A new experiment from Harvard, Caltech, MIT, and QuEra researchers is being described as one of the clearest steps yet toward practical quantum computers. The team demonstrated an integrated system that layers dozens of quantum error correction rounds on top of a neutral atom processor, suppressing errors below the threshold needed for scalable computation.
The work appears in a recent Nature paper and expands on several years of progress in neutral atom platforms. The core promise is simple to state and extraordinarily hard to achieve. Quantum systems lose their delicate states easily. Unless engineers can continuously detect and correct those faults, large quantum machines will never scale.
A system that finally reduces errors as more qubits are added
Lead author Dolev Bluvstein says the new hardware combines multiple correction techniques inside a single architecture. The system uses reconfigurable arrays of nearly five hundred atoms, repeated stabilizer measurements, and machine learning based decoding to push performance below the critical point where additional qubits reduce errors instead of introducing new ones.
The group used surface codes and more advanced logical codes to explore entanglement, transversal gates, and teleportation based logic. They also showed mid-circuit reuse of qubits, a feature that dramatically increases cycle rates and allows deeper circuits without runaway entropy.
Senior author Mikhail Lukin describes the work as the first time all essential elements for a scalable, error corrected quantum processor have been demonstrated under one roof. It does not produce a commercial level machine, but it lays out the operational blueprint.
A quiet but intense race inside quantum research
Google’s Hartmut Neven called the progress a meaningful step in the multi-platform race to build useful quantum computers. Superconducting qubits, trapped ions, photonic systems, and neutral atoms each offer tradeoffs in connectivity, stability, and scalability. Neutral atoms have become a serious contender because they can be rearranged freely, entangled in parallel, and cooled as they run.
The Nature results also emphasize the importance of teleportation based logic. Instead of physically routing qubits around a device, instructions move through entangled states. This allows deeper circuits while keeping entropy contained, one of the toughest engineering obstacles in the field.
The path ahead
Even with this progress, fault tolerant machines with millions of physical qubits remain distant. Every part of the architecture must scale together without losing coherence. But for the first time, several leading researchers say the blueprint looks practical.
After three decades of theory and incremental demonstrations, the foundation for universal, error corrected quantum computation feels closer than ever.
Related Quantum Stories on Laterstack:
IBM pushes quantum chips to new limits as IonQ and Quantum Art race ahead
Salt Unlocks Metallic Nanotubes for Quantum and High Speed Tech
While the world argues about AI, another transformation is unfolding far from the usual tech centers. solar startup Africa is expanding rapidly as local founders build energy systems designed for people who have spent decades waiting for reliable power. This shift is not theory. It is visible in homes, clinics, markets, and schools that once went dark every evening.
The growth of solar startup Africa comes from simple forces. The cost of panels has dropped. Mobile payments work everywhere. Community trust accelerates adoption. Most importantly, the builders grew up with the problems they are solving. They remember outages that lasted days. They know what a broken grid means for a small shop or a rural clinic.
A ground level Solarpunk reality
This is Solarpunk without the aesthetic filter. It is not a design trend or a social media fantasy. It is real infrastructure for communities that cannot wait for national grids to catch up.
Shops can stay open at night. Clinics can preserve vaccines. Farmers can cool crops instead of watching them spoil. Students can study without burning candles. And people can work remotely because their power no longer cuts out every afternoon.
solar startup Africa creates opportunity by making electricity stable. That one change shifts everything else.
Why this acceleration is happening now
A few key shifts pushed this movement forward.
Affordable solar panels made small scale grids possible.
Mobile money created a simple pay as you go system for families and businesses.
Local founders understood exactly where the grid fails and why diesel generators drain income.
Communities saw results quickly, which gave the model legitimacy.
The result is a wave of village scale and neighborhood scale solar grids that operate independently from national systems that move far slower.
The next phase of energy growth
The future of solar startup Africa is already visible in early projects.
Smart community grids allow operators to track usage remotely.
Solar bundles pair panels with fridges, pumps, or tools to raise income.
