For three months, the Federal Reserve warned the country’s largest banks about a cybersecurity threat it could not fully assess itself. The Fed Anthropic Mythos alarm was real, urgent, and public. What was not public until now is that the regulator issuing the warning did not have direct access to the very model it was concerned about.
This gap is central to the story of Claude Mythos banks’ cybersecurity risk. It shows how even the world’s most powerful financial regulator can fall behind when new artificial intelligence develops faster than the institutions designed to oversee it.
What Happened in April
The episode began with an emergency summons. In April, the Federal Reserve and the Treasury Department convened an extraordinary meeting with the chief executives of the country’s largest banks. Then-Treasury Secretary Scott Bessent and then-Fed Chair Jerome Powell called that meeting specifically to warn bank leaders about cyber risks tied to Anthropic’s newest AI model. The trigger was Anthropic Mythos Preview April, the model’s public debut, which the company paired with a cybersecurity initiative called Project Glasswing.
Anthropic designed the model to find weaknesses in software before criminals could exploit them. The company said the tool could spot vulnerabilities in operating systems and browsers more quickly than any human security team. Anthropic revealed that about 50 organizations had access to the model at launch but only named a few publicly. Among those named were JPMorgan, Amazon, Apple, and Google, who received early access. This put some of the world’s largest financial and technology firms ahead of the regulator that oversees several of them.
Project Glasswing Banks Got In. The Fed Did Not
That imbalance is the most striking part of the story. Project Glasswing banks, including JPMorgan Chase, started using Mythos on their own systems almost right away, fixing flaws the model found before attackers could exploit them. Meanwhile, the institution responsible for the stability of the entire U.S. banking system had to wait.
Financial-sector cybersecurity teams pointed out the irony: banks in Project Glasswing were actively fixing vulnerabilities the model found, while the Fed was locked out of the same tool. Industry observers sum up the situation: “Fed rang alarm Anthropic Mythos model” describes the warning phase, while “Claude Mythos banks months delay access”describes what happened next. The regulator raised the flag, then had to watch from outside the room.
The Fed AI Model Access Delay, By the Numbers
The Fed AI model access delay stretched at least three months past the April meeting. CNBC reported the central bank was still trying to secure access to Mythos as recently as July 15, and neither the Fed nor Anthropic offered further comment on whether access has since come through. New Fed Chairman Kevin Warsh, who took over from Powell earlier this year, told lawmakers the central bank had been actively seeking access to Mythos and other frontier models so it could repair vulnerabilities in its own systems and in those connected to the wider financial sector. Warsh put it: the Fed needs to “do all we can to patch any vulnerabilities that we have.”
The timing is important. Anthropic briefly suspended access to Mythos in June to comply with export-control rules. The Commerce Department later lifted those restrictions, and the company restored access to trusted partners on July 1. This months-long gap also overlapped with a time when even outside partners with credentials could not reach the model.
Why the Access Gap Matters Beyond One Model
It may have been a simple bureaucratic issue, such as a paperwork delay or a problem with onboarding a vendor. But that view misses the bigger problem. Regulators are being asked to supervise institutions that already have tools regulators themselves do not have. For example, a bank examiner reviewing JPMorgan’s cyber defenses this summer could ask what vulnerabilities Mythos found and whether they were fixed. Until the Fed gained access, it had no independent way to check those answers against the model itself.
This is not simply a hypothetical concern for one agency. Frontier AI models are now often seen as dual-use tools. The same ability that helps defenders find flaws can also help attackers if misused. Anthropic’s decision to limit Mythos’ access to about 50 vetted organizations reflected this risk. The company trusted banks, cloud providers, and some government contacts with early access. It took longer to extend that trust, or finish onboarding, for the regulator whose job is to oversee all of them.
Warsh’s testimony suggests the Fed sees this as a repeated problem, not merely a one-time event. He described the Mythos episode as proof that financial regulators need to build ongoing relationships with AI labs before the next model is released, instead of rushing for access after a warning. Congress has also shown more interest in whether banking regulators have the technical skills to supervise AI-driven risks. The Mythos timeline now gives that debate a real-world example.
The Forward-Looking Question
Delays like this rarely affect only one institution or one model. As more banks start using AI-driven vulnerability scanning as a standard practice, the agencies that supervise them will need quicker and more reliable ways to access the tools to mold these risks. It is still unconfirmed whether the Fed has since gained access to Mythos. What is clear is that this episode has become a test case for how financial regulators can keep up with an industry that moves faster than Washington’s usual pace.
A manufacturing strategy can survive delays. It rarely survives a lack of customers. That reality has shadowed Intel’s foundry ambitions for years, making every external contract far more important than its budgetary value alone. The announcement that Intel foundry Fortinet customer is now a reality marks the first publicly identified foundry client under CEO Intel Lip-Bu Tan foundry win, signaling that Intel’s manufacturing business is beginning to earn industry confidence. The agreement also means Intel firewall chips designed for Fortinet will be manufactured using Intel’s fabrication capabilities, creating an early benchmark for the company’s renewed foundry strategy.
Intel foundry Fortinet customers mark an Important Milestone.
Intel has confirmed that Fortinet, a leader in cybersecurity, has chosen Intel Foundry to make its future firewall security processors. The financial details are not public, but the partnership is important for reasons that go beyond the contract’s size.
Intel has spent billions over the years building a contract manufacturing business that could compete with TSMC and Samsung. Still, many potential customers were hesitant. Most chose established manufacturers with long track records and reliable results.
This reservation made the announcement of Intel’s first public foundry customer particularly meaningful.
Instead of just talking about plans, Intel can now point to a real customer that trusts it to make high-performance networking chips.
This is also the first clear sign that Lip-Bu Tan’s foundry win strategy is starting to work after months of changes inside Intel.
Why Fortinet Selected Intel
Fortinet designs specialized processors that power enterprise firewall appliances protecting corporate networks against cyberattacks.
These processors need to be highly reliable, made with consistent quality, and delivered through dependable supply chains. Any disruption could impact cybersecurity for governments, banks, hospitals, and global companies.
The agreement means Fortinet security chips Intel facilities will manufacture are expected to support future generations of firewall products.
Intel has not shared which manufacturing process it will use, but analysts think Fortinet’s choice shows trust in Intel’s better technology, packaging, and U.S.-based factories.
This partnership now makes the phrase “Intel manufactures firewall chips Fortinet” an accurate way to describe one of Intel Foundry’s first public business relationships.
Lip-Bu Tan’s Strategy Focuses on Customer Confidence
Since taking over as CEO, Lip-Bu Tan has focused on delivering results instead of making big promises.
Earlier Intel leaders often talked about long-term manufacturing goals, but they faced delays and production problems.
Tan has chosen to focus on strong operations, building customer relationships, and making engineers accountable.
This new approach seems to be working.
The Intel Lip-Bu Tan foundry win demonstrates that rebuilding credibility often starts with modest but tangible victories rather than headline-grabbing announcements.
Industry experts point out that foundry customers seldom change manufacturers without a thorough technical review.
Before choosing a manufacturer, companies look at things like production yields, defect rates, packaging technology, supply chain strength, teamwork with engineers, and long-term capacity.
Fortinet’s choice suggests that Intel has managed to solve many of these issues.
Why the Foundry Business Matters
Intel’s old business model was to design and make its own processors in-house.
But the rest of the semiconductor industry changed differently.
Companies like Nvidia, AMD, Qualcomm, Broadcom, and Apple started using a ‘fabless’ model, where they design chips but hire other companies to make them.
In response, Intel created Intel Foundry Services to become an independent manufacturer for outside customers.
But having advanced factories is not enough for success.
Success also depends on trust.
Every new contract shows that customers believe Intel can deliver on time and meet high-quality standards.
