News - STARTUP HAKK

Google Stock Drops as AI Threatens Search Revenue

Spencer Thomason — Thu, 29 Oct 2026 17:17:23 +0000

Google is experiencing one of the worst days in its financial history. After releasing earnings, the stock fell sharply and closed with a huge loss. Only one company has ever lost more in a single day: Nvidia, on the day Deep Sea came out of China. The core issue is search—the business that has long paid for everything at Google. Search missed expectations, and by a lot.

When analysts asked why, Google’s business chief pointed to the company’s own AI. AI mode answers questions directly, and Google makes less money on those answers than on a traditional page of ads. In some cases, it can actually cost the company money because of the heavy spending required to build the system. Google built the technology that is now eating into its own primary revenue source.

Search Growth Goes Flat While Spending Surges

For 25 years, Google relied on one main money machine: search ads. That revenue funded YouTube, Android, AI labs, and more. This quarter, that machine essentially stopped growing. Search generated about $63 billion in the spring quarter and about $63 billion this quarter—flat. At the same time, Google is spending more cash than it brings in, one of the first times this has happened in the company’s history, as it builds AI data centers for technology that will generate less revenue.

The situation creates a trap. If Google does not push AI forward, competitors such as ChatGPT take market share. If Google does push AI, it reduces the profitability of its own search ads. Either path leaves the company, and its advertisers, paying the bill. Some advertisers say they have felt the pressure for months.

It is not entirely negative. Google Cloud more than doubled. The Gemini app passed 1 billion users. Some analysts are calling the stock a buy opportunity. Both sides of the story matter, though this is not financial advice.

A Brutal Earnings Reaction and Heavy AI Investment

Alphabet’s earnings were brutal—the first double miss in four years. The company missed last quarter and missed again this quarter. AI spending has been extreme. Long-term debt rose 479% in a year, the most Google has ever spent. Free cash flow is negative by both percentage and amount, and 2027 CapEx guidance is significantly higher. The stock dropped more than 10% after hours, with one-day declines reported around 11% and broader talk of a 13% fall that would rank among the largest post-earnings drops in the company’s history.

Berkshire Hathaway is investing an extra $10 billion into Alphabet, deepening its AI bet. Alphabet reached an agreement to sell 10 billion shares to Berkshire in a private placement. The move is part of a broader $80 billion stock announcement by Alphabet and continues a position Berkshire has been building rapidly over the past three quarters.

Search Is No Longer the Majority of Google’s Revenue

For the first time, search is less than half of Google’s money. A chart of search as a percentage of the business versus the rest of the company shows the lines crossing. The rest of the business is now larger than Google search. Some view this as positive because Google has long wanted to diversify beyond its main money machine. Cloud revenue is a key driver of the change. The cost, however, is the pressure on the core search business that has sustained the company.

Advertisers Report Declining Quality and Higher Costs

A person who spends $500,000 a month on Google ads posted that search growth was artificial and extremely unhealthy for Google’s long-term business. The post reached more than 4 million people on the same day Google had previously suffered its worst day. Search volumes are declining as legacy search is increasingly replaced by non-monetized large language model queries. Many people are turning to AI instead of traditional search.

Google’s response, according to that advertiser, has been to manufacture revenue growth through short-sighted, customer-hostile tactics. The company has charged advertisers more for lower-quality clicks, including clicks they did not want and did not approve. For years, Google used a second-price auction: if an advertiser bid $5 cost per click and the next highest bid was $1, Google charged roughly $1.01. This encouraged advertisers to enter their true maximum willingness to pay. Recently, Google moved away from that model and began charging advertisers as much as their bid and budget caps allow, regardless of other bids. It is described as a short-sighted cash grab that harms the long-term health of the advertising ecosystem and may push people toward other search providers.

Google also reduced keyword targeting precision. Advertisers once could select precise search phrases. Now, even with the strictest match settings, Google shows ads across thousands of unrelated keywords labeled as “exact match close variant.” The result is that advertisers are billed for clicks that are irrelevant to their business and their intended audience.

Google operated a benevolent monopoly for most of 25 years. The value it captured from search was only a small fraction of the value it created, supporting expectations of strong earnings growth. That is no longer the case. At the altar of AI CapEx, Google is sacrificing the golden goose. Paid click search volume has significantly deteriorated and decelerated—described as 13% paid and 5% cost per click quarter over quarter.

Gemini 4 Carbon Release and User Counting

Google tried to offset the earnings news by releasing Gemini 4 Carbon. Benchmarks look strong, and even Elon replied that it was impressive. Most people cannot actually test it yet. Critics on Hacker News noted that the Gemini app on a phone, despite high ratings, remains difficult to use effectively for real prompts. Google can claim 1 billion users in part because Gemini is woven into Gmail, search, and other Google apps. Every search that surfaces a Gemini response counts toward that total, even if users never deliberately open the Gemini app.

Meta Ads Catch Up as Search Growth Flattens

Meta ad revenue has caught up with Google search revenue. Meta ads have grown faster than Google search every quarter since Q1 2023 and are likely to surpass Google search next quarter. Google search revenue has still grown overall but has largely flattened near $63 billion over the past year. Meta is closing the gap quickly.

Why Google Held Back on Its Own Breakthrough

Google originally developed the transformer ideas that ChatGPT later scaled. When the transformer papers were published and peer-reviewed, it was fairly common knowledge. Google DeepMind led in AI for years, yet the company did not release a ChatGPT-scale product. The fear that AI would cannibalize search—exactly the outcome now visible in the revenue charts—kept Google from being the one to launch it. The current market result is the scenario the company tried to avoid.

Many users are already shifting. The host’s wife, for example, uses Gemini extensively because it is built into Google apps and is free. She rarely performs traditional Google searches anymore and is part of the cannibalization already underway. It remains to be seen how the picture changes once Google weaves ads more deeply into Gemini.

Looking Ahead

Google’s once-dominant position in search revenue is under pressure from its own AI tools and from faster-growing competitors such as Meta. The company is spending heavily on AI data centers while search revenue stays flat and free cash flow turns negative. Berkshire’s additional investment shows continued belief in the AI strategy, yet the near-term earnings reaction has been severe. Advertisers are already adjusting to higher costs and lower-quality traffic. Whether the diversification into cloud and AI ultimately replaces the old search money machine, or simply accelerates its decline, is the central question the numbers now pose.

The post Google Stock Drops as AI Threatens Search Revenue first appeared on STARTUP HAKK.

Microsoft Offers Buyouts to 60,000 Employees Amid AI Spending

Spencer Thomason — Wed, 28 Oct 2026 18:17:06 +0000

Microsoft has made roughly 60,000 employees eligible for a voluntary buyout, one of the largest such programs in its history. The offer applies the same severance math to everyone, with no age cutoff, no tenure cutoff, and no consideration of performance. Eligible people can accept a package and leave by December 31 or decline and stay. Nobody is forced out.

This approach immediately raises a practical question: who is most likely to take the check? Strong engineers who know they can find another role, and those who already have an offer from an AI lab in their inbox, are expected to move first. Lower performers are more likely to stay. Microsoft has acknowledged this dynamic; a spokesperson told Bloomberg that in some cases the company will try to keep top talent even after they say yes.

Business leaders were allowed to decide whether their teams were included. Some teams building Copilot and core AI products, along with certain sales and data-center groups, opted out. In other words, people closest to the AI work Microsoft is pushing into its products looked at their own teams and concluded they could manage if a portion walked out.

Capital Spending and the Numbers Behind the Offer

Microsoft has planned about $190 billion in capital expenditure this year, mostly on chips and data centers. An accounting change moved the reported figure to roughly $175 billion. Either number, divided across a workforce of about 225,000 people, works out to roughly $785,000 to $850,000 per employee in a single year. At the same time, the company has booked substantial profits—figures cited in discussion range from $133 billion in profit to $137 billion in net revenue.