Energy payments evolve into financial identities that help residents access credit.
Local assembly plants begin to keep more of the value inside the region.
AI supports maintenance and prediction quietly in the background instead of becoming the primary story.
These developments point toward a world where reliable power is not something communities hope for. It is something they build and own.
Why the rest of the world should pay attention
The rise of solar startup Africa offers a model for regions with weak infrastructure. It demonstrates that large grids are not the only path to progress. It shows that communities can build upward using energy systems designed for their own needs rather than imported assumptions.
This is progress without permission. It is change without waiting for someone else to deliver it.
Related Stories on Laterstack:
Startup CEO accidentally names new project after famous OnlyFans creator
How startups can attract top talent without big tech salaries
A rare breach inside China’s cyber ecosystem has exposed a detailed archive of hacking tools, stolen data, and government linked operations. Around 12,000 internal documents from KnownSec, a long running contractor for state agencies, appeared online this week and quickly drew global attention.
The leak was first posted on a Chinese language blog before spreading to Western researchers. The files outline a catalog of remote access Trojans, data extraction programs, and surveillance utilities. They also include a list of more than 80 targets the contractor claims to have infiltrated.
Among the reported data sets are 95 GB of Indian immigration records, three terabytes of call logs from South Korean telecom provider LG U Plus, and hundreds of gigabytes of planning data from Taiwan. The documents also appear to reference direct contracts between KnownSec and Chinese government bodies, removing any ambiguity about who the work served.
An uncommon look inside a closed system
China’s intelligence network has avoided anything resembling a Snowden style exposure. For analysts, any glimpse into the tools and priorities of its contractors is unusual. The KnownSec leak offers evidence of broad regional surveillance and points to an organized system for harvesting and analyzing large data troves.
Researchers say the archive reinforces what many suspected. China continues to rely heavily on private security companies to carry out offensive operations. These firms operate quietly in the background, giving the government distance while providing technical reach.
AI takes a darker turn
The leak follows another notable disclosure this week. Anthropic reported detecting a China linked hacking group using its Claude platform to write malware, analyze stolen files, and prepare intrusion tools. According to the company, the campaign relied on minimal human oversight and attempted to mask its activity by framing all requests as defensive research.
Claude eventually stopped the activity, but not before the group breached four organizations. Even with the low success rate and some hallucinated data, the campaign marks a turning point. State operators are beginning to test how far AI can automate intrusion work.
Growing pressure across the security landscape
The KnownSec leak lands during a year already marked by layoffs, consolidation, and rising tensions in the cybersecurity world. Several major firms have downsized as they shift investment toward automated detection and large scale AI systems.
What remains clear is that governments and contractors are accelerating their offensive capabilities at the same pace. The tools are faster, the targets broader, and the lines between state and private actors increasingly blurred.
For researchers and defenders, the leak is both a warning and an opportunity. It exposes methods that were never meant to be seen and offers a brief window into operations usually sealed behind thick walls.
Related Cybersecurity Stories on Laterstack:
Cybersecurity firm Deepwatch lays off staff as AI reshapes the industry
12 Steps to Better Cybersecurity in 2025
The cybersecurity world just got another wake-up call. Deepwatch, a security company known for its AI-powered detection and response platform, has laid off roughly 80 employees out of its 250-person team. The company says the move is about “accelerating investments in AI and automation.”
For many in the field, it’s a sign of a bigger shift happening across the cybersecurity industry — one where traditional analyst-heavy teams are giving way to AI-driven platforms that promise faster detection and fewer human bottlenecks.
The story behind the layoffs
In an email to TechCrunch, Deepwatch CEO John DiLullo explained that the company is “aligning our organization to accelerate our significant investments in AI and automation.”
Employees inside the company painted a more complicated picture. One current staffer told TechCrunch that the layoffs “feel like an AI reshuffle more than a clear product plan.” The move comes as Deepwatch develops new “agentic AI” features that supposedly make its security platform more autonomous though insiders say details are still unclear.
