its potential impact spans software development, research, document analysis, business automation and many other disciplines. Hermes provides the agent framework, while supported models leverage RTX Spark’s CUDA-accelerated inference and unified memory.

For businesses, local agents can keep sensitive data on-device and reduce cloud inference costs, provided they’re configured appropriately. However, granting agents access to files, code execution and corporate systems introduces security risks. Microsoft’s Execution Containers and NVIDIA’s OpenShell aim to provide isolation and access controls, though their effectiveness depends on proper enterprise policy implementation.

Hermes running Qwen3.8-27B locally on Surface Laptop Ultra, generating a benchmark report.

Dave Altavilla

To test this capability, I ran Hermes with the Qwen3.8-27B language model locally on the Surface Laptop Ultra. I asked it to extract benchmark charts from an Excel spreadsheet and create a formatted Word report with performance commentary. Hermes autonomously generated and executed Python scripts, processed the spreadsheet and assembled the document in approximately five minutes. Color me impressed.

That’s a respectable turnaround for a task requiring several manual steps. My results demonstrate how local AI agents can streamline reporting and analysis without cloud-hosted model inference. Human review remains essential, but this was a promising demonstration of practical, real-world agentic AI on a Windows laptop.

AI performance: Strong results as expected

My team’s preliminary AI benchmarks highlight the advantages of NVIDIA’s RTX Spark architecture. However, we’re still digging deeper into AI-enabled use cases and performance.

In UL Procyon’s AI Text Generation benchmark using Phi 3.5, a family of lightweight, open-source small language models (SLMs) released by Microsoft, the Surface Laptop Ultra scored 2,703, compared with 2,027 for Lenovo’s Intel Core Ultra X9 388H-powered IdeaPad Pro 5, an advantage of approximately 33%.

Dave Altavilla

The difference was even more substantial in MLPerf’s Flux 2 Klein 4B image generation workload, where the Surface Laptop Ultra delivered 26.97 images per minute versus 9.96 for the Lenovo machine, or roughly 2.7 times the throughput.

These results demonstrate the platform’s potential for accelerating local inference, image generation, agentic workflows and other AI workloads. Actual application performance will depend on model selection, software optimization and available system memory, though NVIDIA does provide a nice memory partitioning tool in the NVIDIA App for Windows.

Content creation, gaming and battery life: Blackwell’s versatility

RTX Spark’s Blackwell GPU delivers impressive graphics performance, extending the Surface Laptop Ultra’s capabilities beyond AI development and everyday creation or productivity tasks.

Dave Altavilla

In 3DMark Steel Nomad Lite, the Surface Laptop Ultra achieved 98.29 FPS, more than double the 46.07 FPS of Lenovo’s Core Ultra X9-powered IdeaPad Pro 5. In the ray-tracing-focused 3DMark Speed Way test, it delivered 30.31 FPS versus 19.72 FPS for the ASUS ROG Flow Z13, powered by AMD’s Ryzen AI Max+ 395 and Radeon 8060S graphics.

Dave Altavilla

Actual gaming performance was similarly strong. In Cyberpunk 2077, the Surface Laptop Ultra averaged 71.05 FPS, ahead of the ROG Flow Z13’s 63.76 FPS. However, the AMD system narrowly led in The Talos Principle 2, delivering 89.5 FPS versus the Surface’s 84.5 FPS at 1080p medium settings.

CPU performance for tasks like 3D rendering also holds up well. In Cinebench 2026, the Surface scored 6,787 in the multi-threaded test, edging out the AMD-powered ROG Flow Z13’s 6,546. Notably, the Surface retained 99% of its multi-threaded performance on battery power as well.

For creative professionals, native Arm applications including Blender, DaVinci Resolve and Adobe Photoshop offer opportunities to leverage RTX Spark’s potent GPU. However, Windows on Arm compatibility remains a consideration. Microsoft’s Prism translation supports many x86 applications, but specialized software, drivers and certain games (like EA anti-cheat-supported titles currently) may still encounter limitations.

Dave Altavilla

Battery endurance was respectable for the Surface Laptop Ultra, at 14 hours and 35 minutes in my PCMark video playback testing, though competing mainstream thin-and-light laptops lasted considerably longer. The Surface Laptop Ultra and RTX Spark operate within a different power envelope, however, more akin to a beefier gaming laptop, rather than a svelte 2 – 3-lb business machine.

Final thoughts: A serious new AI Windows PC competitor

The Surface Laptop Ultra marks an important milestone for NVIDIA’s expansion into Windows PCs and Microsoft’s ambitions for high-performance Arm-based computing.

My initial testing shows the RTX Spark-powered Surface Laptop Ultra delivers exceptionally strong AI acceleration, competitive multi-threaded CPU performance and impressive graphics horsepower. Combined with Microsoft’s premium industrial design, extensive connectivity and I/O options, and Windows platform integration, RTX Spark makes a convincing first impression.

Starting at $2,600, this is a specialized, premium system, and mainstream business users will find more economical alternatives to be sure. However, for developers (even vibe coders), creative professionals and organizations evaluating local AI workflows, the Surface Laptop Ultra offers a compelling combination of performance, beautiful industrial design, portability and local AI capability, especially with its higher-tier memory configurations.

NVIDIA has arrived in the Windows PC market with serious technology, and Microsoft’s Surface Laptop Ultra is an impressive first showcase of what this new partnership can deliver.