These figures sit alongside the buyout. The company is not claiming it has already achieved large efficiency gains from AI that allow it to shrink. Instead, the scale of recent capital spending has made the ongoing cost structure harder to sustain at current headcount levels. AI adoption in business is described as a multi-year process even under favorable conditions, and revenue at some AI labs remains modest relative to the infrastructure investment required.

This is the second buyout of the year. An earlier program in the spring targeted longer-tenured employees and saw roughly 30 percent uptake. Microsoft expects a lower acceptance rate this time. Even a 10 percent take-up would mean about 6,000 people leaving by the end of the year; 20 percent would mean 12,000. For context, the company cut about 4,800 jobs in July.

Who Qualifies and How the Offer Is Structured

The package matches Microsoft’s standard severance: base pay calculated by tenure and level, some stock vesting, and health coverage for a period. Eligibility is described as covering level 70 and below on Microsoft’s internal system. Level 70 on the engineering individual-contributor ladder is comparable to a distinguished engineer; the range includes principal engineers, partner-level technical leaders, directors, and senior directors. Titles vary by organization. The vast majority of the workforce falls at or below this threshold; level 80 (technical fellow) sits above it.

One report noted that the internal portal listing the offer was under heavy traffic and crashing, suggesting many people were checking their options quickly. The company has said the offer is not an admission that it simply needs fewer people overall, yet headcount has already declined and leaders expect fewer takers than in the spring program.

Insider Views on Bloat, Speed, and Agentic Work

People who have worked inside Microsoft describe a large organization with many small groups, layers of decision-makers, and silos. One concrete example given is the rollout of managed identity. The Entra team shipped the capability in 2017 as a clear security improvement that removed secrets from code. Getting the rest of the company to adopt it took years; most teams moved slowly because each had its own priorities. Only after major security incidents and the Secure Future Initiative did broad adoption occur. Today nearly every Microsoft app and service supports it, but the process spanned six years and multiple breaches.

In an environment with fewer people and heavier use of agents, the same kind of change could move across the company in months rather than years. Microsoft already has the customers, infrastructure, and scale. The argument from those who have been inside is that it needs fewer people focused more tightly on agentic work. Competitors are shipping at a pace the current structure struggles to match, so the company must become more nimble to stay competitive.

A former employee who left recently noted that after his departure his former team in Australia was let go, creating real disruption for colleagues and their families. He and others who have observed the company from the inside say Microsoft has been bloated for a long time even while remaining profitable. The current capital-expenditure level is what has changed the equation.

Workforce Composition and the H1B Question

Microsoft has thousands of H1B visa applications on file. Employees have questioned, in internal discussions that circulated, why the company is asking people to resign voluntarily while reporting high profits, and why it is not first reducing its H1B population before offering packages to U.S. workers. The program will affect workers in Washington state. Government officials, including JD Vance, have previously called out Microsoft as one of the heavier users of the H1B and related programs. The timing of a large voluntary buyout has brought those questions back into view.

A 60,000-person reduction, if realized at the upper end of eligibility, would put a large number of engineers into the job market. How many actually leave, which functions lose people, which roles are not backfilled, and where hiring continues will clarify what Microsoft is prioritizing.

Conclusion

Microsoft’s voluntary program gives the same exit package to high and low performers and leaves the decision to accept or stay with each eligible employee. Leaders of some AI-focused teams have already decided their groups can operate without a portion of their people. The capital spending required to support AI infrastructure is large relative to current headcount, adoption timelines for business use remain measured in years, and internal voices describe an organization that has long struggled with speed and coordination. Whether the result is mainly a reduction in bloat, a shift toward fewer people doing more agentic work, or something else will become clearer as acceptance rates, replacement decisions, and subsequent hiring patterns emerge.

The post Microsoft Offers Buyouts to 60,000 Employees Amid AI Spending first appeared on STARTUP HAKK.

Anthropic’s IPO: The $800 Billion Compute Cost Challenge

Spencer Thomason — Tue, 27 Oct 2026 19:06:20 +0000

Anthropic’s recent IPO filing contains a striking sentence about a day that has not yet arrived: “our run rate revenue was $ as of October 31st, 2026.” The company filed the document on October 26, meaning it already knows the Halloween figure but has not printed it. The same approach applies to the third quarter. July, August, and September remain blank in the version available to ordinary readers, even though the filing gives the third-quarter picture away through three separate indicators that all point to roughly the same number.

An AI critic online highlighted dates to watch from November 11 through the 16th. Those are the days when the blanks must be filled if the IPO is to proceed on the current timeline. The discussion below walks through what the filing shows, the deadlines that force an update, the compute commitments buried in the document, and the numbers that observers have already flagged.

The Convenient Cutoff and the Third-Quarter Gap

Anthropic’s newest full numbers in the S-1 stop on June 30. In the second quarter the company generated $11.6 billion in revenue and spent $5.6 billion on compute, producing its first operating profit of about $1 billion. Share price and the amount of capital to be raised are routinely left blank in initial filings; that is standard. What stands out is the decision to highlight third-quarter strength on the first page while leaving the corresponding revenue figure blank beside it.

Big investors reportedly heard a specific number in an October meeting: $16.8 billion. The only audience still missing that figure is the public that would buy shares on the first day of trading.

SEC Deadlines That Force the Numbers into the Open

SEC rules require IPO financial statements to be updated once they are 135 days old. June 30 plus 134 days lands on November 11, the last day an IPO could launch using only the June figures. A second clock applies to companies that began the process confidentially: they must file publicly at least 15 days before the roadshow. The public filing arrived on October 26, so the roadshow cannot begin before November 10. That places the practical window for updated numbers—and the earliest realistic business day for a debut—between November 11 and November 16. CNBC has reported that the IPO is slated for November, which aligns with the two-week window before any go-live date. Some observers note the possibility of timing the update just ahead of Black Friday.

Signals from the Filing and Outside Commentary

The S-1 itself notes that Anthropic delayed pre-training on a new frontier model for more than a month to strengthen defenses against theft of model weights. The company warns that security-driven pauses can delay revenue and cause it to lose ground to competitors.

Sawyer Merritt highlighted the compute agreement with SpaceX: Anthropic expects to spend $120 billion on SpaceX AI compute through 2029, with monthly payments rising as high as $3.1 billion, or $37 billion annually. SpaceX is expected to deliver all contracted capacity by December 2026. Either party can terminate the agreement with 90 days’ notice after the initial contract period ends in early 2027.

Gary Marcus compared the published trajectory through the second quarter with an extended view that includes the third. The version without third-quarter data shows a steeply rising line; the version that incorporates the third quarter shows deceleration relative to that rise. Second-quarter results marked the peak of heavy token usage; the third quarter appears to mark the start of a slower regime. The draft S-1 did not report the third-quarter figures, which lets the company emphasize the steeper line.

Other metrics in the filing are striking. Revenue rose 1,300 percent year over year, gross margins reached 85 percent, and revenue retention stood at 900 percent. Revenue per employee hit $6.6 million. Customers spend roughly nine times more after a year, and net revenue retention is approximately 900 percent. Three years ago the company generated $41 million in revenue over a six-month stretch; more recently it reached $16 billion. Nearly half of revenue—about 48 percent—comes through cloud marketplaces. The published run-rate formula multiplies the last 28 days of usage by 13 and adds subscription revenue. The document also suggests Cowork is being merged into Claude. Roughly one thousand employees are about to become liquid at the same time, potentially producing one of the largest waves of new startups and angel investment San Francisco has seen—larger, in the host’s view, than the Facebook IPO. Even an IPO valued at only a trillion dollars would create a substantial number of millionaires and billionaires.

