Presented by F5

As enterprises pour billions into GPU infrastructure for AI workloads, many are discovering that their expensive compute resources sit idle far more than expected. The culprit isn’t the hardware. It’s the often-invisible data delivery layer between storage and compute that’s starving GPUs of the information they need.

“While people are focusing their attention, justifiably so, on GPUs, because they’re very significant investments, those are rarely the limiting factor,” says Mark Menger, solutions architect at F5. “They’re capable of more work. They’re waiting on data.”

AI performance increasingly depends on an independent, programmable control point between AI frameworks and object storage — one that most enterprises haven’t deliberately architected. As AI workloads scale, bottlenecks and instability happens when AI frameworks are tightly coupled to specific storage endpoints during scaling events, failures, and cloud transitions.

“Traditional storage access patterns were not designed for highly parallel, bursty, multi-consumer AI workloads,” says Maggie Stringfellow, VP, product management – BIG-IP. “Efficient AI data movement requires a distinct data delivery layer designed to abstract, optimize, and secure data flows independently of storage systems, because GPU economics make inefficiency immediately visible and expensive.”

Why AI workloads overwhelm object storage

These bidirectional patterns include massive ingestion from continuous data capture, simulation output, and model checkpoints. Combined with read-intensive training and inference workloads, they stress the tightly coupled infrastructure upon which the storage systems are reliant.

While storage vendors have done significant work in scaling the data throughput into and out of their systems, that focus on throughput alone creates knock-on effects across the switching, traffic management, and security layers coupled to storage.

The stress on S3-compatible systems from AI workloads is multidimensional and differs significantly from traditional application patterns. It’s less about raw throughput and more about concurrency, metadata pressure, and fan-out considerations. Training and fine-tuning create particularly challenging patterns, like massive parallel reads of small to mid-size objects. These workloads also involve repeated passes through training data across epochs and periodic checkpoint write bursts.

RAG workloads introduce their own complexity through request amplification. A single request can fan out into dozens or hundreds of additional data chunks, cascading into further detail, related chunks, and more complex documents. The stress concentration is less about capacity, storage system speed, and more about request management and traffic shaping.

The risks of tightly coupling AI frameworks to storage

When AI frameworks connect directly to storage endpoints without an intermediate delivery layer, operational fragility compounds quickly during scaling events, failures, and cloud transitions, which can have major consequences.

“Any instability in the storage service now has an uncontained blast radius,” Menger says. “Anything here becomes a system failure, not a storage failure. Or frankly, aberrant behavior in one application can have knock-on effects to all consumers of that storage service.”

Menger describes a pattern he’s seen with three different customers, where tight coupling cascaded into complete system failures.

“We see large training or fine-tuning workloads overwhelm the storage infrastructure, and the storage infrastructure goes down,” he explains. “At that scale, the recovery is never measured in seconds. Minutes if you’re lucky. Usually hours. The GPUs are now not being fed. They’re starved for data. These high value resources, for that entire time the system is down, are negative ROI.”

How an independent data delivery layer improves GPU utilization and stability

The financial impact of introducing an independent data delivery layer extends beyond preventing catastrophic failures.

Decoupling allows data access to be optimized independently of storage hardware, improving GPU utilization by reducing idle time and contention while improving cost predictability and system performance as scale increases, Stringfellow says.

“It enables intelligent caching, traffic shaping, and protocol optimization closer to compute, which lowers cloud egress and storage amplification costs,” she explains. “Operationally, this isolation protects storage systems from unbounded AI access patterns, resulting in more predictable cost behavior and stable performance under growth and variability.”

Using a programmable control point between compute and storage

F5’s answer is to position its Application Delivery and Security Platform, powered by BIG-IP, as a “storage front door” that provides health-aware routing, hotspot avoidance, policy enforcement, and security controls without requiring application rewrites.

“Introducing a delivery tier in between compute and storage helps define boundaries of accountability,” Menger says. “Compute is about execution. Storage is about durability. Delivery is about reliability.”

The programmable control point, which uses event-based, conditional logic rather than generative AI, enables intelligent traffic management that goes beyond simple load balancing. Routing decisions are based on real backend health, using intelligent health awareness to detect early signs of trouble. This includes monitoring leading indicators of trouble. And when problems emerge, the system can isolate misbehaving components without taking down the entire service.