Eight former employees have already announced their departures on LinkedIn. The total reduction, between 60 and 80 workers, represents a major downsizing for a company that had been growing steadily over the last few years.
A broader cybersecurity reset
Deepwatch isn’t alone. 2025 has seen a string of cybersecurity layoffs, even as demand for threat defense keeps rising.
Earlier this year, CrowdStrike cut about 5% of its workforce, despite posting record revenue and cash flow. Other firms, including Deep Instinct, Otorio, ActiveFence, SkyBox Security, and Sophos, have also reduced headcount.
At first glance, that might sound like an industry contraction. But the deeper story is a realignment. Security platforms are increasingly being rebuilt with AI automation at the core, aiming to do what used to require entire SOC teams.
The pitch is simple: AI models can flag anomalies faster, handle repetitive triage work, and even automate response playbooks. That means fewer humans, more software, and a new kind of efficiency that’s starting to define the next phase of cybersecurity.
What it means for the future of cybersecurity work
The shift comes with big questions for cybersecurity professionals. What happens when AI becomes the analyst?
For now, most companies still rely on human oversight, but the ratio is changing. Fewer people are being hired to monitor dashboards. More are being asked to train AI models, review outputs, and handle only the complex or ambiguous cases.
In the near future, the cybersecurity job market may look less like traditional operations centers and more like data science and model management teams. The people who stay will be those who understand both domains: security and machine learning.
This mirrors what’s happening in other tech sectors, where AI adoption doesn’t eliminate roles immediately it reshapes them.
The bigger picture
If this sounds familiar, it’s because this same pattern has played out before, when cloud automation replaced on-prem infrastructure jobs, and when DevOps merged development and operations.
The cybersecurity industry is now undergoing its own version of that shift. Deepwatch’s layoffs aren’t an isolated event; they’re a preview of what happens when security meets automation at scale.
Over the next few years, expect more platforms that market themselves as “AI-native,” fewer manual SOC workflows, and a growing divide between traditional analysts and the next generation of cyber-AI engineers.
Related Laterstack Reads
If you want to go deeper on how AI is reshaping the cybersecurity landscape, check out these related Laterstack stories:
Congressional Budget Office cyberattack raises alarm
12 Steps to Better Cybersecurity in 2025
Quantum computing has quietly stepped up again. IBM followed through on promises made back in June, debuting two new quantum processors that mark a leap toward error-corrected systems. Meanwhile, IonQ and Quantum Art made their own announcements that signal how quickly quantum innovation is moving from the lab into something that looks more like the next computing industry.
IBM’s new Quantum Loon and Nighthawk chips
IBM confirmed the large-scale manufacturing of its Quantum Loon processor, the company’s new architecture for logical qubits. Loon represents a major shift in IBM’s hardware strategy, moving to a square grid layout that connects each qubit to four neighbors instead of the older hex pattern. This setup allows for faster operations and a smoother path toward error-corrected quantum computing.
The second chip, Nighthawk, uses the same grid design but without Loon’s long-distance connections. It’s focused on improving error rates so researchers can begin testing algorithms for quantum advantage, cases where quantum systems outperform classical ones.
IBM also launched a public GitHub repository that lets researchers share performance data across classical and quantum algorithms. It’s part of a broader effort to track where quantum hardware actually provides a measurable edge.
IonQ’s record-breaking error rate
While IBM was refining architecture, IonQ made headlines with a new record-low error rate for two-qubit gates achieving over 99.99 percent fidelity. That number may sound small, but in quantum computing it’s enormous. Each percentage point gained means fewer hardware qubits are needed to stabilize a system, which brings practical quantum hardware closer to reality.
IonQ’s method builds on technology acquired from Oxford Ionics, which used electromagnetic fields instead of lasers to handle qubit operations. The breakthrough reduces the time required for cooling ions between operations, allowing the entire machine to run faster while maintaining stability.
Quantum Art and Nvidia team up
The final announcement came from Quantum Art, which revealed a partnership with Nvidia to create a more efficient compiler for its trapped-ion systems. While Nvidia isn’t directly building quantum chips, it’s using GPUs to model, optimize, and support the computations quantum hardware needs.