One observer noted that Anthropic remains firmly unprofitable, yet compute costs as a percentage of revenue have fallen sharply, producing narrower losses. The IPO offers a chance to raise a large sum, but the company will still need to reach sustained profitability.

Hundreds of Billions in Future Compute Commitments

The core financial discussion begins on page 161 under liquidity and capital resources. As of June 30, Anthropic had signed roughly $587 billion in compute leases and minimum purchase commitments that had not yet begun. It has since signed an additional $245 billion, bringing total future compute costs to $800 billion. None of these amounts have appeared on the income statement. The company acknowledges that its compute providers also sell competing services. In the SpaceX example, a 90-day termination notice after early 2027 could remove capacity on which the business depends.

This is the rent-versus-buy problem. Anthropic has identified its most significant strategic decision as the move to substantial capital expenditure. So far it has spent zero on owned compute. It can no longer rely solely on leasing, whose magnitude and timing are uncertain, so it intends to build its own capacity. That shift will change the company’s cash-flow profile. Sustaining the growth trajectory visible in the published numbers still requires massive ongoing spend.

A Host Perspective on Hype, Competition, and Local Alternatives

Spencer of Startuphakk has been moving away from Anthropic’s products even while acknowledging that the models and tools (including Claude Code and Co-work) are strong. The preference is for local AI that keeps data inside the organization rather than sending it to the cloud. From that vantage point the expected sequence is heavy hype, a rush into the IPO, and a subsequent crash once later quarters reveal the difficulty of profitable operation. Anthropic and OpenAI lack the profitable side businesses of Facebook, Google, and Microsoft. Google already runs its own TPU stack; Elon is building processing capacity, has partnered with Jensen Huang for large-scale compute, and can draw on Tesla and SpaceX cash flows. Meta can fund its AI efforts from existing advertising revenue. Without comparable alternative revenue streams or cheaper owned compute, the pure-play AI companies face structural pressure.

Startuphakk itself builds custom software and treats AI as owned infrastructure rather than a bolted-on API. Its open-source platform OpenMonoAgent.AI runs as a terminal-native coding agent on local models, with zero API costs and full ownership. The broader point is that durable results come from solid engineering and controlled infrastructure, not from chasing the latest external model.

What to Watch When the Blanks Are Filled

The November 11–16 window is the moment the third-quarter and run-rate numbers must appear if the IPO is to proceed on the current schedule. Those figures will show whether the three indicators already in the filing converge on the $16.8 billion level reported to large investors, and whether the deceleration noted by outside observers is confirmed. At the same time the $800 billion in future compute commitments, the planned shift from leasing to ownership, and the still-narrow path to profitability will remain on the balance sheet. The filing has already laid out the scale of the commitments and the strategic pivot; the upcoming update will simply make the latest revenue line visible to everyone.

Conclusion

The November 11–16 window is when the missing third-quarter and October 31 run-rate numbers must appear if the IPO moves forward. Those figures will clarify whether they align with the $16.8 billion already reported to large investors and whether the deceleration noted by observers is confirmed. The filing has already disclosed the $800 billion compute obligations and the shift toward owned infrastructure. The host expects heavy hype around the IPO followed by a later correction once profitability pressures and competition from companies with other revenue streams and cheaper compute become clearer.

The post Anthropic’s IPO: The $800 Billion Compute Cost Challenge first appeared on STARTUP HAKK.

The PHP 2 Shell Exploit Was Almost Impossible—Until It Wasn’t

Spencer Thomason — Mon, 26 Oct 2026 17:45:14 +0000

A developer placed roughly $2,000 in Bitcoin on a PHP server still vulnerable to the PHP 2 shell holes and publicly dared the internet to take it. For about 19 hours, attackers crashed the server repeatedly, yet no one claimed the wallet. Then one person succeeded.

The episode raises a sharp question: was this the end of the internet for anyone still running affected PHP, or an exploit so difficult that only a highly skilled operator could pull it off? The details show why calm analysis matters more than panic.

Two Critical Holes in PHP Itself

PHP 2 shell refers to two critical vulnerabilities in PHP, tracked as CVE-2026-den-1618 and CVE-2026-den-9619. Both carry a critical score of 9.2. An attacker on the open internet can run code on a web server with no login required.

Patches shipped on October 21 for PHP 8.5.12, 8.4.27, 8.3.6, and 8.2.35. PHP 8.1 and older receive no fix from the PHP project. According to W3Techs, PHP powers about 69% of websites where a server-side language can be detected. A census counted more than 41 million PHP services still showing affected version numbers, making the exposed surface enormous.

A third issue, CVE-2026-1115-560, remains unfixed by the October 21 patches. Vendors such as AWS have advised a settings change that includes turning off PHP’s upload progress feature. As of CESA’s October 23 update, no successful attack in the wild had been observed.

What the Internet Got Wrong

Once the holes became public, GitHub filled with purported working exploits. Many of these were AI hallucinations. Security researcher Marcus Hutchins highlighted the same fake technique appearing across multiple repositories. That older method relied on a session upload progress attack from years earlier. It shared a feature name with the real chain but constituted an entirely different attack and was not PHP 2 shell.

The actual chain differs. According to the winner’s own write-up, the attack broke PHP’s memory during the upload progress step, leaked two hidden memory addresses, and then forged fake PHP data to take control of the server. Upload progress therefore forms part of the genuine sequence, yet it is not the old session trick that people copied into ChatGPT.

The Steal My Bitcoin Challenge

The operator left a real Bitcoin prize on an unpatched PHP 8.4.26 server. The pot began around $500 and grew through donations to about $2,021. Early on, the site could not be won at all because it served only plain HTML. Once the operator added a real PHP page, the fair clock started. The site was claimed after 19.28 hours, at 8:47 a.m. Arizona time on October 25.

Fence, the WordPress security company, outlined the precise steps required. The winner had to corrupt memory at exactly the right moment, leak two secret addresses from the server, defeat the protection that shuffles where code sits in memory, and forge fake PHP components to hijack the system. In the winner’s own lab, the technique needed at most 20 web requests and succeeded roughly seven times out of ten. It worked only on that exact server build. A Debian system reporting the same PHP version string failed because its memory layout differed; no shell was obtained.

A final plot twist concerned the money itself. The key stored on the server did not match the funded wallet. The winner still recovered the correct flag, and the operator paid the prize anyway, confirming confidence that the solve was genuine.

Hard Is Not the Same as Safe

The exploit proved very difficult for amateurs yet possible for one skilled person. Difficulty does not equal safety. The winner published a write-up, giving the next operator a map. Meanwhile, floods of fake proofs-of-concept can leave teams numb just as a real attack arrives.

Anyone running PHP on the open internet should check whether their server remains on a version that still allows unauthenticated code execution. Recommended actions include patching to PHP 8.5.12, 8.4.27, 8.3.6, or 8.2.35 and restarting the workers. Systems still on 8.1 or older should be removed from the open internet until they can be upgraded. Random AI-generated exploit repositories should be treated as guilty until proven otherwise. Settings related to the third bug deserve review, and teams should not wait for GitHub star counts before acting.

Moving from Awareness to Action

This episode illustrates why security cannot remain an afterthought while teams ship quickly. Founders focused on product-market fit often move fast with code, yet familiarity can create blind spots. A single SQL injection, exposed secret, or broken access control can place a company in the news for the wrong reasons.