“An independent, programmable data delivery layer becomes necessary because it allows policy, optimization, security, and traffic control to be applied uniformly across both ingestion and consumption paths without modifying storage systems or AI frameworks,” Stringfellow says. “By decoupling data access from storage implementation, organizations can safely absorb bursty writes, optimize reads, and protect backend systems from unbounded AI access patterns.”

Handling security issues in AI data delivery

AI isn’t just pushing storage teams on throughput, it’s forcing them to treat data movement as both a performance and security problem, Stringfellow says. Security can no longer be assumed simply because data sits deep in the data center. AI introduces automated, high-volume access patterns that must be authenticated, encrypted, and governed at speed. That’s where F5 BIG-IP comes into play.

“F5 BIG-IP sits directly in the AI data path to deliver high-throughput access to object storage while enforcing policy, inspecting traffic, and making payload-informed traffic management decisions,” Stringfellow says. “Feeding GPUs quickly is necessary, but not sufficient; storage teams now need confidence that AI data flows are optimized, controlled, and secure.”

Why data delivery will define AI scalability

Looking ahead, the requirements for data delivery will only intensify, Stringfellow says.

“AI data delivery will shift from bulk optimization toward real-time, policy-driven data orchestration across distributed systems,” she says. “Agentic and RAG-based architectures will require fine-grained runtime control over latency, access scope, and delegated trust boundaries. Enterprises should start treating data delivery as programmable infrastructure, not a byproduct of storage or networking. The organizations that do this early will scale faster and with less risk.”

Sponsored articles are content produced by a company that is either paying for the post or has a business relationship with VentureBeat, and they’re always clearly marked. For more information, contact sales@venturebeat.com.

from the monetizing-imaginary-friends dept

My biggest complaints with AI tend to be with the human beings who are rushing language learning models into mass adoption without doing their basic due diligence. Like AI toy maker Bondu, the creator of “AI” enabled stuffed animals, which recently left the stored chat logs children have with their polyester-filled automated friends openly available online to anybody with a Gmail account:

“[security researcher Joel Margolis] made a startling discovery: Bondu’s web-based portal, intended to allow parents to check on their children’s conversations and for Bondu’s staff to monitor the products’ use and performance, also let anyone with a Gmail account access transcripts of virtually every conversation Bondu’s child users have ever had with the toy.”

At this point there’s just no excuse for this sort of thing. We’ve been writing for more than a decade about how most “smart,” internet-connected toys were being rushed to market without adequate privacy and security safeguards, creating OpSec risks for kids before they’ve even been adequately potty trained.

Now, as we’ve done in sectors like health insurance and journalism, we’ve slathered half-cooked language learning models all over existing dysfunction we refused to address, called it innovation, and then ignored the fact we’ve introduced entirely new problems.

In this case, the included exposed data included kids’ names, birth dates, family member names, and even the detailed summaries and transcripts of every previous chat between the child and their Bondu stuffed animals.

On the plus side, once alerted, the company quickly fixed the issue in a matter of minutes. And when asked by journalists about it, didn’t try to lie about the problem (a low bar, but still):

“When WIRED reached out to the company, Bondu CEO Fateen Anam Rafid wrote in a statement that security fixes for the problem “were completed within hours, followed by a broader security review and the implementation of additional preventative measures for all users.” He added that Bondu “found no evidence of access beyond the researchers involved.”

If hackers are clever they don’t leave many footprints, so that last bit might not be worth much.

One recent survey found that 84 percent of Americans want tougher privacy laws. But corruption has ensured that the country still lacks even baseline internet-era privacy protections. The powers that be have decided, repeatedly, to prioritize mass commercialized surveillance over public safety, and it’s only a matter of time before those chickens come home to roost in ways we can’t even begin to consider.

Filed Under: ai, automation, privacy, security, smart, surveillance, toys

Companies: bondu

The frantic speed of the Trump Administration’s education policy changes is leaving K-12 school officials confused about what orders are legal, whether they need to alter district policies to stay in compliance, and what, if any, federal funding might be at risk.

