[Matt Denton]’s SpoolBot is a surprisingly agile remote-controlled robot that doesn’t just repurpose filament spool leftovers. It looks exactly like a 2 kg spool of filament; that’s real filament wound around the outside of the drum. In fact, Spoolie the SpoolBot looks so much like the real thing that [Matt] designed a googly-eye add-on, because the robot is so easily misplaced.

SpoolBot works by rotating its mass around the central hub, which causes it to roll forward or back. Steering is accomplished by tank-style turning of the independent spool ends. While conceptually simple, quite a bit of work is necessary to ensure SpoolBot rolls true, and doesn’t loop itself around inside the shell during maneuvers. Doing that means sensors, and software work.

To that end, a couple of rotary encoders complement the gearmotors and an IMU takes care of overall positional sensing while an ESP32 runs the show. The power supply uses NiMH battery packs, in part for their added weight. Since SpoolBot works by shifting its internal mass, heavier batteries are more effective.

The receiver is a standard RC PWM receiver which means any RC transmitter can be used, but [Matt] shows off a slick one-handed model that not only works well with SpoolBot but tucks neatly into the middle of the spool for storage. Just in case SpoolBot was not hard enough to spot among other filament rolls, we imagine.

The googly-eye add-on solves that, however. They clip to the central hub and so always show “forward” for the robot. They do add quite a bit of personality, as well as a visual indication of the internals’ position relative to the outside.

The GitHub repository and Printables page have all the design files, and the video (embedded just below) shows every piece of the internals.

The kind of hardware available nowadays makes self-balancing devices much more practical and accessible than they ever have been. Really, SpoolBot has quite a lot in common with other self-balancing robots and self-balancing electric vehicles (which are really just larger, ridable self-balancing robots) so there’s plenty of room for experimentation no matter one’s budget or skill level.

For the last six months, enterprises wanting to deploy high quality AI image generation at scale have faced an uncomfortable trade-off: pay premium prices for Google’s Nano Banana Pro model, or settle for cheaper (sometimes free), faster, but noticeably inferior alternatives — especially in terms of enterprise requirements like embedded accurate text, slides, diagrams, and other non aesthetic information.

Today, Google DeepMind is attempting to collapse that gap with the launch of Nano Banana 2 (formally Gemini 3.1 Flash Image) — a model that brings the reasoning, text rendering, and creative control of the Pro tier down to Flash-level speed and pricing.

The release comes just sixteen days after Alibaba’s Qwen team dropped Qwen-Image-2.0, a 7-billion parameter open-weight challenger that many developers argued had already matched Nano Banana Pro’s quality at a fraction of the inference cost.

For IT leaders evaluating image generation pipelines, Nano Banana 2 reframes the decision matrix. The question is no longer whether AI image models are good enough for production — it’s which vendor’s cost curve best fits the workflow.

The production cost problem: why Nano Banana Pro stayed in the sandbox

When Google released Nano Banana Pro in November 2025, built on the Gemini 3 Pro backbone, the developer community was impressed by its visual fidelity and reasoning capabilities.

The model could render accurate text in images, maintain character consistency across multi-turn conversations, and follow complex compositional instructions — all capabilities that previous image generators struggled with.

But Pro-tier pricing created a barrier to deployment at scale. According to Google’s API pricing page, Nano Banana Pro’s image output is priced at $120 per million tokens, working out to roughly $0.134 per generated image at 1K pixel resolution.

For applications generating thousands of images daily — think e-commerce product visualization, marketing asset pipelines, or localized content generation — those costs compound quickly.

Nano Banana 2, built on the Gemini 3.1 Flash backbone, dramatically undercuts that pricing. Flash-tier image output is priced at $60 per million tokens, approximately $0.067 per 1K image per image — roughly 50% cheaper than the Pro model. For enterprises running high-volume image generation workflows, that’s the difference between a proof of concept and a production deployment.

What Nano Banana 2 actually delivers

The model is not simply a cheaper Nano Banana Pro. According to Google DeepMind’s announcement, Nano Banana 2 brings several capabilities that were previously exclusive to the Pro tier while introducing new features of its own.

The headline improvement is text rendering and translation. The model can generate images with accurate, legible text — a historically weak point for AI image generators — and then translate that text into different languages within the same image editing workflow. 

Subject consistency has also improved significantly. Nano Banana 2 can maintain character resemblance across up to five characters and preserve the fidelity of up to 14 reference objects in a single generation workflow.