Startuphakk Security offers practical starting points. Two free AI-powered services deliver engineer-verified findings within an hour. The AI site scan accepts a URL and runs non-destructive, security-focused playbooks against the live attack surface, looking for injection flaws, broken access controls, cross-site scripting, and open redirects; the free tier returns the top three highest-severity findings with evidence. The AI code scan review accepts a zipped codebase (no GitHub access required) and examines every file for hard-coded secrets, vulnerable dependencies, injection flaws, and broken access control; the free tier again supplies the top three issues plus clear remediation guidance. Both run on a private AI harness rather than a public cloud service.

Automated scans catch low-hanging fruit quickly, yet they do not replace humans who think like attackers. When deeper work is required, engineer-driven penetration tests, architectural assessments, real attack simulations, and prioritized remediation roadmaps address the risks that matter when money or compliance is at stake. Security, handled correctly, need not slow progress; it can become the reason teams move faster without constant concern.

The PHP 2 shell story shows both the difficulty of a high-end exploit and the continuing size of the vulnerable population. Checking versions, applying available patches, and verifying settings remain the immediate steps for anyone still exposed.

Conclusion

The Bitcoin challenge showed that PHP 2 shell is difficult but not impossible once the right conditions and skill are present. The large number of still-affected services, the gap between fake public exploits and the real chain, and the existence of a published write-up mean the practical risk remains real for anyone still exposed. Patching the supported versions, removing older PHP from the open internet, and verifying related settings are the direct steps the situation calls for.

The post The PHP 2 Shell Exploit Was Almost Impossible—Until It Wasn’t first appeared on STARTUP HAKK.

Unauthorized AI Agents in Banks: Who Is Responsible?

Spencer Thomason — Fri, 23 Oct 2026 19:07:56 +0000

A viral Axios report revealed that major US banks are finding AI agents inside customer accounts. The post, shared by reporter Madison Mills, quickly surpassed millions of views, thousands of likes, and thousands of reposts. Many readers saw the phrase “unauthorized AI agents” and imagined external hackers forcing their way into bank systems. In most cases, that is not what occurred.

Customers themselves entered their usernames and passwords into AI tools and instructed the bots to handle financial tasks. Banks later detected these agents contacting call centers and performing actions that did not always appear to match clear user requests. The discovery has prompted banks to restrict access while executives prepare for scenarios in which customers later claim they no longer control their own funds.

What the Banks Actually Observed

Major US banks, including Morgan Stanley, JP Morgan, Bank of America, and Citi, have been detecting third-party AI agents operating inside customer accounts. Some of these agents have attempted actions that appear independent of the instructions the account holder originally gave. Bank executives told Axios they are racing to limit access because AI is moving faster than their existing systems can respond. Several banks are already preparing for worst-case and “day-after” situations in which a customer discovers money is missing after giving an agent full login credentials.

Banks struggle to tell human users from AI agents. They also have difficulty determining whether a given agent was authorized by the customer or is operating more like a rogue process. Morgan Stanley has stated it has not seen rogue AI agent access on its platform, while other institutions are actively blocking or restricting outside agents. Chase has advised customers not to give their username and password to anyone. On the same day the Axios story circulated, One Password released a tool that lets agents sign in without ever seeing the actual password—an acknowledgment across the industry that password sharing is the core issue.

A Recent Timeline of Agent Activity and Bank Reactions

On September 8, Metamuse launched a personal AI agent capable of booking, buying, and negotiating on a user’s behalf. The app quickly reached the top of the Apple US App Store. Bank stocks fell the same day as investors considered the possibility that agents moving idle cash could affect bank revenue.

On October 13, JP Morgan CEO Jamie Dimon said on an earnings call that the bank does not want rogue agents and that there could be thousands of them. He described the situation as a liability transfer.

On October 21, the security company Straker published a report describing a crime group that used AI agents to target more than 400 banks and payment companies in Latin America.

Shortly afterward, Axios published the broader scoop on agents appearing inside US customer accounts. Most of the agents the banks identified had been let in by the customers themselves using their own passwords. The banks never approved these connections, even when the customer had consented. An agent that logs in with a customer’s credentials typically violates bank rules. Some of those agents contacted call centers or appeared to act without being asked. Axios reported no specific count of cases and no confirmed losses.

Two Very Different Public Responses

Public reaction split quickly. Bernie Sanders called the situation insane and argued that AI agents must be paused because, left unchecked, they could endanger the entire global financial system. That view treats the appearance of agents as an external threat requiring an immediate industry-wide halt.

David Sachs offered a different framing. He noted a clear difference between user-authorized agents—such as those from Muse, Instinct, or Grokbot—and true unauthorized rogue agents that could constitute cyber attacks. Most of the activity described in the Axios reporting appears to fall into the first category: customers sharing credentials so agents can act on their behalf. This may violate bank terms of service, but it is not a cyber attack. Sachs argued that banks should support user-directed agents rather than blur the line between authorized use and attacks. American Express, for example, has already authorized a connection inside Muse that lets users grant, limit, and revoke access. Read-only permission levels are described as a sensible interim step, and several banks are already moving in that direction. Full write access using raw user credentials is a higher-risk category that should be handled more carefully.

Why the Password Changes Everything

A bank password is not a limited permission. It does not merely allow someone to view a balance. It tells the bank that the person presenting the credentials is the account holder and may perform any action the account permits. When a customer pastes those credentials into an AI agent, the agent becomes indistinguishable from the customer. The bank cannot tell the difference, and under current legal treatment the law generally cannot either.

If an agent that has been given login credentials moves money in a way the customer did not intend, the liability rests with the customer. Jamie Dimon’s reference to liability transfer points to exactly this outcome: once the credentials are handed over, responsibility shifts. Under US banking rules, actions taken by a bot that has received the customer’s login generally count as authorized. If the agent hallucinates a transfer or takes an unintended step, ordinary fraud protections are unlikely to cover the loss. Schwab is already blocking unauthorized bots dozens of times a week, and more institutions are expected to tighten controls quickly—through read-only APIs for agents and human verification before money can move.

Tools such as Grokbot illustrate the practical reality. The agent runs on a computer or virtual machine that the user can access. Once the user logs the agent into the bank and supplies credentials, the agent is operating as the user. From the bank’s perspective and from a legal standpoint, that is the account holder.

Responsibility Versus Calls to Pause Everything

The claim that AI itself must be paused is treated as unrealistic. Local systems and the tools that can already be built and run on individual machines cannot simply be switched off. The more practical path is learning how to use the technology, how to set boundaries around it, and how to remain responsible for the access granted to it.

At Startuphakk, the same principles guide the construction of AI systems. Agents are designed to run locally so that customer data does not leave the user’s environment. A sandbox is enforced so the agent cannot escape the boundaries placed around it. The resulting framework has been open-sourced as openmonoagent.ai, allowing others to run a terminal-native coding agent on local models with no API costs and no telemetry. The goal is infrastructure that an organization owns and controls rather than a wrapper around someone else’s remote service.

Most companies facing these questions do not lack technology. They lack clear ownership of how that technology is integrated and governed. When AI is added, it is designed into the architecture and kept inside the organization’s own environment instead of being bolted onto an external API.

What Comes Next

Banks are already restricting outside agents and warning customers against sharing passwords. At the same time, some institutions are exploring controlled connections that let users grant limited access without handing over raw credentials. Read-only options are viewed as a lower-risk starting point, while full transactional access remains a separate category that requires greater caution.

The central fact remains straightforward. In the cases Axios described, the majority of the agents inside customer accounts were not the result of external break-ins. They were the result of customers voluntarily supplying the single credential that gives complete control of the account. Once that credential is given, the agent’s actions are treated as the customer’s actions. The resulting liability stays with the person who handed over the keys.

The post Unauthorized AI Agents in Banks: Who Is Responsible? first appeared on STARTUP HAKK.