The April 3 letter that sought to disqualify any K-12 schools with diversity, equity and inclusion programs from receiving federal funding is a good example. The Department of Education missive, titled in part Reminder of Legal Obligations Undertaken in Exchange for Receiving Federal Financial Assistance, requested certification from school officials that they were meeting various federal laws in order to receive Title VI funding.

Within days, at least two school districts, City Schools of Decatur in Georgia and Haldane Central School District in Cold Spring, New York, altered policies to try to comply with current government discrimination definitions. Decatur rescinded two policies about equity and school board governance, while Haldane’s Board of Education voted unanimously to suspend its wide-ranging DEI policy, which sought “to advance inclusive and culturally responsive teaching and learning.” But when a trio of rulings from federal court judges found that the Trump administration could not withhold federal funds for these reasons, both districts rushed to reinstate their policies.

“It’s a really complicated landscape right now,” said Ray Li, policy counsel for education equity at the Legal Defense Fund. He reiterated that DEI policies, affinity groups, and racially inclusive curricula are not illegal and never have been for school districts.

Not an Either/Or

One of the confusing issues in this dispute is who’s defining DEI. Carlos Hoyt, who has worked as a diversity and equity trainer at several private schools in Massachusetts, maintains that making DEI a binary question between having a policy or not having a policy is a false choice. He said even policies that hold that some people are more deserving of resources and protections than others is a form of DEI, albeit a restrictive method. An expansive method of DEI would be more inclusive, he added.

“No one can be against DEI. It’s just a matter of one’s approach to it,” he added.

In Decatur, Board Vice Chairman Hans Utz admitted in a public meeting on April 23 that the board’s original decision to cut equity policies was “wrong.” The board had rescinded two policies and amended three others on April 4, just a day after the Department of Education request. The actions were not made public until an April 15 meeting. When one parent shouted her displeasure about eliminating the DEI policies during this meeting, three city police officers carried her out of the room.

“We were counseled to be quiet,” Utz said, according to a story in Decaturish.com. “The risk that we faced of dissolution was great enough that we needed to not draw attention to it,” he said. “We followed that [advice] and we were wrong.”

Decatur officials said the 10-school, 5,500-student district gets about $3.2 million in federal funding, mostly for special education. The school board unanimously agreed to reinstate all its DEI policies on April 29, four days after the three court rulings issued preliminary injunctions against Department of Education requests.

Both Utz and board chairman Carmen Sulton declined to comment further on the board’s actions when reached recently via email.

Nia Batra, the student representative on Decatur’s board, called the board’s original decision “hasty” but added that it was “a reflection of the panic this current administration has instilled, where many governing bodies feel compelled to comply without proper information or legal counsel,” she said. “Boards should be more willing to stand up to this administration’s bullying.”

Removal, then Reversal

While DEI’s roots extend back to the Civil Rights era, in 2020, after George Floyd’s murder, there was a “major shift” in government, companies, and schools adopting DEI policies, according to the Center for Urban and Racial Equity. In 2021, then-President Joe Biden signed an executive order calling for federal departments to advance racial equity. Just two years later, however, the Supreme Court ruled against the affirmative action programs from Harvard and the University of North Carolina. Haldane adopted its policy in 2022 while Decatur passed its equity policy in 2023.

In New York, Haldane’s school district suspended its DEI policy on April 22 and reinstated it on May 6. The 800-student, three-school district feared losing about $450,000 in federal funds.

While Haldane officials did not respond to questions about the issue, in a letter released to the public on April 26, the Board of Education said it suspended its DEI policy to investigate “minor language revisions that would not alter the spirit of the policy.” The letter added, “Given the speed with which the federal government is canceling funding for projects and activities, we were genuinely concerned that they would withhold this funding from Haldane.”

Five days later, in another letter, the board announced that it would reinstate the policy in full after the Education Department was blocked from withholding funds over this issue. Board members met with the public and high school students in separate sessions to discuss the issue and respond to comments.

New York was one of 19 states to sue the Trump Administration, claiming that DEI policies did not violate the Civil Rights Act of 1964. Title VI of the Civil Rights Act prohibits discrimination based on race, color, and national origin in programs and activities receiving federal funding.

