Intel is set to raise $19.7 billion by selling new common stock in a bid to finance the building out of new production capacity, the development of next-generation leading-edge process technologies like 14A and others, and day-to-day operations. While the company does not assign money to a particular project, Intel needs to build capacity to land orders from large external clients, so capacity expansion will likely be a priority. According to Bloomberg, the share sale attracted $100 billion in demand.

Intel will sell 210,526,315 shares for $95 apiece through an underwritten public offering. In addition, participating banks have 30 days to acquire as many as 31,578,947 more shares at the same $95 price, minus applicable underwriting discounts. Should they exercise all their options, Intel could sell approximately 242.1 million shares altogether and increase the proceeds to roughly $23 billion. Without the additional shares, Intel expects net proceeds of approximately $19.7 billion after underwriting discounts, commissions, and estimated expenses. The transaction is scheduled to close on August 12, 2026.

Intel's market capitalization increased from roughly $90 billion last August to $491 billion at press time, so the time is right to sell some shares and raise some much-needed cash, as the company must compete against giants like TSMC and Samsung, which spend tens of billions of dollars every year on new fabs and advanced process technologies. Meanwhile, Intel's capitalization reached its all-time high of $673 billion on June 20, 2026.

Intel has not assigned the money it is going to raise to particular projects and says the capital can be used across the business, including for capital expenditures and working capital. The company is currently ramping up its Fab 52 in Arizona and is on track to start using adjacent Fab 62 when it needs to. In addition, the company still has to build its fab complex in Ohio, which is expected to cost over $100 billion when fully built, so it badly needs money.

In its risk disclosures, the company specifically mentioned Intel 14A — which is due to enter mass production in 2028 — and other advanced process technologies, manufacturing expansion required to support them, and the need to secure design wins and volume commitments from major external foundry customers. While Intel does caution that these long-term investments amounting to tens of billions may not generate adequate returns, it is impossible to land sizeable contracts from external customers without having production capacity readily available.

Interestingly, Intel also mentioned alternative financing arrangements, government grants, and the U.S. government's significant equity position in the company among relevant factors.

The FCC is drafting a proposal that would expand its list of equipment and services covered by the Secure Networks Act to include imports of new-model optical transceivers manufactured in China, according to a TrendForce report. Although the specifics are still unknown, such as what exactly constitutes a Chinese company and what is considered a "new model," this import block could have far-reaching implications. It's likely to impact hyperscaler AI companies, which are investing heavily in optical interconnect technologies, seen by many as the next battleground in the race for advancing AI performance, latency, and efficiency.

That's thought to be the core reason behind this potential blockade. With Chinese optical module manufacturers thought to make up around 56% of the global manufacturing capacity for this key technology in 2026, the U.S. administration wants to decouple U.S. reliance on this supply chain and strengthen its alternatives: Both domestic and international.

The long tail of this decision, however, could see the FCC refuse to authorize Chinese networking hardware entirely, following a ban on all foreign-manufactured routers in late 2025, and a ban on all advanced foreign-produced robotics in July this year.

Shifting bottlenecks

The story of the AI industry's rapid buildout in recent years has arguably been one of ever-changing bottlenecks and attempts to circumvent them. There are general GPU, CPU, and memory shortages that we've all had to contend with. But there have also been utility difficulties faced by hyperscaler companies, from power, to water, and local-infrasturcture.

There have also been more international key material bottlenecks, like shortages of glass cloth, of specialized PCB drill bits, and power inverters. Raw material shortages like helium, aluminum, and copper.

But those bottlenecks have also been felt within the servers within the data centers, too. And copper is a key feature there, because as modern data centers become ever more powerful, the new bottlenecks are the wires between them, more so than the raw speed of the processors or memory chips.

That's where optical interconnects come in. Designed to replace the simplicity, relative resistance, and physical limitations of copper wiring with lasers, optical interconnects can increase bandwidth, reduce latency, and reduce power consumption of existing networking hardware dramatically.

Although many of the world's largest optical interconnect developers, including TSMC, Intel, Samsung Foundry and Global Foundry, have different methods for the ways they plan to produce next-generation versions of these important networking interconnects, the underlying shift is much the same. Replace the bottleneck of copper wiring with optical communication lines and bring them as close to the processor as possible.

Within a few years, effectively integrating an optical networking interface into the chip itself, making latency all but disappear over distances up to several miles. That would allow for components to be located wherever they were best placed within a system, no longer needing to be near other components to reduce latency. That also allows for easy scaling up and out after deployment, as components from various servers could be co-opted as required. Multiple data center campuses within a few miles of one another could collaborate on particularly demanding workloads, too.

But at the moment China has a firm grasp on this emerging key technology, and the U.S. appears ready to respond with a heavy hand.

The ban and its impact

The FCC has banned the import of other products of specific types and from specific manufacturers before, but if it were to ban the import of optical transceivers, that would be the first time it had enacted such a measure. The term "new models" suggests such a ban would include the latest in interconnect technologies, including co-packaged optics (CPO) and near-packaged optics (NPO) integrated into network switches.