PHP Vulnerability 2026: Why Patching Is Only Half the Battle

Spencer Thomason — Thu, 22 Oct 2026 19:20:25 +0000

Yesterday was the leak and the patch. Today is the cleanup. And the cleanup is where most companies get hurt, because they patch, feel safe, and never check if somebody got in.

That is the plain-English take on what unfolded around a long-running PHP memory problem that finally went public in mid-October 2026. The story stretches back to 2024, accelerates through an embargo, a honeypot detection, early disclosures, government site takedowns, and millions of attack attempts, then lands in boardrooms asking how the vulnerable packages were there in the first place.

How the PHP Flaw Moved from 2024 Audit to October Chaos

A security audit funded by Germany’s sovereign tech agency first turned up signs of a memory problem in PHP back in 2024. The team could not agree on a fix, and the work stalled and restarted for months. PHP later called the delay a structural failure.

Everything changed on August 11, 2026. A firm called Caliph and OpenAI both reported the same root cause on the same day. Wiz came in with more attack paths over the next two months. The PHP Foundation, the researchers, and the big distributors then worked under embargo to prepare a real patch.

On Saturday, October 17, Amazon’s Madpoint Honeypot—a fake server built to catch hackers—logged a PHP-to-shell attempt. It came from an IP on a commercial anonymized network. Nobody had gone public yet, and Amazon says it is still unclear how the actor got the details.

The Days the Embargo Unraveled

On Tuesday, October 20, around 8 a.m. Arizona time, two CVE IDs started circulating online with the wrong release time and a wrong score of 10. No technical details had leaked yet. PHP moved the release up a full day from Thursday to Wednesday at 7 a.m. Arizona/Pacific time, because distributors were still packaging the fix and an immediate drop would have left many people without easy update paths.

Later that same day, around 1 p.m. Pacific time, Caliph—one of the firms that found the bugs—published a video of its exploit working against real targets. The copy spread. PHP said the videos “contributed to accelerated development” of more exploits. Mitigation advice PHP had given a few critical infrastructure providers may have leaked as well.

Overnight from Tuesday into Wednesday, a researcher using a crypto-style name posted that he created both bugs and “lab/field tested” them. The post blew up into the millions.

Chain Guard Breaks the Embargo and the Patch Goes Live

Wednesday, October 21, at 2:41 Arizona time, Chain Guard published a built image containing the fix more than six hours early. PHP said Chain Guard “unilaterally broke the embargo” and that any attacker could now study the fix in binary form while every other PHP user was still waiting for the patch.

Chain Guard CEO Dan Lorex told The Register he did not leak it. His rule is coordinated disclosure “unless there’s evidence of active exploits in the wild.” He said people were already getting hit, so he made the call and made the fixed images free to everybody. He also said he woke up and saw people blaming his company for the whole mess.

At 7 a.m. Pacific time the embargo lifted. PHP shipped versions 8.5.12, 8.4.27, 8.3.36, and 8.2.35. It stated there are no complete mitigations; updating is the only safe choice.

Governments Pull Sites Offline as Attack Volume Spikes

Within minutes, governments started taking services offline. The Netherlands took the National Cyber Security Center offline, along with Port Rotterdam, the government portal, and other systems. Japan’s Prime Minister posted on X telling people to check if their public website runs PHP. In the UK, the cyber minister told Parliament that 32 government services were taken offline. The impact reached China and beyond.

The attack curve went vertical. Cloudflare said its firewall rules went live about nine hours before the release, including on the free plan. In the 10 hours before the announcement it saw thousands of attempts, tens of thousands in the first 45 minutes after, hundreds of thousands within two hours, and millions by the end of the day. The Cloudflare CEO’s post passed 1.5 million views that morning.

Amazon later said that after the patch dropped, PHP-to-shell was being broadly exploited in the wild and told everyone to start their investigation from October 17, not from Wednesday.

Millions of Attempts Are Not Millions of Break-Ins

Here is the twist almost nobody is saying out loud: millions of attempts is not millions of break-ins. Watchtower says the exploits being sprayed around mostly do not work or just crash the target. Germany’s cyber agency says it has seen attacks but no successful use, and that it is a little more complicated than expected to exploit. A lot of the news hype claims this is really easy to exploit. It is a real vulnerability, but exploiting it is not trivial, and at the moment it does not look like exploitation has been widespread, according to The Guardian.

As of that morning, census data still showed about 1.5 million internet-facing servers advertising a PHP version older than the fix. Voluntech says at least 1 million run versions PHP itself will not patch. CISA urged everyone to immediately take action: if you have older systems that cannot be patched, take them offline.

The researcher who posted overnight later fired back at critics. He said he did not do anything illegal, he had no contract with any program, and social contracts are meaningless in today’s world. A Middle Eastern threat intelligence account shared the reply; by morning the post had over half a million views.

Cleanup Reality and the Questions Boards Are Asking

A former CISA figure who goes by Blackroom Tech put the boardroom mood in one post that was already circulating in security circles: leadership is asking how they did not know they had all these vulnerable packages. For 30 years, software developers, CISOs, and system admins have been saying systems must be taken down to patch and stay secure. For decades, executives have said no—we need to keep moving forward, we will take the risk, we will pay the insurance claim. This has been the push-pull in the boardroom for decades. Then the story hits the press, the sky appears to be falling, and everyone asks why systems have to come offline.

In practice, some organizations that integrate with third-party PHP systems saw those systems taken down as soon as the news hit, then brought back once the patch was available. That is the correct behavior. Others had clients screaming about the downtime even though the responsible step had been taken. WordPress sites were among the larger footprints that needed rapid patching. When many people tried to update within minutes of the release, PHP servers kept timing out; responsible teams had to wait an hour or two.

Amazon’s threat intelligence noted that an embargo protects defenders only until someone else finds the same bug. Coordinated disclosure remains the right model, but defenders should not assume the public disclosure date is the earliest date of exploits. They need to go back further. The time from patch to working exploit is shrinking for structural reasons. In the past, most observed exploitation came after a public proof of concept; a small number of capable actors built exploits and everyone else waited.

What the Embargo Break and AI Change About Disclosure

The PHP Foundation said Chain Guard disregarded weeks of work by many parties toward a coordinated disclosure. Chain Guard had received the fix for the two critical issues under embargo like every other distributor, then published built images more than six hours early. From that point any attacker could study the fix in binary form.

The first signs dated to the 2024 German-funded audit, and the release turned into chaos: a flaw reference leaked on October 20, a security company published an attack video, and Chain Guard broke the embargo. The security of a language that powers banks, shops, and public services rests on a small group of developers. Those same public services that states had to shut down that morning rely on a handful of people to protect the servers.

On the researcher who published: he clarified that he did nothing illegal and had no contract or agreement with any program or platform, and that social contracts have no value. Legally he is right—there is no law forcing coordination before publishing. Coordinated or responsible disclosure has never been a law; it is a security community custom so affected parties can release an update before the exploit becomes available to everybody. It has been a gentleman’s agreement.

The speaker agrees 100 percent that the embargo period for zero-days should never be broken by trusted parties. At the same time, the security industry needs to reevaluate how much time these embargoes can realistically afford to wait now that cyber models exist. During a recent KVM escape vulnerability, it was possible to independently discover the issue within a few hours simply by giving a semi-redacted tweet from security researchers to a cyber monitor. The embargo was respected and details were not shared, but if that could be done, anyone motivated enough can do it. Embargo periods need to be measured in hours, not weeks or months. Disclosures now travel as fast as rumors. Humans have to sleep. AI can find a vulnerability and report it, but getting a fix deployed, coordinated, and out to distributors still takes human time. That line between what AI can do and what still requires people is becoming clearer.