So, announcing Intel’s first public foundry customer is more than just another deal. It shows that Intel’s investments in manufacturing might finally be bringing in outside business.
Firewall Chips Represent a Strategic Segment
Enterprise firewall processors are very different from the CPUs found in consumer devices.
Instead of focusing on general computing power, these chips are built for functions such as inspecting data packets, speeding up encryption, handling lots of network traffic, and keeping delays low.
As cyberattacks get more advanced, organizations need hardware that can handle huge amounts of encrypted data without slowing things down.
By making Intel firewall chips for Fortinet, Intel is strengthening its place in networking and opening up new opportunities beyond just PCs and servers.
This deal also lets Intel show off its cutting-edge packaging and manufacturing skills in specialized chip markets.
The phrase “Intel foundry lands Fortinet first customer” is more than just marketing. It marks an important moment for Intel’s manufacturing strategy.
The Road Ahead Stays Difficult
One customer alone will not undo years of tough competition.
TSMC continues to dominate global foundry manufacturing, serving nearly every major semiconductor designer. Samsung also remains a strong competitor with advanced process technologies and a significant manufacturing scale.
Intel still has big challenges ahead, like improving manufacturing yields, planning new technologies, finding more customers, and making sure its huge investments pay off.
So, Intel’s foundry turnaround is still ongoing.
But successful foundry businesses usually grow by adding customers little by little, not through sudden big changes.
If Intel keeps delivering quality products for Fortinet, other companies in networking, automotive, industrial, and AI sectors would start to consider Intel as a manufacturing partner.
This possibility makes today’s announcement important, even if the deal itself is not huge.
Investors Will Watch Customer Momentum
Wall Street has always judged Intel’s foundry plans by one simple measure: how many customers sign up.
Spending factories and new technology only matters if outside companies actually pick up Intel over its rivals.
The Intel foundry turnaround depends on turning engineering progress into real business deals.
If Intel can announce deals with bigger chip designers in the future, investors will feel more confident that its manufacturing business is truly competitive.
For now, Fortinet gives Intel something it did not have before: a public customer willing to show faith in its manufacturing abilities.
That makes a real difference.
What This Means for the Semiconductor Industry
Governments and tech companies are looking more and more for multiple sources of semiconductor supply.
When most manufacturing happens in one region, it raises worries about political risks, natural disasters, and possible production problems.
Intel’s growing foundry business gives customers another option, especially those who want more variety in where their chips are made.
The relationship highlighted by Fortinet security chips and Intel manufacturing also supports wider efforts to expand advanced semiconductor production in the United States.
If Intel makes this partnership work and brings in more customers, it could slowly build a stronger place in the global foundry market.
The news that Intel now has Fortinet as an official foundry customer will not change the semiconductor industry overnight. Real turnarounds usually start with proof that customers are willing to trust a new strategy, not with huge deals. Fortinet’s choice gives Intel that proof. The next step is to turn this early win into a steady stream of new manufacturing contracts. As the industry looks for more diverse and resilient manufacturing, Intel now has its first public example to build on.
Santa Clara, California | Wednesday, July 22, 2026
A server chip that uses as much power as a household oven and has enough memory to run 375 iPhones at once might not seem efficient. But that’s exactly how Nvidia is presenting Vera, its first in-house CPU in almost ten years. This move has chip analysts now saying Intel and AMD are on the defensive.
On Tuesday, Nvidia published the Nvidia Vera CPU details that engineers and cloud buyers have been waiting for since its announcement in May. The company confirmed Vera CPU 250-450 watts of configurable power draw and, more strikingly, Vera CPU 1.5 terabytes memory capacity per chip. These numbers, along with a technical white paper and independent benchmarks, show Nvidia making its boldest move yet into a server processor market long dominated by Intel and AMD.
Why Nvidia Built Its Own CPU
For years, Nvidia has sold graphics processors to data centers, usually pairing them with CPUs from Intel, AMD, or Arm-based partners to handle other tasks. Vera changes this setup. It’s Nvidia’s own chip, built around a custom core called Olympus, and designed specifically for the AI agents that now answer customer questions, write code, and oversee complex tasks across data centers.
This difference is more important than it seems. Traditional server CPUs were made to run many virtual machines at a low cost, focusing on having lots of cores instead of fast single threads. Agentic AI needs the opposite: a processor that can quickly handle one task after another, since a slow CPU can leave a costly GPU waiting. Ian Buck, Nvidia’s vice president of hyperscale computing, said at a recent briefing that the rise of AI agents has made the CPU “much more integral” to how quickly a system can answer a single question.
The Memory Bet
The notable number 1.5 terabytes of memory on a single chip—comes from an unusual choice. Vera uses low-power DDR5X memory, the same type found in laptops and smartphones, instead of the more power-hungry DDR5 modules that typically fill out server racks. Nvidia says the resulting low-power memory AI chip design draws under 30 watts for its entire memory subsystem, compared to over 100 watts for DDR5. With a data fabric that moves information at 3.4 terabytes per second, the chip can supply thousands of AI agents with data at once without using too much power.
Nvidia is clear about what this means. The company frames Vera as Nvidia’s new CPU class territory — a processor built around agent throughput rather than the raw core density that has defined server chips since the cloud-computing boom of the 2010s. Independent tests from the cloud platform DeepInfra support this, showing that Vera can run up to 1.6 times more AI agents at the same time compared to other processors in early trials.
Nvidia Challenges AMD Intel on Their Own Turf
The stakes are not small. AMD currently holds roughly a third of the server’s CPU market. At the same time, Intel controls close to two-thirds, and both companies have spent years cultivating relationships with the hyperscale cloud providers that buy chips by the tens of thousands. Karl Freund, founder of Cambrian AI Research, has argued that Nvidia’s move is less about winning general-purpose server workloads and more about cutting dependency. “The CPU is something they’ve done to kind of unhook their customers from using Intel or AMD CPUs, and they covet that revenue,” Freund said.
That framing captures why Nvidia challenges AMD Intel so directly with this launch. Vera is not meant for running basic websites or databases. Instead, Nvidia is selling it together with its Rubin graphics processors as the Vera Rubin platform, as well as offering standalone chips, dual-chip servers, and liquid-cooled racks that can hold 256 Vera chips at once. Each setup encourages customers to buy both the CPU and GPU from Nvidia, rather than mixing Nvidia graphics with a competitor processor. It leans heavily on companies it already supplies with GPUs — OpenAI has said it plans to deploy Vera chips in large quantities this quarter. Anthropic and SpaceX have also received early units. Analyst Remy Cout has described the rollout as still in its “early innings,” and Nvidia has not yet locked in a major public cloud provider beyond Oracle. Getting Amazon, Microsoft, or Google to redesign server fleets around a new CPU architecture is a slower, more political process than shipping graphics cards, and both rivals have institutional relationships that won’t unwind overnight.
What Intel and AMD Can Still Do
Neither competitor is standing still. AMD’s latest Epyc processors are still the benchmark Nvidia used for its own SPEC CPU 2026 results, which shows where the real competition is. Intel still has the largest presence in data centers worldwide, a scale that Nvidia has not matched outside its GPU customers. Both AMD and Intel have something Nvidia does not: decades of experience selling CPUs to buyers who care about price per core, not just AI performance.
The Wider Signal for AI Infrastructure
Tuesday’s announcement shows that the debate over AI hardware is not simply about graphics processors. For the last three years, GPU supply has limited how quickly companies could train and use AI models. With Vera, Nvidia now sees the CPU and its memory as the next important factor to control. A white paper about the Nvidia Vera CPU and its challenge to AMD and Intel argues that CPU design, not just GPU power, will determine which companies can run AI agents at scale and at a reasonable cost.