This enables storyboarding, product photography with multiple SKUs, and brand asset creation where visual continuity matters. Google’s documentation highlights the ability to provide up to 14 different reference images as input, allowing the model to compose scenes incorporating multiple distinct objects or characters from separate sources.

On the technical specification side, the model supports full aspect ratio control, resolutions ranging from 512 pixels up to 4K, and two thinking levels that let developers balance quality against latency.

One notable addition that Nano Banana Pro lacks is an image search tool — the model can perform image searches and use retrieved images as grounding context for generation, expanding its utility for workflows that require visual reference material.

The Qwen-Image-2.0 factor: why Google needed to move fast

Google’s timing is not coincidental. On February 10, Alibaba’s Qwen team released Qwen-Image-2.0, a unified image generation and editing model that immediately drew comparisons to Nano Banana Pro — but with a dramatically smaller footprint.

Qwen-Image-2.0 runs on just 7 billion parameters, down from 20 billion in its predecessor, while unifying text-to-image generation and image editing into a single architecture.

The model generates natively at 2K resolution (2048×2048 pixels), supports prompts up to 1,000 tokens for complex layouts, and ranks at or near the top of AI Arena’s blind human evaluation leaderboard for both generation and editing tasks.

For enterprise buyers, the competitive dynamics are significant. Qwen-Image-2.0’s 7B parameter count means substantially lower inference costs when self-hosted — a critical consideration for organizations with data residency requirements or high-volume workloads.

The Qwen team’s previous model, Qwen-Image v1, was released under Apache 2.0 approximately one month after its initial announcement, and the developer community widely expects the same trajectory for v2.0. If open weights materialize, organizations could run a Nano Banana Pro-competitive image model on their own infrastructure without per-image API charges.

The model’s unified generation-and-editing architecture also simplifies deployment. Rather than chaining separate models for creation and modification — the current industry norm — Qwen-Image-2.0 handles both tasks in a single pass, reducing latency and the quality degradation that occurs when outputs are passed between different systems.

Where Qwen-Image-2.0 currently trails is ecosystem integration. Google’s Nano Banana 2 launches today across the Gemini app, Google Search (AI Mode and Lens), AI Studio, the Gemini API, Google Antigravity, Vertex AI, Google Cloud, and Flow — where it becomes the default image generation model at zero credit cost. That breadth of distribution is difficult for any challenger to replicate, particularly one whose API access is currently limited to Alibaba Cloud’s platform.

What this means for enterprise AI image strategies

The simultaneous availability of Nano Banana 2 and Qwen-Image-2.0 creates a decision framework that IT leaders haven’t had before in the image generation space.

For organizations already embedded in Google’s cloud ecosystem, Nano Banana 2 is the obvious first evaluation. The cost reduction from Pro pricing, combined with native integration across Google’s product surface, makes it the path of least resistance for teams that need production-quality image generation without re-architecting their stack. The model’s text rendering capabilities make it particularly well-suited for marketing asset generation, localization workflows, and any application where legible in-image text is a requirement.

For organizations with data sovereignty concerns, high-volume workloads that make per-image API pricing prohibitive, or a strategic preference for open-weight models, Qwen-Image-2.0 presents a compelling alternative — provided Alibaba follows through on open-weight availability. The model’s smaller parameter count translates to lower GPU requirements for self-hosting, and its unified generation-editing architecture reduces pipeline complexity.

The wild card is Nano Banana Pro itself, which isn’t going away. Google AI Pro and Ultra subscribers retain access to the Pro model for specialized tasks, accessible via the regeneration menu in the Gemini app. For use cases demanding maximum visual fidelity and creative reasoning — think high-end creative campaigns or applications where every image needs to look bespoke — Pro remains the ceiling.

The provenance layer: a quiet but important enterprise differentiator

Buried in Google’s announcement is a detail that may matter more to enterprise legal and compliance teams than any quality benchmark: provenance tooling. Nano Banana 2 ships with SynthID watermarking — Google’s AI-generated content identification technology — coupled with C2PA Content Credentials, the cross-industry standard for content authenticity metadata.

Google reports that since launching SynthID verification in the Gemini app last November, the feature has been used over 20 million times to identify AI-generated images, video, and audio. C2PA verification is coming to the Gemini app soon as well.

For enterprises operating in regulated industries or jurisdictions with emerging AI transparency requirements, baked-in provenance is no longer optional. It’s a compliance checkbox — and one that self-hosted open-weight alternatives like Qwen-Image-2.0 don’t natively provide.