The Uncomfortable Boardroom Conversation and the Practical Path Forward

This morning, boardrooms across many places are having uncomfortable conversations. Leadership asks how the organization did not know it had all these vulnerable software packages. Technical people who have been constrained for years by the same leadership—on patching, updating, documentation, and asset management—are the ones being asked to explain. How can you protect assets if you do not know how many you have, what they are, and where they are? Budgets have gone to other priorities. Production systems cannot come down in the middle of the day. Overtime is not approved. The result is that updates that require downtime do not happen cleanly.

Companies that brag about how secure they are often are not, in reality. That is not the fault of the technical staff; it is the fault of the people at the top who are supposed to be managing the business, including its security, and who rarely get called out. This is not the moment to lay off the people who will help clean up the vulnerabilities that still need patching in the coming days, weeks, and months. This incident is a precursor. More will come. Stop focusing on projects that do not protect what already exists and start investing in the people and the ability to move quickly. Nimble teams that can patch and respond are the ones that stay strongest. Security has to be first, not an afterthought. Listen to the cybersecurity staff.

The cleanup is not finished when the patch is applied. Organizations that treat the public disclosure date as the start of their investigation, or that patch and assume they are safe without checking for earlier activity, are the ones most likely to get hurt.

Conclusion

Yesterday was the leak and the patch. Today is the cleanup. Most companies get hurt in this phase because they apply the update, feel safe, and never check whether someone got in earlier. Updating remains the only safe choice, as there are no complete mitigations. Millions of attempts do not equal millions of successful break-ins, but the gap between public disclosure and the first real activity is shrinking. Organizations that treat security as an afterthought, refuse downtime for patching, and fail to know their assets will keep facing the same boardroom questions. Nimble teams that can move quickly and put security first are the ones positioned to handle what comes next.

The post PHP Vulnerability 2026: Why Patching Is Only Half the Battle first appeared on STARTUP HAKK.

Why DeepSeek’s Open-Source AI Strategy Could Win the Long Game

Spencer Thomason — Thu, 22 Oct 2026 14:00:54 +0000

DeepSeek follows a much simpler strategy than the American labs. The company builds models, open sources the models and the infrastructure, and provides inference to make money. In short, it builds it, gives it away, and sells the meter. That is the entire plan.

This stands in contrast to the multi-tier systems, subscriptions, licensing deals, and outcome-based pricing that define other approaches. One lab runs consumer and team subscriptions, a free tier paid for by ads and shopping, an API on top, licensing and IP deals, and outcome-based pricing. Its Pro offering alone now comes in one-hundred-, two-hundred-, and five-hundred-dollar-per-month subscriptions. Forecasts for that lab show it has burned twenty-five billion dollars this year—a negative burn of twenty-five billion dollars.

The question is which plan is working. Reports indicate that DeepSeek’s API business ran an 82 percent gross margin from January through July. The other lab’s gross margin in the first quarter was 39 percent. DeepSeek’s revenue was about ten times all of last year, and by September its run rate hit one billion dollars. It achieved this while handing out the models, the research, and, as of now, the engine they run on.

Selling the Meter While Giving Away the Recipe

American labs sell a subscription, a speed tier, an ad slot, a plug-in store, and a meter—the whole kitchen sink. DeepSeek just sells the meter. Its models are MIT licensed, so anyone can download them, run them, and even sell them against DeepSeek if they want.

One reply raised the obvious question: if inference is the revenue strategy, why open source the inference so everyone can use it for free? The response was that the revenue strategy is to be awesome. When everybody has the recipe, the only way to win is to run the best and cheapest kitchen in town.

The spending numbers highlight the difference. One lab expects to spend 665 billion dollars training and running its models through 2030. DeepSeek spent about 1.6 billion on AI infrastructure in the first seven months of this year. To be fair, DeepSeek still lost about 106 million over those seven months. One of these plans fits on a napkin.

Open Sourcing the Entire Stack

DeepSeek is effectively open sourcing the whole company at this point. It is sharing its product stack to help the community lower compute costs, reduce energy use, and make AI more affordable. The CEO is very loaded. This has caused a stir.

DeepSeek released DeepInfer, its in-house inference framework, now fully open source. It powers DeepSeek V4.1 Flash at scale with high throughput and low latency. The company is sharing its production stack so the community can lower compute costs, reduce energy use, and make AI more affordable. This is the long game.

DeepSeek is the biggest threat to the US frontier labs. A DHS order should come as no surprise. The US government bans DeepSeek, claiming it is because the servers are in China. It bans even US inference providers if they are using DeepSeek, and it bans anything built with or based on DeepSeek models. Someone higher up was convinced that the models themselves are calling home. Anyone running a decent firewall would see that if it happened. Many open-weight models have been run without anything calling home, and it would be visible if it did.

People think DeepSeek just makes cheap models distilled from Claude and ChatGPT. They are missing that DeepSeek is playing an entirely different game. The founder is the wealthiest frontier lab founder if you do not count certain others. He is wealthier than Dario Amodei and Sam Altman and has majority control. The company can play a much longer game than the other labs.

Look at the choices. It releases model weights under an MIT license. Others, such as Kimi and GLM, keep commercial conditions on theirs. DeepSeek is willing to let anyone run the model and compete with its own API. It publishes unusually detailed research. Its original DeepSeek R1 research helped every lab, including the Western ones—OpenAI, Anthropic, and others. It ruthlessly focuses on low inference costs. The company is clearly compute constrained, and its architecture choices are optimized for cost. It is forced to be efficient because it does not have the ability to throw unlimited amounts of compute at the problem the way OpenAI and Anthropic do. That forces actual efficiency.

American labs are wasteful. They are burning so fast because the VCs keep feeding them money. A founder who can afford to wait, take decades, who has already built a true challenger, and who is willing to put the model and research out in the open has a different position. The biggest constraint has been the chips. Huawei’s 950 DT is coming. When DeepSeek gets the chip, it is going to kick everybody’s butt.

The Ban and the Alternative

If America is worried about China winning, the answer is not a banned list. It is better American open-weight models. Banning it does not make better models. DeepSeek keeps shipping open-weight models. Anyone can download, run, and build on them. That is why it is everywhere. Ship the weights, or lean into the tools they are giving, see what they are doing, learn from it, and keep iterating. Banning is a stick-your-head-in-the-sand tactic, and the DHS approach is silly.

DeepInfer is the inference tool and engine released so people can run and build out their own inference. The company has given the tools: the original DeepSeek V3, the original DeepSeek R1, and now the DeepSeek Harness to run these models. It is not just giving the models. It is giving away free harnesses and the whole stack.

Sam Altman is watching DeepSeek slowly open source the entire AI stack. Today it open sourced DeepInfer, the framework used to run V4.1 Flash at scale. It includes optimized inference kernels, distributed inference, speculative decoding, async scheduling, structured outputs, and NVIDIA plus Ascend support. It is all Apache 2. DeepSeek has already released model weights, research, training recipes, its coding harness, and now the inference framework running V4.1 Flash. Every few months another layer of DeepSeek becomes public. At this rate, it is going to open source the whole stack piece by piece.

Chip Partnership and the Long Game

DeepSeek and Huawei are teaming up on a new AI chip designed to challenge NVIDIA. DeepSeek would be working with China’s biggest chip maker to reduce reliance on US chips. This comes a week after China’s CXMP chip progress sparked a global sell-off in stocks. If they can build out this chip and start to mass-produce it with what DeepSeek produces and what it has given away, they are going to become a very big challenger to the Western US frontiers.

It is official: DeepSeek is the actual OpenAI now. The simple plan is going to win the long game.

A Similar Open Approach

The same principles of playing the long game appear elsewhere. Open sourcing a harness shows how to build on AI by putting the work out in the open. The approach is local first, always—never pump customer data into the cloud. Customer data is the number one asset and should not be given away. Unlimited tokens change the game: stop and think about what you can run with AI if you let it run forever. Sandbox by default, and fully open source.