Quantum Art’s design is unusual instead of performing operations on one or two qubits at a time, its system groups large clusters of ions into what it calls quantum cores. Each core operates on many qubits at once, using laser “pins” to isolate and move groups efficiently. The result could be multicore quantum computing, a concept that mirrors classical chip architecture.
FULL ARTICLE HERE
This move also reflects a growing trend, traditional computing giants like Nvidia are embedding themselves deeper into the quantum ecosystem, preparing for a hybrid future where quantum and GPU systems coexist.
The next stage of the race
Taken together, these updates show that 2025 is shaping up to be a year of practical momentum for quantum hardware. IBM is building the foundation for scalable, error-corrected machines. IonQ is proving the physics can handle real-world precision. And Quantum Art is testing entirely new models for how quantum cores might work in the future.
If these companies keep their current pace, the next few years might finally deliver what the field has promised for decades: quantum computing that actually does something classical machines cannot.
Related Laterstack Reads
For more context on the evolving quantum race:
Salt Unlocks Metallic Nanotubes for Quantum and High Speed Tech
Will quantum be bigger than AI?
– John Timmer Senior Science Editor –
In one of the more bizarre startup moments of the year, Inference CEO Sam Hogan accidentally gave his company’s newest project the same name as a famous OnlyFans creator and Substack researcher.
The initiative was called Project AELLA, part of an open science push designed to make research more accessible using structured summaries generated by large language models. Within hours of the launch, Hogan learned that “Aella” wasn’t just a cool Greek name, it was also the online identity of a high-profile figure in tech and adult content circles.
The mixup that went viral
After the announcement, Aella herself noticed and responded with a simple post: “Lmfao.”
Hogan quickly replied that he “didn’t know who you were until today,” before changing the project’s name to Project OSSAS. The exchange went viral on X, drawing attention from investors, tech founders, and fans of Aella’s work alike.
Aella, who once ranked in the top 0.04 percent of OnlyFans creators, now focuses most of her time on data-driven research about relationships and behavior through her Substack, Knowingless. Her work has gained recognition from tech figures such as Marc Andreessen, who publicly called one of her theories “fantastic.”
When brand naming meets the internet
The situation highlighted how fragile modern branding can be in an era where every name has an online footprint. A single Google search oversight turned into a viral moment connecting the worlds of adult content, tech startups, and venture capital.
To Hogan’s credit, he handled it with humor and transparency. After learning about Aella’s reputation, he even proposed potential collaboration, suggesting data visualizations for her research. Aella responded enthusiastically.
The two exchanged messages that ended on a friendly note and a possible idea for a joint project.
Inference official website
Aella’s Substack “Knowingless”
What it means for founders
The incident became an instant case study in startup brand due diligence. In a world where personal brands often overlap with company names, founders are realizing they need to check not just trademark databases but also social media ecosystems.
It also revealed how online culture now blends effortlessly with venture-backed innovation. Aella and Hogan, despite their very different backgrounds, share a fascination with human data and digital systems.
Hogan’s startup, Inference, recently closed an 11.8 million dollar seed round led by Multicoin Capital and Andreessen Horowitz, describing itself as “the world’s largest GPU cluster for model inference.”
What started as an embarrassing name mix-up could end as a collaboration between two of the most unexpected voices in tech.
For more on startup culture:
How startups can attract top talent without big tech salaries
and
Sizable Energy wants to store electricity in the ocean
The UC San Diego academic decline report just dropped, and the numbers are jaw-dropping. Between 2020 and 2025, the number of freshmen with math skills below middle school level exploded from around 30 to 921. That’s one in eight new students entering one of California’s top universities unable to perform basic arithmetic or work with fractions.
The numbers that stunned UC San Diego
The report, compiled by the Senate-Administration Working Group on Admissions, paints a sharp picture of academic erosion. Faculty members in the Mathematics Department redesigned their remedial program this year to focus entirely on elementary and middle school concepts after realizing incoming students struggled with the fundamentals taught in grades one through eight.