Li with the Legal Defense Fund said departmental orders, such as the Department of Education Reminder of Legal Obligations, “don’t create law and can’t change the application of Title VI itself.”

The ‘Dear Colleague’ Letter

While the April reminder is one example of the administration causing confusion by issuing supposed new rules that have then been stopped by legal challenges, there have been other attempts. In February, the Department of Education issued a Dear Colleague letter to both higher education institutions and K-12 schools to “clarify and reaffirm the nondiscrimination obligations of schools… Discrimination on the basis of race, color, or national origin is illegal and morally reprehensible.”

The four-page letter from Craig Trainor, the acting assistant secretary for civil rights at the Department of Education, specifically mentioned discrimination against white and Asian students, while calling for all institutions to comply with federal civil rights law within 14 days. In April, a federal judge enjoined the DOE from “enforcing and/or implementing” this letter, as well as two other DEI-based DOE rulings.

Not everyone was confused by the flurry of action. In Massachusetts, Gov. Maura Healey and state Attorney General Andrea Joy Campbell rushed to issue joint guidance to the state’s higher education institutions and K-12 schools after the February Dear Colleague letter: “Massachusetts educational institutions should continue their work to foster diversity, equity, inclusion, and accessibility among their student bodies.”

“We believe that bringing people of different backgrounds and perspectives to the table – including women, people of color, and the LGBTQ+ community – is a strength, especially in education,” the governor said.

On July 30, the Department of Justice joined the Education Department’s campaign by issuing guidance that recipients of federal funding can’t “engage in unlawful discrimination.” The nine-page memo specifically called out DEI programs.

But, Li added, this guidance remains an “open legal question” while courts decipher the exact meaning. “It’s messy for that reason.” Even though courts haven’t acted on the Department of Justice guidance, Li said, because it rests on the same arguments as the previously mentioned DOE actions, it is likely to be rejected by a future court decision.

Tracking Cases

Because of the fast-moving actions by the federal government and various court lawsuits and decisions, there are two ways school officials can track ongoing cases. Lawfare, a journal covering legal and policy issues, has a litigation tracker that monitors 227 active cases challenging Trump administration actions, 19 lawsuits by the administration against state or local laws, as well as other cases where the Supreme Court has ruled or judges have ruled for or against the federal government. Also, the Brookings Institution has its own litigation tracker covering all the challenges to Trump’s executive actions that impact K-12.

In the meantime, federal targeting of DEI and gender issues has had a financial impact on schools.In September, the Trump administration said it would cancel more than $65 million in funding for magnet schools in New York City, Chicago, and Fairfax, Virginia.

The administration called for both New York and Fairfax to change gender policies that allow students to participate in physical education and athletic team sports based on their gender identity. In Chicago, the administration called for the district to eliminate its Black Student Success Plan.

“The Education Department is taking a one-by-one approach,” Li said. He wasn’t surprised that officials from Haldane and Decatur refused to comment further about their cases. “I don’t think anyone wants to attract attention right now,” Li said.

Hoyt said the national debate about DEI hasn’t stopped schools he works with from continuing its training programs.

“People know they need to do the work,” he said, adding that he is scheduled to conduct DEI training at the four campuses of the Brooke Charter School in Boston and Chelsea, Massachusetts.

The former associate dean of students at Phillips Academy in Andover, Massachusetts, Hoyt has also taught at four different colleges in the Bay State, including Boston University. He works with Boston College’s Lynch Leadership Academy, a program that annually trains about 30 school officials seeking to become principals. His equity and justice training in the academy’s year-long cohort will continue, he said.

But Hoyt doesn’t expect resolution in this fight anytime soon. Citing the partisanship over the language, policies and funding of DEI today, he said, “I think we’re stuck with this [uncertainty] for a while.”

• Decodo reports 68% rise in digital squatting scams over five years
• Techniques include typosquatting, combosquatting, TLD squatting, and homograph attacks, tricking users into sharing credentials or payments
• WIPO logged 6,200 domain disputes in 2025, the highest ever; Decodo urges brands to register domains beyond .com for protection

Digital squatting is getting increasingly popular among scammers, ruining businesses and their reputations at an unprecedented pace.

This is according to a new report from Decodo, which said that there’s been a 68% increase in these cases in half a decade.