Installation of such a system is straightforward. On the inference side, run a curl script, select the option, and let it auto-detect hardware and dependencies. Then run the tunnel setup to create a secure tunnel, verify with email OTP, copy the relay config commands, and the GPU inference powerhouse is ready. This architecture keeps heavy local GPU inference completely separate from the coding agent via a secure relay. On the client side, open VS Code, search for the extension, install it, connect the relay, and the local setup is accessible from anywhere.

Most companies do not have a technology problem. They have a leadership problem, and they are paying for it in missed deadlines, failed integrations, and AI investments that delivered nothing. The businesses winning right now are not the ones chasing the latest AI trends. They are the ones who built on solid engineering and treated AI as infrastructure. They own control and integrate the software that actually works.

Custom software development covers database architecture, API design, system integration, and scalable infrastructure built the right way. When AI belongs in a solution, it is designed into the architecture, running in the customer’s environment—not a vendor’s cloud and not on someone else’s pricing schedule. An open-source platform that is a terminal-native AI coding agent running entirely on a local LLM delivers zero API costs, zero telemetry, and full ownership. As a fractional CTO, the same standard of leadership is available without the full-time executive costs: strategic architecture, hands-on delivery, and clear accountability.

DeepSeek’s approach of building, open sourcing, and selling the meter while focusing on efficiency and the long game continues to play out as more of the stack becomes public.

Conclusion

DeepSeek’s one-sentence strategy—build the models, open-source the models and the stack, and sell the inference—has produced an 82% gross margin, a $1 billion run rate, and progressive release of its full infrastructure while staying compute-efficient. The same simple plan is still unfolding as more of the stack becomes public and chip partnerships advance.

The post Why DeepSeek’s Open-Source AI Strategy Could Win the Long Game first appeared on STARTUP HAKK.

Why You Should Keep Your AI Local

Spencer Thomason — Tue, 20 Oct 2026 18:14:10 +0000

Technology is moving quickly. Models are releasing almost daily, after previously arriving weekly and then monthly. The speed is increasing, and everybody can feel it. A model like Qwen 3.8 27B was released a little over six weeks ago and remains an impressive option to run on local hardware. Comparing this model—which many now take for granted—against the most powerful models from a year ago shows how the numbers stack up. If the 27B model had come out a year ago, it would have been the best model in the entire world by a large margin.

This perspective highlights how much current AI capabilities, including what runs on accessible hardware, are taken for granted. Local AI remains worth pursuing even as some prices continue to drop and models grow more powerful. While some people discuss using frontier models, many others cannot or should not rely on them.

The Rapid Pace of Model Releases

Before, changes felt quick when they occurred yearly or every six months, with release cycles that were yearly. Even keeping up with Ubuntu releases twice a year once felt difficult. Now, content is released every single day. Going offline for three days could mean missing a handful of frontier models. Things arrive faster and faster, making it useful to zoom out and compare against just a year ago.

Local Models Compared to Last Year’s Frontier Options

A chart comparing frontier models from a year ago—such as GPT-5, Claude 4.3, and Grok 4—against Qwen 3.8 27B shows a striking difference. Some dismiss local hardware as not worth it, yet the comparison is against results from only one year ago. What is available now would have blown away anything the frontier models could have produced even a year ago. This illustrates how quickly local capabilities are developing.

Qwen 3.8 27B compared to all the state-of-the-art models from one year ago makes the point clear. If 27B had been released a year ago, it would have been the best model in the world by a large margin. One comment on the topic noted that a 27 billion model that runs locally is 35% more intelligent than the leading state-of-the-art model a year ago, and asked what will run locally next year.

GPT OSS 120 billion, with only 5 billion active parameters, was the eighth-best model in the world a year ago. It was described as a truly extraordinary model. By most standards now, a 27B dense model runs circles around it. In one year, a 120 billion model is being outperformed by a 27 billion dense model. This shows how quickly the landscape has shifted.

Local Hardware Remains Relevant

Some predict that hardware will go obsolete. Machines currently in use include one with a 1590 and another with a 3090. The 3090 Ti was a cutting-edge card six years ago and is still 100% relevant today. Claims that hardware will be dead in three years are false. Current setups run capabilities that would have been absolutely cutting edge a year ago. Even while discussing 27B, more powerful options are already running on the same hardware.

Models continue to get smaller yet more powerful. More is being squeezed out of local hardware every single day, which benefits end users. Looking ahead to what will run in a year is striking given this trend of smaller, more capable models.

Why Data Should Stay Local

One major reason to run AI locally involves data handling. Cloud platform documents may state that data is retained for only 30 days and that it is not used for training. This is a bold-faced lie, as is widely recognized. Data-related issues have appeared repeatedly. One recent example involved a fabled data set purchased from a Chinese LLM router containing six terabytes of data that included keys allowing access to over seven Chinese government entities and nine top Chinese firms. This came from a single data set.

Thinking that providers such as Anthropic are not training on user data is unrealistic. Time and again, issues have surfaced, including cases of OpenAI stealing from mathematicians. This is not new. As capabilities advance rapidly and power is squeezed into smaller boxes faster and faster, the decision to send data elsewhere carries ongoing risks.

Some point to options like Haiku 5.5 running nearly free as a reason not to build local setups. While the model is powerful, this approach still pumps data into the cloud. For someone with nearly 30 years in software development, that is not acceptable. People in regulated industries are often not allowed to do it either.

Four years ago, before the GPT-3 moment, proposing to stream an entire system, all databases, and every piece of customer data through a third-party service would have been seen as nuts. Now that AI is powerful, some treat it as acceptable. Best engineering practices that have existed for over three decades, four decades, and longer argue against giving this data away. Data should stay local.

Local-First Frameworks and Ownership

OpenMonoAgent.ai was launched as a framework supporting this approach. Base principles include local first, always unlimited tokens, and a zero-cost sandbox by default. It is open source so others can examine it. It is not meant to be a subscription. The infrastructure should be owned and sitting on a desk rather than rented.

Three major new features are launching on OpenMonoAgent. Regardless of the specific framework chosen—whether that one, another option, or a custom harness—running AI locally is worthwhile. OpenMonoAgent offers a practical place to start for those unsure where to begin. Many people in regulated industries cannot simply pump data into the cloud.

Treating AI as Owned Infrastructure

Most companies do not have a technology problem. They have a leadership problem, and they pay for it in missed deadlines, failed integrations, and AI investments that delivered nothing. The businesses winning right now are not the ones chasing the latest AI trends. They are the ones that built on solid engineering and treated AI as infrastructure they own, control, and integrate into software that actually works.

This means real custom software development: database architecture, API design, system integration, and scalable infrastructure built the right way, as it has been done for over 25 years. When AI belongs in a solution, it is built in rather than bolted on or wrapped around someone else’s API. It is designed into the architecture, running in the user’s environment rather than a vendor’s cloud or on someone else’s pricing schedule.

An open-source platform called OpenMonoAgent.ai demonstrates this model. It is a terminal-native AI coding agent running entirely on local LLMs, with zero API costs, zero telemetry, and full ownership. It is growing because serious engineers recognize real infrastructure when they see it. The same standard of leadership can be applied without full-time executive cost through strategic architecture, hands-on delivery, and clear accountability.

Organizations ready for custom software built to last, with AI integrated where it actually makes sense, can focus on technology leadership and AI as true infrastructure that produces real, ownable results.

Conclusion

Local AI has advanced to the point where a 27B model that runs on ordinary hardware would have outperformed last year’s leading frontier systems by a wide margin. At the same time, models continue to grow smaller and more capable, and hardware from six years ago remains fully relevant. Sending data to the cloud still means giving it away—something best engineering practice has advised against for decades. Running AI locally keeps that data under your control and treats the models as infrastructure you own.