For enterprise buyers planning their infrastructure spending over the next year and a half, the main question is not about benchmarks but about vendor lock-in. A Vera CPU with 1.5 terabytes of memory, paired with a Rubin GPU, looks impressive on paper. Whether it becomes the standard for AI operations or just another high-end choice alongside Intel and AMD will depend on how quickly the big cloud providers testing it decide to adopt it. Nvidia has built the chip, but the market—not the white paper—will determine if it truly creates a new category.
A small amount of computer code could become the next trigger for economic sanctions. That is the message coming from Washington after Treasury Secretary Scott Bessent warned that the United States may soon penalize Chinese artificial intelligence companies accused of copying advanced American AI systems. His comments quickly raised concerns in the global tech industry, where such intellectual property is now seen as important as semiconductors and rare earth minerals. Issues like Bessent sanctions China AI, Treasury AI model theft, and US China AI distillation have become key topics for policymakers, investors, and AI developers.
Bessent sanctions China AI Signals Tougher Enforcement.
Treasury Secretary Scott Bessent gave one of the administration’s strongest warnings during a Bessent Fox Business interview, stating that sanctions remain under active consideration if evidence confirms Chinese AI developers gained from American large language models without permission.
Bessent said investigators have found digital “watermarks” from US-developed language models inside several Chinese AI systems. He explained that these markers suggest some developers may have trained their models using outputs from top American platforms instead of only using their own data.
His comments were in direct.
“We are finding watermarks of our U.S. large language models on many of the Chinese models, and that’s unacceptable,” Bessent said during the interview, adding that potential enforcement actions could arrive “in the coming days or weeks.”
His comments raised expectations that the Treasury Department might go further than export restrictions and start using financial penalties against organizations accused of breaking intellectual property rules.
Understanding Treasury AI model theft
The controversy centers on an AI training method known as “distillation.”
Distillation lets a smaller language model get better by learning the answers given by a larger, more advanced model. When used legally, distillation is a common machine learning method for companies building their own AI systems.
The dispute happens when developers allegedly rely on proprietary commercial models without authorization.
People who want stricter rules say that unauthorized distillation lets competitors get the benefit of years of costly research and billions spent on computing for much less money.
This worry is central to the Treasury AI model for theft debate. American AI companies have spent huge amounts of money on advanced models, so protecting intellectual property is now key to remaining competitive as a nation.
US-China AI Distillation Debate Intensifies
The discussion escalated after the widely reported Anthropic-Alibaba distillation claim, which alleged that Alibaba conducted what Anthropic described as the largest known unauthorized distillation attempt involving its commercial AI systems.
Alibaba has not agreed that these claims are true, but the accusation has led to more attention and scrutiny in the AI industry.
If regulators find that commercial AI models were regularly trained with unauthorized outputs from American systems, policymakers might say that current export controls are not enough.
The US-China AI distillation issue is about more than just business competition. It also affects national security, technology leadership, and future economic policy.
AI distillation brings up tough legal questions, unlike patent or software license disputes, because the outputs from language models regularly fall into a gray area for intellectual property.
This legal uncertainty makes it much harder to enforce the rules.
Why Digital Watermarks Matter
Digital watermarks are one of the few technical tools that can help trace where AI-created content comes from.
Many leading AI developers secretly add statistical markers to the responses their models generate. Users usually cannot see these markers, but researchers can sometimes use them to tell if another model has learned from protected outputs.
Bessent’s comments about watermarks suggest that investigators think they have technical proof, not just circumstantial evidence.
If that evidence holds up under independent review, it could give the administration a stronger legal reason for sanctions.
The statement also warns AI developers everywhere that it may now be possible to trace proprietary model outputs.
Markets Watch Washington Closely
Financial markets are starting to see artificial intelligence as a strategic industry, much like aerospace or advanced semiconductor manufacturing.
Sanctions against Chinese AI companies could change how investors approach cloud computing, chip manufacturing, enterprise software, and data infrastructure.
Technology investors are considering several possible outcomes.
First, sanctions could restrict access to American financial systems or software services.
Second, more export restrictions could impact shipments of AI hardware, especially advanced GPUs needed to train large language models.
Third, multinational companies working in both the US and China might have to meet new compliance requirements.
These possible changes are why Bessent sanctions China AI has quickly become one of the most closely watched policy issues in the tech sector.
Bessent warns of sanctions on China AI theft before September Talks.
The timing of these events makes them even more important.
The Treasury Secretary’s remarks arrive shortly before scheduled US-China AI talks September, where officials from both governments are expected to discuss artificial intelligence governance, technology competition, export controls, and cybersecurity.
Diplomatic talks often get harder when public warnings come before official meetings.
Analysts believe Washington may seek stronger commitments from Beijing regarding intellectual property enforcement before broader AI cooperation becomes possible.
The phrase Bessent warns sanctions China AI theft means more than just political talk. It shows a stronger negotiating approach meant to establish clear expectations before talks between the two countries start.
Whether those discussions reduce tensions or heighten existing disagreements is unclear.
The Technology Industry Faces Higher Compliance Standards
The AI industry is now in a new phase.
For years, most discussions were about innovation, model abilities, and commercial use. Now, policymakers are just as focused on accountability, transparency, and who owns the training data.
Companies building advanced AI models are expected to spend more on watermarking, usage monitoring, contracts, and technical tools to spot unauthorized copying of their models.
Cloud providers might also improve their monitoring systems to spot suspicious automated activity that looks like large-scale distillation.
At the same time, companies buying AI services may want more proof that commercial models were built with legally obtained training data.
The effects go far beyond just Washington and Beijing.
European regulators, Japanese tech agencies, and other governments watching AI rules closely may see this dispute as an example for future actions on generative AI intellectual property.
Treasury threatens Chinese AI model makers as Policy Evolves.
Whether sanctions actually happen will depend on what evidence investigators find and the administration’s bigger diplomatic goals.
Still, Bessent’s warning shows that artificial intelligence is now closely linked to economic security and foreign policy.
The idea that the Treasury Department might use financial penalties for AI-related intellectual property disputes is a big change in international technology regulation.
For developers, investors, and multinational companies, the message is clear. Governments now see advanced AI models as more than just commercial software—they are strategic national assets that need strong protection. As the policymakers prepare for the US-China AI talks in September, the tech industry will be watching to see whether the Treasury threatens Chinese AI model makers or uses the threat as a bargaining tool in bigger negotiations.
Wall Street Just Got a $60 Billion Reason to Rethink Super Micro.
It’s unusual for a single earnings update to change a stock’s price overnight. But on Tuesday, Super Micro Computer delivered one that did exactly that. SMCI stock surges 18 percent in after-hours trading when the San Jose-based server maker disclosed Super Micro’s $60 billion orders booked during its fiscal fourth quarter — a number so large that even experienced analysts needed a second read to confirm it. The headline is simple: Super Micro surges 18 percent on $60 billion orders, and the AI infrastructure buildout just found its newest exclamation point.
After spending the past two years dealing with accounting questions and investor doubts, this week’s jump was more than merely a rebound for Super Micro. It sent a clear message.
The Numbers Behind the Surge
Super Micro’s early business update, filed ahead of its complete fiscal fourth-quarter and year-end results on August 11, laid out three figures that mattered most to investors. First, the company confirmed an AI server backlog record, with orders received in the June quarter pushing commitments to an all-time high. Second, management raised its Super Micro gross margin of 15-17% guidance, more than doubling a prior forecast of 8.2% to 8.4%. Third, revenue is expected to be near the low end of the $11 billion to $12.5 billion range, which helped keep excitement in check and the rally realistic.
The jump in margins stands out on its own. Doubling gross margin guidance is a big deal and shows a real change in what Super Micro is selling and who is buying. Management said the improvement comes from a better mix of customers and products, which usually means more high-value, high-margin AI rack sales instead of lower-margin standard servers. Investors who were frustrated by low margins finally saw the shift they wanted.