The bottom line

Nano Banana 2 doesn’t represent a generational leap in image generation quality. What it represents is the maturation of AI image generation from a creative novelty into a production-ready infrastructure component. By collapsing the cost and speed gap between Flash and Pro tiers while retaining the reasoning and text rendering capabilities that make these models useful for actual business workflows, Google is making a calculated bet: the next wave of enterprise AI image adoption will be driven not by the models that produce the most beautiful images, but by the ones that produce good-enough images fast enough and cheaply enough to deploy at scale.

With Qwen-Image-2.0 pushing from the open-weight flank and Nano Banana Pro holding the quality ceiling, Nano Banana 2 occupies exactly the middle ground where most enterprise workloads actually live. For IT decision-makers who’ve been waiting for the cost curve to bend, it just did.

DIY store chain ManoMano is notifying customers of a data breach that was caused by hackers compromising a third-party service provider.

The company confirmed to BleepingComputer that it learned of the hack in January 2026. An investigation into the incident determined that 38 million individuals are affected.

“We can confirm that ManoMano has recently notified customers about a security incident involving one of our third-party customer service providers (a subcontractor),” the company told BleepingComputer.

“In January 2026, we identified unauthorized access linked to this provider, which resulted in the unauthorized extraction of certain personal data associated with customer accounts and customer service interactions.”

ManoMano is a French e-commerce firm operating an online marketplace specializing in DIY, home improvement, gardening, and related products. It operates in France, Belgium, Spain, Italy, Germany, and the United Kingdom, and its e-stores reportedly have 50 million unique visitors per month.

Earlier this month, someone using the alias “Indra” claimed the ManoMano attack on a hacker forum, alleging that they were holding details on 37.8 million user accounts, as well as thousands of support tickets and attachments.

According to unconfirmed reports, the compromised organization was a Tunis-based customer support service provider that suffered a Zendesk breach.

Cybersecurity firm Hackmanac posted that ManoMano started notifying customers this week that their data had been stolen.

A spokesperson of ManoMano explained to BleepingComputer that the exposed information varies per individual, depending on the type of interactions they had with the platform. Exposed data types include:

• Full name
• Email address
• Phone number
• Customer service communications

ManoMano emphasizes that no account passwords were accessed and that no data modifications occurred on the company’s systems.

“Upon discovery, we took immediate steps to secure our environment, including disabling the relevant access, revoking the subcontractor’s access to customer data, and strengthening access controls and monitoring,” said a ManoMano spokesperson.

“We also notified the relevant authorities, including the CNIL and ANSSI, and informed impacted customers with guidance to remain vigilant against phishing and social engineering attempts.”

The notification sample ManoMano shared with BleepingComputer contains recommendations for customers, including verifying incoming communications and sender identity, monitoring bank accounts for fraudulent transactions, and avoiding clicking on suspicious links or downloading email attachments.

ManoMano notes that the investigation is ongoing and that they cannot share additional technical details at this stage.

Modern IT infrastructure moves faster than manual workflows can handle.

In this new Tines guide, learn how your team can reduce hidden manual delays, improve reliability through automated response, and build and scale intelligent workflows on top of tools you already use.

Advertisement

Get the guide

The raise will fund Allica’s expansion plans outside the UK.

London’s Allica Bank has joined the European unicorn league with a $155m Series D raise that values the company at $1.2bn. New investors Ventura Capital, GLG and Sona AM, and existing investors TCV and Blue Owl, took part in the round.

Allica’s digital banking services are geared toward small, and medium enterprises (SMEs), currently offering services to more than 30,000 SMEs across the UK. The 2011-founded company has been named the fastest growing technology company in the UK by Deloitte in both 2023 and 2024.

According to Allica, SMEs are an underserved customer base in the fintech market.

The digital bank said the capital will enable continued investment into AI to develop newer lending mechanisms for SMEs. The infusion will also fund the company’s expansion plans outside the UK.

“We’re building the category defining digital bank for established SMBs, and are excited to be taking our proprietary platform into new markets,” said Allica CEO Richard Davies.

“This Series D investment is a major vote of confidence in Allica’s strategy and performance.”

Allica joins the fintech unicorn league alongside companies such as the UK’s Cleo and Denmark’s Flatpay. It was one of the new firms invited to participate in a new UK scale-up unit designed to support fast-growing, innovative financial services firms to scale and create high-skilled jobs in the country.