The post Why You Should Keep Your AI Local first appeared on STARTUP HAKK.

Is the AI Bubble About to Burst? Demand, Financing Walls, and Open Weight Models

Spencer Thomason — Mon, 19 Oct 2026 17:56:26 +0000

The debate over whether AI is in a bubble has become one of the most active discussions online. A short clip of Ben Horowitz commenting on the topic received more than half a million views in just a couple of days. A post suggesting America may need to spend $3.5 trillion a year on AI by 2032 has been seen more than 250,000 times. The conversation is loud on both sides, and people keep returning to it month after month.

One side argues that demand is simply too high for this to fail. The other side says the financing does not add up and the required revenue numbers are unrealistic. A third factor has entered the picture: free and open-weight models that change the math for everyone involved.

The Bull Case: Sold-Out Compute and Unmatched Demand

Ben Horowitz has shared that Amazon CEO Andy Jassy told him Amazon’s AI compute is sold out beyond 2028. The shortage is severe enough that Jassy indicated they do not expect supply to return even into the early 2030s. Horowitz contrasted this with the dot-com era, when more bandwidth was built than consumers could absorb. This time, he said, there is more demand than supply.

Amazon has reinforced the point in public statements. AWS AI demand is outpacing capacity through 2027, and the demand already booked for 2028 is described as striking. Multiple reports point to the same conclusion: compute is constrained.

Horowitz has also argued that OpenAI and Anthropic threaten nearly every stock people own. He noted that the world has never seen companies grow this large this fast, raising the question of whether there is enough capital to finance them. At the same time, he expressed confidence in the demand for their potential IPOs. Another investor added that people need to own these companies precisely because they sell intelligence broadly, and nearly every business is built on intelligence.

Model performance rankings continue to shift quickly. In one recent stretch, OpenAI’s GPT-6 and Astra brought OpenAI level with Anthropic roughly three weeks after launch. Then Claude Opus 5.5 pushed Anthropic back into the lead within three weeks. Trackers such as Vercel, Ramp AI, and OpenRouter show the two companies staying neck and neck, with the lead changing every few weeks. Anthropic has told investors that quarterly revenue is up nearly 50 percent, and Opus 5.5 is widely regarded as a strong release. Yet people show little loyalty to any single model. Users move rapidly to whatever is currently best.

The Bear Case: Revenue Requirements and Financing Risks

The other side focuses on the scale of the required returns. Gartner has estimated that more than $3.5 trillion in AI infrastructure spending could be exposed to potential impairment by 2030 if current investment trends continue. To recover the investments, cover financing costs, and earn a 12 percent return on invested capital, data-center owners would need roughly $8.5 trillion in cumulative revenue through 2030. That would require growing nearly nine times as fast as the industry grew over the past decade just to break even.

A Columbia professor has put the total cost of the AI build-out at about $10 trillion from 2025 to 2035. Relative to the size of the economy, that is larger than the railroad, highway, or internet build-outs—and three times larger than the internet boom. Other figures put the needed annual U.S. spending on AI services at about $3.5 trillion by 2030, or 8.8 percent of GDP. No single product category has ever reached that share of GDP. The build-out itself is estimated at $10 trillion from 2025 to 2032, again larger relative to the economy than previous major infrastructure booms.

Financing has shifted toward joint ventures, private credit, and special-purpose vehicles. That structure makes risks harder to see. While it may be premature to label the system a clear risk, better measurement and disclosure are needed while capital structures are still forming. Roughly 80 percent of NeoCloud providers are expected to cease independent operations. AI providers are projected to spend nearly $1 trillion on infrastructure, with an additional $6 trillion forecast between 2026 and 2030. Current combined run rates for OpenAI and Anthropic sit around $125 billion. Some projections call for roughly 5x growth over the next 14 months, but recent monthly figures show revenue growth tapering rather than accelerating. Anthropic’s reported quarterly numbers have moved from about $4 billion in one quarter to $11 billion in the next and still remain well short of $100 billion annualized.

Corporations are beginning to pull back on AI spending rather than doubling down. Revenue appears to be trading back and forth between the leading companies rather than expanding in a straight line.

Open Weight Models Change the Equation

A third category is now affecting both the bull and bear arguments: open-weight and freely available models. Meta released a model capable of running on two Mac Studios (or equivalent high-end hardware in the 40,000–50,000 range). While that hardware cost is not trivial, it shows that capable models can already run outside the large closed providers. Google offers substantial free services and can fund much of its own infrastructure. The debt-to-income ratios for these companies differ sharply from those of pure-play labs that have raised large amounts of capital.

Frontier closed models still lead open-weight models by roughly 10 percent on many benchmarks, yet the gap is narrowing. Multiple new open-weight models have appeared in a single week. Businesses are learning that these models can deliver strong results on hardware they control. For the trillion-dollar revenue targets to be met, spending must reach deeper into corporate budgets. If companies instead adopt open-weight options and slow their paid usage, the path to those revenue numbers becomes steeper.

What History Shows About Technology Bubbles

Major technology bubbles have never eliminated the underlying industry. They have revalued market capitalizations. The video-game crash of 1983–1985 saw revenue fall more than 97 percent and nearly wiped out the new entertainment category. Nintendo later revived it. The same pattern appeared in the dot-com, fiber-optic, and clean-tech cycles. Even if AI valuations drop 20–30 percent around an IPO, the technology itself is expected to continue advancing. The industry is more consequential than video games, and AI revenue is not projected to fall 97 percent.

A Middle Path: Technology Stays, Financing Faces a Wall

All three perspectives contain elements that are currently true. Demand exceeds available supply today. That demand has been supported by venture capital that cannot fund the build-out indefinitely, so costs are likely to rise and some loosening will occur. Open-weight models are being released by organizations with different motives—ranging from large U.S. technology companies to others outside the U.S.—and those models are already usable.

The technology itself is not going to disappear. People predicting total collapse are overlooking the persistent demand and the rapid improvement in open-weight systems. At the same time, the circular financing that has propped up spending cannot continue unchanged. A funding wall is likely. Prices should keep rising. OpenAI and Anthropic are expected to reach an IPO eventually. When they do, the offering is likely to be difficult; valuations could drop roughly 50 percent once the economics are fully visible. Companies currently spending two to three dollars to earn one dollar face a model that has never worked indefinitely. Revenue, not hype, will determine the outcome.

The practical winners are likely to be smaller businesses that adopt open-weight models running on hardware they control. Capable systems are already producing results that would have seemed futuristic only a few years ago, including on hardware that is six to eight years old. The advantage sits with organizations that treat AI as infrastructure they own, integrate into their own software, and run in their own environments rather than relying solely on external APIs and pricing schedules.

The debate will continue because the numbers are large and the stakes are high. Demand is real and currently constrained by supply. The revenue required to justify the capital deployed is also real and extremely high. Open-weight models are already shifting where that demand lands. The most durable path is building systems that remain useful even after financing conditions change.

Conclusion

The technology is not going to disappear. Demand remains higher than supply today, yet the financing that has supported the build-out cannot continue unchanged, and a funding wall is likely. OpenAI and Anthropic are expected to reach an IPO eventually, and that offering is likely to be difficult once the economics—spending two to three dollars to earn one—are fully visible.

The practical advantage sits with businesses that treat AI as infrastructure they own. Open-weight models already deliver strong results on modest hardware. Organizations that learn to run and integrate these models into their own systems, rather than relying only on external services, are best positioned as valuations adjust and costs rise.

The post Is the AI Bubble About to Burst? Demand, Financing Walls, and Open Weight Models first appeared on STARTUP HAKK.