The AI server backlog also addresses a question that has lingered since Super Micro’s 2024 governance issues: can the company still attract big, reliable customers? With a $60 billion order book, even if built over several quarters and clients, the answer seems to be yes. Not all orders are binding, and the company warned that some could be canceled or delayed. Still, the size of this update changed what the market expects for fiscal 2027.
Charles Liang and the SpaceX Connection
Earnings updates don’t happen in isolation, and this one came with a subplot that retail and institutional investors alike found impossible to ignore. SMCI SpaceX data center speculation had been building since June, when Charles Liang’s AI servers were being used in Elon Musk’s projects. The Super Micro CEO shared on social media that he was proud to help build a new gigawatt AI data center for SpaceX and xAI within a year, calling it the company’s fastest build yet.
This news is important because, after merging with xAI earlier this year, SpaceX has become one of the biggest buyers of AI computing power in the world. A gigawatt-scale data center order from a company like SpaceX is far from ordinary. It’s the kind of major customer relationship that can support a supplier’s technology plans for years. Liang is known for moving quickly, and Super Micro’s reputation is built on delivering new server designs faster than Dell or Hewlett Packard Enterprise. The SpaceX project demonstrates strength.
Competitors felt the impact right away. Shares of Dell and Hewlett Packard Enterprise also rose on Tuesday, showing that investors saw Super Micro’s news as a sign of strong AI infrastructure demand throughout the industry, not just for one company.
Why the Market Reacted So Sharply
Big stock moves like this usually don’t happen because of just one thing. Wednesday’s strong reaction came from three signals at once: record demand, much better margins, and a major customer relationship with one of the world’s most watched companies. Each of these could have moved the stock a little, but together, they changed the whole investment story.
Think about how this works. A server company with low margins can increase sales but still let investors down if profits don’t keep up. Super Micro faced this problem for much of the last 18 months. Now, with gross margins jumping to 15% to 17%, every extra dollar of backlog turned into revenue brings about twice as much profit as it did a year ago. That’s how a big order book leads to higher earnings.
There is still some doubts. Some traders and analysts wonder if the backlog will hold up and if all $60 billion in orders will turn into actual shipped and recognized revenue. Analysts at Raymond James pointed out possible risks with Nvidia’s chip supply, which could slow down deliveries even if demand stays high. Super Micro says its chip allocation hasn’t changed, but these concerns help explain why Wall Street’s overall rating on the stock remains cautious, even as the price rises.
What Comes Next
August 11 is now the key date that will confirm or challenge the excitement from Wednesday. The full fourth-quarter and year-end results will reveal whether the AI server backlog record equals the audited numbers and whether the 15-17%gross margin guidance holds up after the books are closed. Investors who pushed the stock higher based on early numbers are basically betting on management’s credibility before everything is verified.
But the bigger message goes beyond just one company’s results. Large tech firms and AI-focused companies are scrambling to get more computing power, often faster than suppliers can keep up. Super Micro’s record AI server backlog and its new relationship with SpaceX show that this rush is still going strong. In fact, the latest update suggests that the next wave of AI infrastructure spending will come with better margins, which could be more important to long-term investors than the size of the orders. Super Micro says its record backlog for fiscal 2026 is a turning point, not a peak, and that claim will be tested in three weeks.
A public listing is only meaningful if investors see a clear path to growth. Many media companies find it hard to show they can compete with worldwide streaming giants, falling ad revenue, and rising cost of digital content. That makes K Wave Media’s Nasdaq approval, K Wave’s strategic AI initiative, and Nasdaq Capital Market approval in 2026more than just routine news. Together, these moves show that the South Korean media company is working to combine financial soundness with artificial intelligence as part of its extended strategy.
On Monday, July 20, 2026, K Wave Media announced from Seoul that it received approval from the Nasdaq Capital Market. The company also shared plans to strengthen its finances and grow its AI-driven operations. This update helps investors better understand the company’s strategy as technology plays a bigger role in how entertainment is made, shared, and sold.
K Wave Media’s Nasdaq approval represents an Important Milestone.
Getting Nasdaq Capital Market approval in 2026 is a major step for any international company that wants better access to U.S. capital markets. To be listed, companies must meet strict financial, governance, and statutory standards.
For K Wave Media, this approval means it can keep trading on a well-known stock exchange and helps build investor faith in its compliance and governance.
For a Korean media company, a Nasdaq listing carries additional strategic value. U.S. exchanges expose companies to institutional investors, global analysts, and higher market visibility. That increased exposure can improve liquidity while creating opportunities to raise funds for future expansion.
This approval comes at a time when international investors are looking for media companies that combine entertainment with advanced technology, instead of depending only on traditional ad or broadcast revenue.
Financial Position Improves as K Wave Media’s balance sheet strengthened.
Alongside its Nasdaq announcement, the company reported initiatives that leave the K Wave Media balance sheet strengthened through financial restructuring and capital improvements.
A stronger balance sheet is important because media companies often need to invest a lot before new projects make money. Creating original content, licensing intellectual property, building technology platforms, and expanding overseas all take significant funds.
With better financial leeway, management can put more money into innovation instead of focusing on debt or short-term financing.
Investors frequently evaluate three factors when assessing publicly traded media companies:
Revenue growth potential
Financial soundness
Investment in future technologies
By announcing both Nasdaq approval and balance sheet improvements at the same time, K Wave Media seems to be tackling all three key areas in one move.
The K Wave strategic AI initiative Goes Beyond Automation.
Artificial intelligence is now a key competitive factor in entertainment, publishing, and digital media. But just announcing AI investments is not enough to impress investors. Markets want to know exactly how companies plan to create real business value.
The K Wave strategic AI initiative intends to use artificial intelligence across many parts of the media business, not solely as a separate tech project.
Possible uses include AI-assisted content creation, customized recommendations, multilingual localization, automated metadata, predictive analytics, and more efficient content distribution.
For example, recommendation algorithms can help streaming service platforms present viewers with programming based on behavioral patterns rather than simple genre preferences. AI-powered localization tools can accelerate subtitle creation and language adaptation, allowing Korean productions to reach larger international audiences more efficiently.
Machine learning can also help production teams spot trending themes, understand audience engagement, and choose the best release times.
These kinds of improvements show the type of strategic AI growth initiative increasingly adopted across the global media industry.
Why “K Wave Media gets Nasdaq approval AI” Matters to Investors
The phrase “K Wave Media gets Nasdaq approval AI” highlights two things happening at once.
First, Nasdaq approval provides market credibility and continued access to U.S. investors.
Second, the AI initiative shows that management wants to make long-term improvements, not just rely on traditional media revenue.
This mix may attract investors who want to invest in companies that combine entertainment and technology.
Media companies now compete on both creative talent and technology. Those that can analyze customer tastes, automate tasks, personalize recommendations, and distribute content productively may gain real advantages over slower rivals.
This strategy might become even more important given global demand for Korean entertainment keeps growing on digital services and digital media.
AI Could Differentiate K Wave’s Business Model
The entertainment industry is now in a time when technology affects almost every part of content delivery.
Artificial intelligence can make workflows more efficient without taking away creative decisions. Editors, producers, marketers, and distribution teams are using AI tools more often to handle big data, spot audience trends, improve promotions, and work faster.
If executed effectively, the strategic AI growth initiative could enable K Wave Media to:
Deliver customized viewing experiences.
Improve advertising efficiency
Expand multilingual material distribution.
Reduce production bottlenecks
Strengthen audience engagement through analytics-based insights.
These abilities are more valuable now that viewers expect digital experiences customized for them.
Market Confidence and Future Growth
International investors usually favor companies that combine financial discipline with clear innovation strategies.
The announcement that K Wave Media’s balance sheet strengthened alongside K Wave’s strategic AI initiative sends a message that management intends to support future expansion with stronger financial resources rather than speculative growth alone.