“Allica is a world class business that is executing exceptionally well in a large, underserved market,” said Mo El Husseiny, the managing partner of Ventura Capital.

Earlier this week, Irish fintech Stripe, founded by brothers John and Patrick Collison, announced it had hit a $159bn valuation – up from $106bn a year ago. Bloomberg reported shortly after that Stripe is considering acquiring PayPal, which has been struggling to grow for a few years.

Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.

Robert Walters’ data suggests that microshifting could be the next evolution of flexible working.

The global pandemic and AI wave have in many ways altered how modern-day employees approach working life. Remote and hybrid opportunities have given professionals greater control over their hours, creating a stronger sense of work-life balance. 

New research from recruitment platform Robert Walters indicates that there may now be a new working trend impacting the professional space: microshifting. This is defined as an approach to hours which sees the traditional working day split into shorter blocks of time, based around a professional’s personal obligations or energy peaks.

Robert Walters collected data from 850 white-collar, full-time, permanent professionals based in Ireland between December of 2025 and January of this year. What the report discovered is that more than half (59pc) of contributing Irish employees want their place of employment to adopt a microshifting schedule. 

Some respondents have noted that available flexible working models are not operating as efficiently or effectively as they could. Only 36pc of Irish workers stated that their company’s current policy is fit for purpose.

“Despite years of debate around flexible working, many organisations still measure commitment by visibility rather than results,” said Suzanne Feeney, the country manager at Robert Walters Ireland. “Trends like microshifting will continue to emerge as professionals seek flexibility that actually works, instead of policies that look progressive on paper but fail in practice.”

Shifting values

Despite concerns that flexible working results in lower engagement, Robert Walters’ report highlights that flexibility in hours can lead to increased office attendance. Of those who contributed their data, 42pc responded that switching to a microshifting approach would encourage them to increase the number of days they spend in the office each week.

Feeney said: “Offering flexible hours may feel counterproductive for employers looking to increase office attendance. Yet, a more adaptable schedule, without the pressure of rush hour commutes or staying at their desk all day, could motivate professionals to attend the office more frequently.”

More than a third of line managers (37pc) surveyed by Robert Walters thought microshifting could help improve engagement within their teams, while a further 44pc said they were open to testing it out.

Feeney added: “Microshifting is a more transparent version of the unofficial flexible working arrangements that already exist in many organisations. For managers and senior leaders, the question is whether it should be governed by trust and outcomes or quietly negotiated between colleagues.”

According to the data, however, worries persist, as nearly 50pc of surveyed managers expressed concerns that microshifting could result in higher instances of “quiet quitting“ and “slacking”. 

Feeney said: “While fears of microshifting fuelling disengagement are justified, the reality is that rigid working patterns are already pushing professionals to seek workarounds.

“Our findings suggest that when expectations are clear and performance is measured by results rather than visibility, microshifting has the potential to increase engagement, accountability and even time spent in the office.”

Additional research published today (26 February) by CPL also explored how organisations have to do more to encourage key talent loyalty. CPL’s Salary Guide for Ireland 2026 found that while compensation and benefits continue to be the top priority for 35pc of contributing employees, 24pc of professionals said that leadership and culture are the most important factors to consider when choosing an employer.

CPL’s research also found that flexible working has evolved from a perk to a critical component of employee packages, ranking as the second most important benefit overall among contributing participants. 

Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.

Google API keys for services like Maps embedded in accessible client-side code could be used to authenticate to the Gemini AI assistant and access private data.

Researchers found nearly 3,000 such keys while scanning internet pages from organizations in various sectors, and even from Google.

The problem occurred when Google introduced its Gemini assistant, and developers started enabling the LLM API in projects. Before this, Google Cloud API keys were not considered sensitive data and could be exposed online without risk.

Developers can use API keys to extend functionality in a project, such as loading Maps on a website to share a location, for YouTube embeds, usage tracking, or Firebase services.

When Gemini was introduced, Google Cloud API keys also acted as authentication credentials for Google’s AI assistant.

Researchers at TruffleSecurity discovered the issue and warned that attackers could copy the API key from a website’s page source and access private data available through the Gemini API service.

Since using the Gemini API is not free, an attacker could leverage the access and make API calls for their benefit.

“Depending on the model and context window, a threat actor maxing out API calls could generate thousands of dollars in charges per day on a single victim account,” Truffle Security says.