It didn’t stop with math. Nearly one in five first-year students needed remedial writing instruction in 2024, levels not seen since before the pandemic. Professors across departments are reporting that students have increasing trouble reading and analyzing long or complex texts.
What caused the UC San Diego academic decline?
The study links the deterioration to a perfect storm of factors:
Remote learning during the COVID-19 pandemic disrupted core skill development.
The UC system’s removal of SAT and ACT testing in 2021 made it harder to measure academic readiness.
Grade inflation in high schools created misleading transcripts.
UC San Diego expanded enrollment from under-resourced schools, doubling LCFF+ admits between 2022 and 2024.
Combined, these changes created what faculty now describe as “a silent collapse in academic foundations.”
Why this matters for the future of higher education
The report warns that admitting large numbers of underprepared students can harm both those students and the university’s learning environment. Faculty fear that remedial overload could drain limited resources, forcing departments to divert funding from research to basic skill recovery.
The UC San Diego academic decline also raises broader questions for universities nationwide. How do you maintain standards in an era of test-optional admissions and algorithm-driven grading systems? Could predictive models built with AI and big data be used to spot underprepared applicants early or would that create new biases?
UC San Diego Senate Report PDF
University of California Testing Policy Overview
The big picture
Education experts argue that the next phase of innovation must focus on adaptive learning systems, using data to personalize remediation at scale. Whether through advanced tutoring models, machine-assisted placement systems, or AI-powered learning analytics, the challenge now is rebuilding the foundation that technology helped erode.
For more on how education and tech are colliding:
and
The world’s largest illegal sports streaming platform just collapsed under its own crypto trail. Streameast, with more than 130 million global users, was dismantled after investigators traced its blockchain footprint through offshore accounts, digital wallets, and shell companies across Egypt and Dubai.
The investigation was led by the Alliance for Creativity and Entertainment alongside Egyptian authorities. What began as a piracy probe turned into one of the largest blockchain-based financial crackdowns in years.
How blockchain broke the blackout
For nearly two years, Streameast operated as the underground home for free sports streams, covering everything from Premier League to Champions League matches. Behind the curtain, the platform used cryptocurrency to mask payments, reroute ad revenue, and obscure its true operators.
When investigators followed the digital breadcrumbs, the illusion of anonymity fell apart. “The funds left a fingerprint,” said investigator Dani Bacsa. “They used crypto to hide, but the blockchain records everything.”
Each transfer told part of the story. Wallets led to shell companies, which led to bank accounts, which led to real people. Authorities seized cash, gold, laptops, and crypto wallets holding more than £450,000. Ad revenue from malware-driven popups reached almost £7.6 million.
The blockchain that never forgets
The raid took place west of Cairo and resulted in two arrests. Those individuals allegedly managed the Streameast empire’s infrastructure and crypto movement. The operation revealed how decentralized finance tools once meant for innovation had been repurposed to power cybercrime.
Anti-piracy experts now see this as a turning point. Illegal streaming has evolved from sketchy file hosting to sophisticated, blockchain-powered ecosystems that run like startups. But the very tech that gave them cover also sealed their fate. Blockchain transparency meant investigators could follow every token, no matter how deep it was buried.
The ripple effect
Since the takedown, dozens of Streameast copycats have surfaced across Europe and Asia. Each new site recycles the brand name, hoping to capture the same loyal traffic. Investigators are already tracking new wallet activity tied to these offshoots.
The Streameast story is no longer just about piracy. It is about the convergence of digital entertainment, decentralized finance, and cybercrime. And how the same code that enables freedom can also expose everything.
For deeper insights, explore two related stories on Laterstack:
Crypto markets stumble as AI bets and Wall Street confidence fade and Crypto.com ruling reveals blockchain’s legal gray zone.
A former employee has accused Bank of America of quietly taking millions in unpaid labor from its remote workforce. The alleged culprit? The time it takes to start a computer.