For a Korean media company, Nasdaq recognition also provides greater international visibility, possibly supporting business partnerships, licensing opportunities, and wider investor participation.
As more media is consumed digitally, companies that mix strong intellectual property with scalable technology may do better than those sticking to classic models.
What “K Wave Media strengthens balance sheet growth” Signals
The phrase “K Wave Media strengthens balance sheet growth” points to a bigger company strategy, not just one financial event.
Improving the balance sheet gives companies more flexibility to invest when market conditions change.
For K Wave Media, this flexibility can help fund technology investments, content production, international growth, acquisitions, or AI-related partnerships.
The company seems focused on creating a business model where economic security and technology innovation support each other, instead of competing for resources.
This moderate approach could become more important as investors look more closely at AI investments and expect real results, not just big promises.
Gazing Forward
With Nasdaq approval, a stronger balance sheet, and a clear AI growth plan, K Wave Media is at a key stage in its development. Success will depend on how well the company carries out these plans, but the latest announcements show a real effort to build both financial strength and technology skills. If management can use AI well in content creation, audience engagement, and global distribution, K Wave Media could become a stronger contender in the global digital entertainment market.
A growing number of smaller public companies face the same challenge: investors expect meaningful artificial intelligence capabilities, yet building proprietary AI platforms from scratch requires years of development, specialized talent, and considerable capital. That reality has fueled a wave of strategic takeovers. Against this backdrop, Brookmount acquires Mentor AI, constituting one of the more notable transactions in the emerging small-cap AI acquisition 2026landscape. The BMXI Mentor AI Corp acquisition demonstrates how targeted mergers can accelerate technology adoption while reshaping competitive strategy for smaller publicly traded companies.
Brookmount Explorations Completes Mentor AI Acquisition
Brookmount Explorations Inc. (BMXI) has officially announced that the Mentor AI Corp acquisition completeprocess has been finalized, bringing Mentor AI’s artificial intelligence capabilities under Brookmount’s corporate umbrella. The transaction represents more than a routine acquisition. It demonstrates a strategic decision to expand through technology integration instead of depending solely on lengthy internal software development.
The announcement that Brookmount Explorations completes the Mentor AI acquisition immediately attracted attention because it corresponds with a wider trend of small-cap AI consolidation, where publicly traded companies seek established AI expertise through mergers and acquisitions.
For investors, the BMXI Mentor AI Corp acquisition serves as another example of how smaller companies are attempting to position themselves for a market increasingly driven by intelligent automation, predictive analytics, and enterprise software solutions.
What Mentor AI Brings to Brookmount
Acquisitions work best when they bring in skills or technology that would otherwise take a lot of time and money to develop. Mentor AI seems to offer just that for Brookmount.
Mentor AI’s technology is built to help businesses make better decisions using data analysis, automation, and smart workflow tools. Instead of starting from scratch, Brookmount now has access to technology it can use and expand in different business areas.
For Brookmount Explorations Inc. (BMXI), integrating AI capabilities may create several competitive advantages.
Using Mentor AI’s existing technology, Brookmount can develop products faster instead of building everything itself. The company may also be able to license this technology or form new partnerships. Having access to skilled AI developers and engineers can help Brookmount keep innovating and make hiring easier.
Most importantly, this acquisition gives Brookmount more flexibility. The company can now explore how AI can support new projects in different industries, rather than just sticking to its usual business.
Brookmount Acquires Mentor AI During Rising Small-Cap Consolidation
Why acquisitions have become the preferred strategy
Building advanced AI platforms for businesses takes a lot of investment in machine learning, cloud computing, cybersecurity, data management, and skilled engineers.
For many smaller public companies, these costs can quickly become too high to manage.
This is why small-cap AI consolidation has accelerated over recent years. Rather than spending years building their own systems, companies are buying others that already have proven technology, skilled teams, and products ready for the market.
The fact that BMXI acquires Mentor AI Corp reflects this evolving corporate strategy.
Instead of seeing AI as just an experiment, more management teams now view it as a key part of their business that affects efficiency, customer relationships, product development, and future revenue.
Investor Perspective on the Acquisition
Investors usually look at three main questions when judging an acquisition.
First, does the company being bought bring valuable technology?
Second, can the management team successfully combine the two businesses?
Third, will the deal increase value for shareholders over the long term?
While financial details about the Mentor AI Corp acquisition complete announcement remain limited, the strategic rationale appears straightforward.
If Brookmount can successfully add Mentor AI’s products to its business, the deal could make the company more competitive and create new chances for ongoing software revenue.
Investors usually favor acquisitions that make operations more efficient, open up new markets, or bring in technology that can grow with the company.
On the other hand, problems with combining the companies, duplicate systems, or slow product launches could lower the expected benefits.
As with most tech deals, how well the plan is carried out will decide if the acquisition meets its goals.
AI Mergers Are Changing the Small-Cap Market
Big tech companies often make news with huge AI investments, but smaller public companies are also making important moves.
The small-cap AI acquisition 2026 market increasingly shows a shift toward focused acquisitions rather than speculative research spending.
Rather than trying to compete with big tech firms, smaller companies often look for specialized AI businesses that can offer practical solutions in specific markets.
The announcement that Brookmount Explorations completes the Mentor AI acquisition fits squarely within this trend.
Experts have noticed more interest in buying companies that already have their own algorithms, automation tools, analytics platforms, or specialized AI knowledge.
These deals often lower the risks of development and help companies bring products to market faster.
Strategic Opportunities for Brookmount
The long-term value of the BMXI Mentor AI Corp acquisition will depend on how effectively Brookmount leverages Mentor AI’s technology.
Brookmount could use this deal to expand its AI software, create smart automation tools for clients, enhance decision-making with analytics, and form new partnerships to reach more markets.
This acquisition also helps Brookmount build its standing as a tech-focused company. Public companies now compete not just on growth, but also on how innovative they appear to investors.
Artificial intelligence is still a major investment trend worldwide, so technology acquisitions like this are especially important.
Challenges That Should Not Be Neglected
Every acquisition comes with risks in how it is carried out.
Bringing new technology into a company often means matching up software systems, engineering teams, security standards, and company culture.
If Brookmount struggles to combine Mentor AI’s technology with its own, the expected benefits could take longer to appear.
Competition in the market is still very strong.
The AI software market changes quickly, so companies need to keep investing to stay up to date. Businesses that buy AI technology must keep improving it, not just rely on what they have now.
Investors should watch for new product launches, partnerships, financial results, and progress in combining the companies after the acquisition.
What This Means for the AI Industry
The announcement that BMXI acquires Mentor AI Corp strengthens a broader reality within artificial intelligence markets.
Innovation is no longer limited to in-house research labs.
Tactical acquisitions are now a key way for companies to quickly enter AI-driven markets.
As business customers keep asking for automation, prognostic tools, and better workflows, deals like this could become more common in healthcare, finance, tech, cybersecurity, and business software.
For smaller public companies, buying established AI firms is often a quicker and less risky option than building similar technology themselves.
This trend will likely lead to more small-cap AI mergers in the next few years.
Viewing Ahead
Finishing the Mentor AI Corp acquisition is a big step for Brookmount Explorations Inc (BMXI). Whether this deal leads to better financial results will depend on how well the companies are integrated, ongoing product development, and how effectively Brookmount turns its new AI tools into real business growth. As small-cap AI deals pick up speed in 2026, Brookmount’s move may represent a bigger change in how new public companies compete in an AI-focused economy.
Chronic wounds impact about 1% to 2% of people worldwide at any time, and treating them in the United States costs over $20 billion each year. That figure is the background against which Medical Care Technologies’ AI wound care ambitions now play out. On Monday, the Mesa, Arizona-based company, listed as OTC PINK:MDCE, laid out a case for Medical Care Technologies’ diversified strength, pairing active revenue from two consumer subsidiaries with ongoing development of the imaging technology it hopes will eventually reach hospitals, wound clinics, and home-care patients.