The researchers warn that these API keys have been sitting exposed in public JavaScript code for years, and now they have suddenly gained more dangerous privileges without anyone noticing.

TruffleSecurity scanned the November 2025 Common Crawl dataset, a representative snapshot of a large swath of the most popular sites, and found more than 2,800 live Google API keys publicly exposed in their code.

According to the researchers, some of the keys were used by major financial institutions, security companies, and recruiting firms. They reported the problem to Google, providing samples from its infrastructure.

In one case, an API key acting just as an identifier was deployed since at least February 2023 and was embedded in the page source of a Google product’s public-facing website.

Truffle Security tested the key by calling the Gemini API’s /models endpoint and listing available models.

The researchers informed Google of the problem last year on November 21.  After a long exchange, Google classified the flaw as “single-service privilege escalation” on January 13, 2026.

In a statement for BleepingComputer, Google says that it is aware of the report and has “worked with the researchers to address the issue.”

“We have already implemented proactive measures to detect and block leaked API keys that attempt to access the Gemini API,” a Google spokesperson told BleepingComputer.

Google stated that new AI Studio keys will default to Gemini-only scope, leaked API keys will be blocked from accessing Gemini, and proactive notifications will be sent when leaks are detected.

Developers should check whether Gemini (Generative Language API) is enabled on their projects and audit all API keys in their environment to determine if any are publicly exposed, and rotate them immediately.

The researchers also suggest using the TruffleHog open-source tool to detect live, exposed keys in code and repositories.

Modern IT infrastructure moves faster than manual workflows can handle.

In this new Tines guide, learn how your team can reduce hidden manual delays, improve reliability through automated response, and build and scale intelligent workflows on top of tools you already use.

Get the guide

While Panasonic has recently moved its TV production over to Skyworth, it is taking the opposite approach with Technics. Instead of scaling back, the company is reinforcing its in-house engineering and manufacturing commitment to a brand that remains central to its premium audio business. The new Technics SL-1500CS turntable makes that strategy clear. It replaces the widely adopted SL-1500C (without the ‘s”) and introduces Delta Sigma Drive motor control technology to its direct-drive platform, aiming to further reduce motor vibration and improve rotational accuracy.

Technics continues to produce some of the best-selling turntables in the world, supported by a growing lineup of wireless speakers, network amplifiers, and receivers that compete on measurable performance and build quality, not just brand recognition.

At a time when parts of Panasonic’s business are being outsourced, Technics is being strengthened. That tells you exactly where the company believes the long-term value, and the quality still resides.

Delta Sigma Drive Motor Technology

Delta Sigma Drive first appeared in 2023 with the Technics SL-1200GR2 and Technics SL-1210GR2, marking a meaningful refinement in Technics’ coreless direct-drive architecture. The technology was expanded in 2024 with the Technics SL-1300G, where it was paired with a higher-grade twin-rotor, single-stator motor design. It also found its way into the limited-edition Technics SL-1200GME, elevating the SL-1200G platform with a collector-focused release that quickly drew strong demand.

Now, Technics is bringing Delta Sigma Drive to a more accessible price tier. By adding the technology to the SL-1500 platform, the new Technics SL-1500CS extends motor-control refinements previously reserved for higher-end models into one of the company’s most popular turntable lines.

According to Technics, Delta Sigma Drive represents one of its most significant proprietary developments since the brand returned to the market:

“Delta Sigma Drive technology is one of our most innovative proprietary technological achievements since Technics’ reintroduction to the market. The rotational accuracy of our direct drive turntables, widely enhancing their faithful, audiophile vinyl sound reproduction, is something very special and instantly noticeable. By adding this unique technology to the SL-1500 platform, our range of turntables will be enriched by a truly audiophile new model toward the entry level of our lineup. We are confident the SL-1500CS will mark a new ‘best in its class’ standard.”

SL-1500CS Motor Construction and Control Technology

The SL-1500C’s iron-coreless, direct-drive motor with a single-rotor/single-stator design already minimizes vibration and effectively eliminates cogging. With the addition of Delta Sigma Drive technology, the Technics SL-1500CS takes that refinement further by optimizing the motor’s sine-wave drive signal for greater rotational precision. The result is even smoother operation, with motor vibration components reduced to an absolute minimum.

Mechanical Design

The overall mechanical design of the SL-1500CS closely follows the SL-1500C. Its aluminum die-cast chassis is bonded to a second layer made from ABS resin reinforced with glass fiber, creating a two-layer construction that improves rigidity while enhancing vibration damping.