In a federal class action filed in North Carolina, former Business Analyst Tava Martin claims that hundreds of hourly workers were required to spend up to 30 minutes each day booting up complex systems before their paid shifts began. Those minutes, she says, were never counted or compensated.
Martin describes mornings that began long before the clock started. Employees had to power on company computers, wait for Windows to load, request a security token from their phones, log into a virtual private network, and open a maze of internal apps before they were “call ready.” According to the suit, Bank of America enforced a strict policy that required staff to be available for customer calls the instant their shift began.
Work before work
The filing paints a picture of hundreds of remote analysts spending up to half an hour preparing for their day unpaid. During lunch breaks, the systems would time out, forcing workers to log in again. After hours, they were required to securely shut down each program. It all added up to significant lost time, the suit says.
At Martin’s pay rate of $46 per hour, that missing time could push her into overtime territory. For others in similar roles, the pattern may have lasted for years.
The Department of Labor has already issued guidance stating that tasks like loading computer systems count as part of the workday for hourly employees. Martin argues the bank either ignored that rule or knowingly avoided compliance.
A system bigger than one worker
Bank of America has not commented publicly, but the lawsuit suggests the issue could span hundreds, possibly thousands of employees who worked under the same strict scheduling rules. Many were hired through staffing agencies but followed the bank’s internal policies and timekeeping systems.
The class action seeks back pay for all affected employees, along with penalties and legal fees. If the case is certified, it could become one of the largest remote work wage disputes of the decade.
The lawsuit also hints at a broader shift in how companies handle the hidden labor of remote tech work. When employees spend a chunk of their day waiting for digital tools to load, is that still their time?
For now, the question heads to court. The outcome could change how corporations measure and pay for every click, login, and reboot in the digital workplace.
After months of relentless speculation, the pulse of the crypto market has slowed. Bitcoin’s latest slide toward 100,000 dollars and sharp ETF outflows are signaling a wider pullback from the AI and digital asset frenzy that defined 2025.
Markets did not crash this week, but the tone has shifted. Big tech stocks like Palantir and Oracle took heavy hits, and their losses echoed across the leveraged trades that powered the latest rally. Bitcoin and other major coins fell sharply as retail investors and institutions began scaling back their risk.
Peter Atwater, a behavioral economics professor at the College of William and Mary, called it a confidence break. “AI and crypto live in the same neighborhood of belief,” he said. “When the mood shifts, it hits everything tied to that optimism.”
Retail energy drains from crypto and AI
The retreat is visible in the data. More than 700 million dollars left digital asset ETFs this week, including 600 million from BlackRock’s Bitcoin fund and 370 million from its Ether fund. Solana and Dogecoin products are also down double digits since their launch.
Meanwhile, the Roundhill Meme ETF, marketed as a retail sentiment tracker, is down more than 20 percent just a month after debuting. Indexes that follow speculative tech names and new IPOs also fell hard, with losses not seen since the summer.
Stephen Kolano, chief investment officer at Integrated Partners, said the selloff is not panic but a reset. “The profit taking is coming from trades that ran the most since spring,” he said. “That’s AI, that’s crypto, that’s anything fueled by momentum.”
Bitcoin as a signal
Bitcoin’s 15 percent drop this month has some analysts watching closely. Bloomberg Intelligence’s Eric Balchunas said Bitcoin often acts as an early indicator for shifts in broader market sentiment. “It trades around the clock. It reacts before most other assets do,” he said.
A Citi report noted that large holders, often called whales, have been quietly exiting. That is unusual, since this group tends to ride through downturns. Their selling adds weight to the idea that liquidity and conviction are thinning.
What it means beyond crypto
This is not a collapse, but it is a cooling of risk appetite. Retail traders who flooded into meme stocks and tokenized assets are pulling back. As capital leaves the edges of the market, liquidity tightens and timing begins to matter again.
The total crypto market cap, which peaked at 4.4 trillion dollars in October, has fallen nearly 20 percent. For now, the thrill ride that defined 2025 looks to be slowing.