A Diversified Balance Sheet Behind the AI Bet
While most small-cap health tech companies focus on one product and ask investors to be patient, MDCE is taking a different approach. Its subsidiary Infinite Auctions hosts live online sales of sports memorabilia, with a major auction of collectibles ending this Saturday, July 25, and the company shows strong interest from bidders. Another subsidiary, Real Game Used, has developed an automated photo-match authentication tool for collectors, and submissions to this service are increasing along with its traditional authentication business.
Neither subsidiary provides patient care or wound analysis, but both generate revenue now and rely on the same core skill: extracting reliable signal from a photograph. That shared engineering thread is the connective tissue behind the AI vision platform wound care strategy MDCE is now emphasizing to shareholders. Its AI vision technology is already used in consumer products like the Snapshot Recipes app, allowing management to test image-recognition correctness before applying it to clinical skin and wound analysis.
What the AI Vision Platform Actually Does in Wound Assessment
At its core, the technology is simple, though not easy to perfect. A patient or caregiver takes a photo of a wound with a smartphone. Computer-vision software then measures the wound’s length, width, and depth from the image, tracks changes over time, and identifies patterns that may signal infection or delayed healing, such as shifts in color, swelling, or drainage.
MDCE calls this process sequential imaging integrated with automated edge detection and infection-risk modeling. In practice, a nurse who used to visit a patient’s home, or a patient who had to travel to a clinic for a routine check, can now upload a photo and let the software handle the first assessment. For rural patients, homebound seniors, and people with chronic ulcers, moving from face-to-face visits to remote image review is the main benefit.
From Auction House Photography to Melanoma Screening
The company is also expanding its imaging technology into skin cancer screening through a partnership with Derm Foundation, a dermatology imaging model supported by Google. This cooperation adds a melanoma-risk classification feature to MDCE’s platform, so the same camera that tracks wound healing can also flag suspicious moles for follow-up. Using one imaging system for two clinical needs is efficient and shows how a well-trained computer-vision model can be adapted to related skin conditions at a low cost.
Sizing the Opportunity: The Multi-Billion Wound Care Market
Independent market research supports the scale MDCE is targeting, though exact numbers differ by firm and by how “AI in wound care” is defined. One estimate value the global AI-enabled wound analysis market at about $2.28 billion by 2030, growing nearly 14% each year. Another forecast puts the AI segment at $2.21 billion by 2029, with growth above 22% annually. Looking at chronic wound care overall, not just the AI part, US spending already exceeds $20 billion a year and could reach $30 billion by 2030 as more people need long-term wound care due to diabetes, obesity, and an aging population.
No matter which estimate is most accurate, the trend is clear, which is why MDCE often uses the term “multi-billion wound care market” in its investor updates. This is a real market, not something the company created to support its valuation. Established companies like NATROX Wound Care and Net Health are already active in this area, along with remote-monitoring apps like Tissue Analytics. Hospitals are also willing to pay for tools that reduce the need for in-person wound checks, which are costly and difficult to arrange for patients who cannot easily travel.
Where MDCE Fits Into the Wider AI Diagnostics Trend
The wound care market opportunity MDCE is chasing sits inside a larger pattern reshaping US healthcare: artificial intelligence is being directed disproportionately at conditions that are expensive to manage but chronically underserved by specialist capacity. Chronic wound care fits this pattern well. Wound-care specialists are not evenly spread across the country, face-to-face visits are expensive for patients and insurers, and missing signs of worsening wounds, like an infected diabetic foot ulcer, can lead to amputation or hospitalization.
Studies on wound-monitoring apps for patients have shown that AI-assisted image analysis can match or exceed the consistency of manual clinical measurement while cutting the need for face-to-face consultations. That body of evidence is precisely what gives the AI vision platform positioning MDCE has adopted its commercial logic: build a tool that clinicians trust to replace some in-person visits, and reimbursement and adoption are likely to follow.
Risks Investors and Clinicians Should Weigh
There are still risks. MDCE is listed on the OTC Pink marketplace, which has fewer disclosure requirements than Nasdaq or NYSE, and its wound and skin tools are still in beta, not fully launched. The company has also dropped a provisional patent application in July after its engineers moved past the first idea. While this makes sense from an engineering perspective, it shows the platform is still evolving. Getting regulatory approval for any tool marketed as a diagnostic aid, not just a wellness app, adds extra time and cost that press releases often overlook.
The Way Forward
Medical Care Technologies’ AI wound care platform now has two advantages that many smaller health-tech companies lack: immediate cash flow from its auction and authentication businesses, and a significant clinical challenge—chronic wound assessment—that matters if the technology works as promised. Whether MDCE moves from beta testing to real clinical use will depend more on whether hospitals, clinics, and insurers trust the imaging accuracy enough to change their payment practices than on market estimates. The key milestones to watch are regulatory approval, a completed beta rollout, and the first independent clinical data proving the AI vision platform’s value in the multi-billion-dollar wound care market.
Electricity is now one of the main challenges for artificial intelligence infrastructure. As AI clusters grow and utilities warn about higher demand, operators know that even small gains in power efficiency can impact operating costs. This is why Hillcrest’s 99 percent efficient design, 800V AI data center power, and ZVS architecture next gen have attracted interest beyond engineering circles. A technical breakthrough that cuts energy losses can shape decisions about spending, infrastructure, and the enduring profitability of AI.
Hillcrest Launches a High-Efficiency Power Architecture
Hillcrest Energy Technologies has released a technical paper that describes a power conversion platform built for new AI infrastructure. The paper explains a single-stage ZVS architecture that can deliver over 99% efficiency for high-voltage uses in today’s AI facilities.
Hillcrest’s technical paper AI comes as large-scale operators keep investing billions in GPU clusters and face rising electricity costs. Each new rack in an AI setup adds energy use, cooling needs, and system complexity. Boosting electrical efficiency at the power conversion stage helps tackle these monetary challenges.
The announcement surrounding Hillcrest publishes 99 percent efficient AI design shows an engineering approach that cuts energy losses without making systems more complicated. For investors looking at AI infrastructure, better efficiency is more than an engineering win—it’s a financial factor that can affect long-term profits.
Why Hillcrest 99 percent efficient design Matters
Power conversion often gets less attention than AI chips or networking gear. However, every watt lost as heat raises operating costs and puts more pressure on cooling systems.
A power system that stays above 99% efficiency wastes less electricity during nonstop use. For example, a big AI data center running thousands of accelerators all day can save a lot of energy each year with even a small boost in efficiency.
The Hillcrest’s 99 percent efficient design seeks to cut switching losses, a main challenge in high-power electronics. Using less energy means less heat, which also lowers cooling needs. These savings add up over the years.
For operators running facilities that use hundreds of megawatts, better efficiency is a clear financial benefit, not just a small engineering improvement.
Understanding 800V AI data center power architecture
The move to 800V AI data center power shows how the industry is responding to higher computing needs. AI servers now need much more power than older enterprise hardware.
As graphics processors get stronger, power systems must deliver more energy while keeping losses low. Higher-voltage systems do this by lowering the current for the same power, which cuts losses in cables and components.
The idea of 800V AI data center power architecture is consistent with broader field trends that concentrate on efficiency, scalability, and less complex systems. Instead of using many conversion steps that each waste energy, engineers now look for designs that keep efficiency high from the energy source to the computing hardware.
Hillcrest’s research adds to this discussion by supplying a design made for the next wave of AI deployments.
How single-stage ZVS architecture Improves Efficiency
Zero Voltage Switching, or ZVS, lets power electronics switch when voltage is near zero instead of at its peak. This greatly cuts switching losses, which are a major source of inefficiency in today’s power converters.