The feet use a spring-and-rubber insulator system tuned for optimal frequency control. This design helps maintain sound quality by resisting acoustic feedback and reducing the impact of external vibrations, even at higher playback levels.

Tonearm 

The SL-1500CS features a static-balance, universal S-shaped tonearm. The tonearm tube is constructed from lightweight, high-rigidity aluminum to maintain strength without adding unnecessary mass.

Its gimbal suspension uses a precision-machined housing and high-tolerance bearings to deliver high initial sensitivity and stable tracking performance. The S-shaped aluminum arm maintains balanced mass on either side of the pivot point, supporting accurate groove tracing and consistent playback.

SL-1500CS Features Automatic Lift and Built-In Phono Stage

Like the SL-1500C, the Technics SL-1500CS includes a practical feature set designed to simplify setup and everyday use.

Phono Equalizer: The built-in phono stage supports MM cartridges, allowing the SL-1500CS to connect directly to amplifiers or powered speakers that do not offer a dedicated phono input. Both Line-level and traditional phono outputs are provided, ensuring compatibility with a wide range of audio systems.

Phono Cartridge: The SL-1500CS ships with a pre-mounted MM cartridge for immediate use. The included universal headshell comes fitted with an Ortofon 2M Red ($109 at Amazon), allowing for straightforward plug-and-play setup right out of the box.

Auto Tonearm Lift: The SL-1500CS provides automatic tonearm lift. When the tonearm reaches the end of the record, the auto lifter automatically raises the tonearm. This prevents unnecessary wear on your valuable vinyl records and stylus.

New Finish: The SL-1500CS is offered in a distinctive metallic grey finish that sets it apart from the standard SL-1500C. The clean, minimal industrial design gives it a technically refined appearance that integrates easily with both black and silver audio components.

Packaging

To reduce environmental impact, the new packaging for the SL-1500CS is free of Expanded PolyStyrene (EPS). Instead, it is made of smartly shaped cardboard, maintaining safe protection of the product, also including a cardboard box for the accessories, allowing ease of removal and storage. 

Comparison

The Bottom Line 

By bringing Delta Sigma Drive to the SL-1500 platform, Technics is adding measurable motor-control refinement to one of its most accessible direct-drive models. The SL-1500CS stands out because it pairs advanced rotational accuracy with a proven iron-coreless motor design that already minimizes vibration and eliminates cogging.

It’s built for listeners who want precision without moving into the price tier of the G or GR series. The static-balance S-shaped tonearm and built-in MM phono stage make it easy to get started with the included Ortofon 2M Red, which is a solid entry-level performer. From there, upgrading is straightforward to something like a 2M Blue, 2M Bronze, or a more ambitious cartridge is an easy swap.

And if you prefer to run the line output into a higher-end external phono preamp, the platform is more than capable of supporting that move. The Technics SL-1500CS doesn’t just replace the SL-1500C, it refines it, giving serious vinyl listeners a clear path to better performance without complicating the experience.

Price & Availability

The Technics SL-1500CS turntable is expected to be available for $1,499 in March 2026 via Authorized Technics Retailers.

The Technics SL-1500C seems to be in limited supply, but is currently available for $1,499 at Audio Advice while supplies last.

For more information: us.technics.com

Related Reading:

Block is the latest business to announce layoffs, with the operator of payment platforms Square and Cash App opting to cut jobs in favor of using more AI tools. The financial tech company, helmed by Twitter founder Jack Dorsey, is slashing its current staff of 10,000 to “just under 6,000.” CNBC highlighted a letter Block sent to shareholders announcing the decision to nearly halve its workforce. According to the message from Dorsey:

“The core thesis is simple. Intelligence tools have changed what it means to build and run a company. We’re already seeing it internally. A significantly smaller team, using the tools we’re building, can do more and do it better. And intelligence tool capabilities are compounding faster every week.”

We learned last year that Block had developed an AI agent called “codename goose” for interacting with LLMs. Leadership is clearly putting high expectations on that project and any other in-house tools to fill the shoes of thousands. “intelligence will be at the core of how the entire company works. How we make decisions, how we build trust and manage risk, how we build products, and how we serve customers,” the shareholder letter states.

Block also reported its latest financial results today. It finished the 2025 financial year with operating income (profit after expenses) of $1.71 billion.

This isn’t the first time the fintech company has made deep cuts in its employee count. Layoffs numbering about 1,000 were rumored both in 2024 and 2025.