The single-stage ZVS architecture described in Hillcrest’s technical publication removes intermediate conversion stages that traditionally consume additional energy.
Having fewer conversion steps brings multiple practical benefits.
Heat generation declines.
Component stress decreases.
System reliability can improve because fewer components experience continuous high-power switching.
Maintenance requirements may also decline over the lifetime of the installation.
The aim is not just to get high efficiency in the lab. The real goal is to keep that efficiency under real-world conditions, even as AI workloads change during the day.
The Investment Case for next-gen data center power
Investors are realizing that the economics of AI depend on more than just how well the chips perform.
As hyperscale data centers grow, utilities in North America are talking about higher electricity demand. Rising rates, limited transmission, and bigger infrastructure investments all add to the uncertainty about future operating costs.
This makes next-generation data center power a more important way for companies to stand out.
When looking at AI infrastructure companies, analysts now check a number of operational metrics in addition to processor performance.
Power utilization effectiveness remains important.
Cooling efficiency continues to receive notable attention.
Electrical conversion efficiency is now an important measure because it directly determines ongoing operating costs.
Technologies that improve these metrics make a stronger case for growing AI infrastructure, even as utility costs rise.
Why the Hillcrest technical paper AI Could Influence Infrastructure Decisions
Technical papers don’t usually move markets on their own. Still, they help build engineering credibility and offer independent proof for new technologies.
The Hillcrest technical paper AI gives investors, engineers, and planners an opportunity to review detailed technical methods instead of just trusting marketing claims.
As AI facilities keep growing, decision-makers are increasingly asking for documented performance data before they adopt new power technologies.
Engineering publications fulfill several purposes.
They demonstrate technical transparency.
They explain design methodologies.
They encourage peer evaluation.
They establish performance expectations for commercial implementation.
If future commercial projects confirm the reported efficiency numbers, this publication could boost confidence in Hillcrest’s technology for developers looking for better electrical systems.
AI Infrastructure Economics Go Beyond GPUs
People often focus on advanced AI processors from companies like NVIDIA or AMD. But it’s the supporting infrastructure that decides if those processors run cost-effectively.
Power delivery, cooling, electrical networks, backup systems, and facility design all play a role in the total cost of AI computing.
Every gain in efficiency helps lower total operating costs over the life of a data center.
If a large operator runs several facilities using hundreds of megawatts each year, even small gains in efficiency could save millions on electricity over time, lower cooling needs, and help meet green objectives.
This bigger picture shows why Hillcrest’s 99 percent efficient AI design is worth attention from both engineers and investors.
The technology tackles a growing challenge in the AI industry: balancing fast-growing computing needs with rising electricity costs.
Viewing Ahead
AI infrastructure will keep growing as companies use more advanced models that need bigger computing clusters. This growth puts new pressure on the electrical systems that support these workloads. Developments like Hillcrest’s 99 percent efficient design, 800V AI data center power, next-gen ZVS architecture, single-stage ZVS, and the Hillcrest technical paper show that technology efficiency is still key to AI’s economic future. As utilities face higher demand and operators look for lower costs, better power architecture could become as important as better computing hardware.
You have probably noticed the prompt by now. Google, Apple, Amazon, your bank — they are all asking you to set up a passkey. Some accounts have already stopped asking for passwords entirely. If you have been dismissing these prompts without knowing what you are agreeing to or skipping, this is the article that explains what is actually different and whether it matters for your accounts.
The short answer: in the passkeys vs passwords debate, passkeys are more secure in almost every measurable way. The longer answer involves understanding why, where passwords still have the edge, and what the transition looks like for the average person managing dozens of accounts across devices they do not fully control.
Here is the full picture.
What a Password Actually Is — and Why It Keeps Failing
A password is a shared secret. You know it, and the website knows it. When you log in, you send the website your secret, it checks it against what it has stored, and if they match, you are in.
That model has one structural problem that no amount of complexity requirements, mandatory resets, or two-factor codes has solved: the secret exists in two places. Your memory — or your password manager — and the website’s database. When that database gets breached, your secret is exposed. When a phishing site tricks you into typing it, your secret is gone. When someone watches you type it over your shoulder or on a public camera, it is compromised without you knowing.
The average person now manages roughly 250 passwords — a number that has doubled in the past five years. That growth has made the fundamental problem worse, not better. More passwords means more reuse, weaker choices, and more exposure across more sites.
A third of consumers experienced a compromise or breach notification in the past year — and that is only the cases people found out about. Password breaches are often discovered months or years after the fact.
What a Passkey Actually Is — Without the Jargon
A passkey is a cryptographic key pair. When you create a passkey for a website, your device generates two mathematically linked keys: a public key that the website stores, and a private key that never leaves your device.
When you log in, the website sends a challenge. Your device uses your private key to sign it and sends back the signature. The website verifies the signature with the public key it already has. You are authenticated — without ever sending a password, without the website storing a secret that could be stolen, and without anything leaving your device that an attacker could use.
The practical experience is simpler than the explanation. You tap a button, use your fingerprint or Face ID to confirm it is you, and you are in. No typing. No remembering. No two-factor code to wait for.
The average passkey sign-in finishes in 8.5 seconds versus 31.2 seconds for the password-plus-code routine most people still use. That speed difference is real, and it adds up across dozens of logins every week.
Passkeys vs Passwords: The Direct Comparison
Passkeys
Passwords
Phishing resistant
Yes — by design
No
Can be stolen in a breach
No — private key never leaves device
Yes — if database is breached
Can be reused across sites
No — unique per site
Often yes
Login speed
~8.5 seconds
~31 seconds with MFA
Login success rate
93–98%
32–63%
Works on legacy systems
No
Yes
Recovery if device is lost
Via backup device or recovery code
Password reset email
Requires remembering something
No
Yes
The security advantage is not marginal. It is structural. Passwords can be phished, reused, guessed, leaked, or cracked. A passkey cannot be phished because there is nothing to steal — the private key never travels over the network. It cannot be reused because each passkey is mathematically tied to one specific site. It cannot be cracked from a server breach because the server only holds the public key, which is useless to an attacker without the private key on your device.
The Real-World Numbers Behind Passkeys in 2026
Passkeys have moved well past the experimental stage. The adoption data from 2026 is specific.
The FIDO Alliance counts 5 billion passkeys in active use as of 2026.
Consumer awareness has climbed to 90% — up from 75% in 2025 — with only 7% of respondents saying they are not familiar with passkeys at all. Seventy-five percent of users have enabled a passkey on at least one account.
Amazon reported 465 million customers now using passkeys, with passkey sign-in roughly six times faster than traditional password login.
Microsoft’s 2024 passkey rollout data showed passkey users signing in eight times faster than those using passwords with MFA, with a 98% success rate compared to 32% for passwords.
Organizations deploying passkeys report an average 73% reduction in sign-in time and 81% reduction in login-related support tickets.
Companies that have completed passwordless adoption report a 91.6% drop in security incidents and 57.3% fewer help desk calls.
These numbers are not projections. They are reported outcomes from live deployments at scale.
Google Passkeys — How They Work and Why They Matter
Google’s passkey implementation is the one most US users will encounter first, because Google accounts touch so many other services. Understanding it is worth the three minutes it takes.
When you set up a passkey on your Google account, the private key is stored in your device’s secure hardware — the Titan chip on Pixel phones, the Secure Enclave on iPhones, the TPM on Windows laptops. It is not accessible to apps, not accessible to Google, and not transmitted over any network.
When you sign in, Google sends a challenge to your device. Your biometric — fingerprint or face — unlocks the secure hardware, which signs the challenge. Google verifies the signature. Done.
What makes Google’s implementation particularly important is scale and cross-device sync. If you use Chrome and have Google Password Manager enabled, your passkeys sync across your signed-in devices through end-to-end encrypted storage. Log into a new phone and your passkeys come with you. This solved the single biggest practical complaint about passkeys: what happens when you get a new device.
Google also supports cross-device authentication — if you need to log in on a computer that does not have your passkey, you can use your phone to authenticate by scanning a QR code. The private key still never leaves your phone.
I Still Use Passwords. Do I Need to Switch Right Now?
Not immediately — but the case for switching where you can is strong, and the practical barriers are lower than most people expect.
Here is the honest situation: 93% of users still type passwords every day in 2026, and most legacy systems, government portals, and smaller services have not added passkey support yet. Passwords are not disappearing this year. You will be managing both for the foreseeable future.
What you can do right now:
Enable passkeys on the accounts that support them and are highest value targets. Google, Apple ID, Microsoft account, Amazon, PayPal, GitHub — these are accounts where a breach causes significant damage. All of them support passkeys. Setting one up takes under two minutes per account and immediately eliminates phishing risk for that account.
Keep using a password manager for everything else. A password manager with strong, unique passwords per site is still meaningfully better than reusing passwords or using weak ones. The manager handles the memory problem; the unique passwords handle the breach problem.
Do not disable two-factor authentication while transitioning. If an account does not support passkeys yet, two-factor authentication is still your best available protection. The risk with passkeys is not that they are weaker — it is that accounts without passkeys are still vulnerable, and removing 2FA from those accounts in the meantime creates exposure.
Where Passwords Still Win
Being honest about where passwords have the practical edge matters, because the complete picture is more useful than a one-sided comparison.
Universal compatibility. Every device, browser, and system on earth accepts a password. Passkeys require hardware support, OS integration, and a service that has implemented the standard. That combination exists on most modern consumer hardware but is far from universal.
Shared account access. If you share login credentials with a family member or colleague — a streaming service, a shared work tool, a household account — passwords make that straightforward. Passkeys are tied to individual devices and biometrics. Sharing access requires additional setup.
Legacy system access. Corporate VPNs, older enterprise software, government portals, and utility company logins are often running authentication systems that are years or decades old. Passwords are what they accept. That is not changing quickly.
Recovery simplicity. Forgetting a password is annoying but recoverable — reset link to your email, new password set, done. Losing access to all devices associated with a passkey is more complicated, though the major platforms have improved recovery options significantly in 2025 and 2026.
What Happens If You Lose Your Phone — The Passkey Recovery Question
This is the question most people ask before switching, and it deserves a direct answer.
Modern passkey implementations have addressed device loss with three mechanisms that work together.
Cloud sync. Apple syncs passkeys via iCloud Keychain, encrypted end-to-end. Google syncs via Google Password Manager with the same encryption model. If you lose your iPhone but have another Apple device, your passkeys are already there. If you lose your Android but log into a new one with your Google account, your passkeys sync over.
Backup device. You can register a passkey on more than one device. If you have both a phone and a laptop set up with passkeys for an account, losing one device does not lock you out — you still have the other.
Recovery codes. Most major services that support passkeys also provide one-time recovery codes when you set them up. These are the equivalent of a spare key — store them somewhere offline and physically secure, and you have a fallback that does not depend on any device.
The scenario where passkeys become a problem is someone who has set up passkeys on a single device, has no backup device registered, discards their recovery codes, and loses or destroys that device. That is a real but avoidable situation. The fix is registering passkeys on at least two devices.
Passwordless Login — What It Means When a Service Goes Fully Passwordless
Passwordless login means the service has removed the password option entirely, not just offered passkeys as an alternative. Microsoft began doing this for personal accounts in 2025. Google is pushing consumer accounts in the same direction.
The distinction matters because many services are currently in a hybrid state — they support passkeys but still allow passwords as a fallback. That hybrid state preserves the password vulnerability because attackers can attempt to social-engineer a password reset on an account that technically has a passkey set up.
True passwordless eliminates that attack surface. There is no password to reset, no SMS code to intercept with a SIM swap, no security question to guess. The only way in is through an authorized device, authenticated with your biometric.
Currently 87% of organizations still use passwords for customer-facing authentication, while only 2% believe passwords adequately balance security and user experience. That gap between what organizations know and what they have actually changed is where the transition is happening now.
Passkey Security for Businesses — What the Enterprise Numbers Show
For anyone managing security for a team or organization, the business case for passkeys has shifted from security argument to financial argument.
Deploying organizations see an average 81% reduction in login-related support tickets — that is a measurable IT cost reduction that shows up in budget lines, not just security posture.
Among organizations that have rolled out passkeys, reported benefits include improved security confidence at 47%, faster logins at 45%, better employee satisfaction with IT at 43%, fewer password-reset tickets at 35%, and reduced phishing incidents at 32%.
The 91.6% drop in security incidents after passwordless adoption addresses the most expensive category of IT cost — breach response, legal liability, customer notification, and remediation. For most organizations, a single avoided breach covers the cost of passkey deployment many times over.
68% of organizations are now deploying, piloting, or rolling out passkeys for employee authentication. The holdouts are primarily dealing with legacy system compatibility, not skepticism about whether passkeys work.
Frequently Asked Questions
1. Are passkeys actually safer than passwords with two-factor authentication?
Yes, in most real-world attack scenarios. Two-factor authentication significantly improves password security but does not eliminate phishing risk — attackers can capture both the password and the 2FA code in real time using adversary-in-the-middle attacks. Passkeys are phishing-resistant by design because the private key never leaves your device and the authentication is cryptographically bound to the specific legitimate website. A fake site cannot capture anything useful.
2. What if a website I use does not support passkeys yet?
Use a strong, unique password stored in a password manager, with two-factor authentication enabled if the site supports it. Do not disable existing security measures while waiting for passkey support to arrive. Check the FIDO Alliance’s passkey directory at passkeys.directory for an updated list of sites that currently support them.
3. Can someone use my passkey if they steal my phone?
Not without your biometric or PIN. The passkey itself is stored in the device’s secure hardware and requires authentication — fingerprint, face, or device PIN — before it can be used. A stolen locked phone does not give an attacker access to your accounts. This is meaningfully different from a stolen password, which gives immediate access without any hardware required.
4. Do passkeys work across different operating systems?
Cross-platform passkey use has improved significantly in 2025 and 2026 through the FIDO Alliance’s credential exchange protocol. Apple, Google, and Microsoft passkey ecosystems now support cross-device authentication where you use one device to authenticate on another. Full interoperability — moving a passkey from an Apple device to a Google device seamlessly — is still maturing but closer than it was a year ago.
5. Should I delete my passwords after setting up passkeys?
Not yet, and not for accounts where passkeys are an alternative rather than the only option. The risk is that if something goes wrong with your passkey access and the service still allows password login as a fallback, an attacker could use the password route. The cleaner solution is a long, complex, unique password stored in your password manager that you never type — effectively a dead credential that is technically still there but unusable in practice.
The Bottom Line
Passkeys win decisively on security and speed. They are phishing-resistant by design, cannot be reused across sites, and cannot be stolen in a server breach the way a password hash can. The numbers from Microsoft, Google, Amazon, and the FIDO Alliance all point in the same direction.
Passwords win on one thing only: they work everywhere, with everything, right now. That universality is real and it matters for the long tail of accounts that have not caught up to passkey support.
The practical strategy for 2026 is not choosing one or the other — it is using passkeys on every account that supports them, starting with the accounts where a breach would do the most damage, and keeping a password manager with strong unique passwords for everything else. That combination gives you the security benefits of passkeys where they are available and the best available protection where they are not.
Consumer awareness has hit 90% and the adoption data shows that when passkeys are offered, users adopt them. The transition is happening — not as a single event but as an account-by-account migration that will play out over the next several years. Getting ahead of it on the accounts that matter most is the move that makes the most sense right now.