Although the U.S. may opt to deploy a transitional ban that only blocks the sale of new optical transceiver designs, leaving existing designs and contracts intact and saleable, it would still present a problematic disruption for American AI companies wanting to build cutting-edge data centers. China's optical module manufacturing, packaging, and testing are all key elements of the global optical interconnect supply chain, especially since a key raw material, indium phosphide, relies heavily on China's indium supply, with it controlling some 70% of the global market.

China restricted the export of indium in 2024, causing a reduction in global sales of the key material.

Even if the materials required can be sourced elsewhere, though, there's no suggestion U.S. manufacturing can scale up anywhere near quick-enough to respond, or indeed, at all. Key U.S. optical transceiver manufacturers, Coherant and Lumentum, are already running at capacity and are still around 30% overbought by their customers. A $2 billion investment from Nvidia will help, but the leaders there don't believe they'll be able to service the increasing demands of a hungry AI industry. Let alone if Chinese supplies are cut.

Nothing happens in a vacuum, except chip production

All of this assumes, too, that the Chinese authorities won't respond - and they certainly have a history of doing so. When America has restricted access to cutting-edge chips and chip design software, China has cut off raw material access. When the U.S. banned imports of Chinese hardware, China doubled down on investing in its domestic market and sources, as well as fighting to build friendly relations with other countries and markets.

That has already been happening in the optical interconnect space. In 2025, Chinese manufacturers only accounted for 16% of the global production capacity of electro-absorption modulated lasers (EMLs) and continuous-wave (CW) lasers. But in 2028, that capacity is expected to reach almost 28%. Will that additional production capacity go to help accelerate China's own domestic AI efforts, or will it end up being sold to major AI companies in Western markets?

It may be possible for Chinese suppliers and Western buyers to circumvent any new optical transceiver restrictions. Almost all the major bans and blockades from the Trump administration have included carve-outs for specific deals that can be struck — often seemingly at the whims of whatever official is involved in the deal-making. But they're there, and it may be that a phone call to the right person allows the right materials through, making any potential ban just the price of doing business.

But if not, it has the potential to put the brakes on the next phase of rapid AI infrastructure build-out that is making data centers more capable, campuses more efficient, and hinting at a future where even in consumer devices, the specific placement of components within them may no longer be constrained by the limitations of copper.

House Select Committee on China Chairman John Moolenaar has written a letter that demands the U.S. government enforce an existing export control measure designed to prevent Chinese companies from getting advanced chips produced by contract chipmakers like TSMC or Samsung Foundry.

When Joe Biden was about to leave the office in early 2025, he signed a law that required chipmakers to determine their end customers in a bid to prevent contract manufacturers from unknowingly fabricating chips using American technologies for companies that served as intermediaries for restricted organizations tied to the Communist Party or the People Liberation Army. The regulation was introduced after it was discovered that chips made by TSMC for China-based Sophgo were actually Huawei's Ascend 910B AI accelerators. However, the Trump administration announced in May 2025 that it would not enforce this semiconductor regulation known as the 'Foundry Due Diligence Rule,' which naturally created uncertainty about whether the foundry-focused requirements would be actively implemented at all.

"The announcement created ambiguity as to whether front-end fabricators like TSMC can export unpackaged advanced dies to non-approved designers located outside of China, without performing the due diligence specified in the Foundry Due Diligence Interim Final Rule (IFR)," the letter reads. "These exports enabled Huawei to obtain millions of controlled Ascend logic dies from TSMC, using its front company, Sophgo in 2023 and 2024."

Before the adoption of the 'Foundry Due Diligence Rule,' Chinese companies and/or their intermediaries could misrepresent chip specifications and end users when placing orders with TSMC or other foundries, which allowed restricted devices to be made despite U.S. export controls. Under the rule, foundries and OSAT providers exporting chips produced on 14/16nm-class process technologies or more advanced nodes must presume those devices qualify as controlled AI processors subject to a global licensing requirement, unless they meet an exemption. According to the letter, foundries have generally complied with these requirements since the rule took effect. As a result, Moolenaar essentially asks the government to enforce the existing controls rather than introduce stricter export controls.

Moolenaar believes that the Commerce Department's Bureau of Industry and Security could clarify its current stance on the Foundry Due Diligence Rule in two ways:

• By issuing guidance, which confirms that the worldwide Regional Stability (RS) licensing requirement still applies to exports from front-end foundries;
• By formally annulling the AI Diffusion IFR and amending §744.23 to explicitly restore that requirement for both foundries and OSAT providers.

Either approach would eliminate conflicting interpretations and reinforce enforcement of existing export controls, Moolenaar believes.

Nvidia late on Monday announced that it had signed memorandums of understanding with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR to establish independent financing platforms that could mobilize more than $500 billion in third-party capital to invest in AI infrastructure. Nvidia's goal is to ensure that its clients building AI data centers (which Nvidia calls AI factories) can get enough money from powerful financial companies. As a result, Nvidia will reinforce its position on the AI hardware market as the funds will exclusively finance Nvidia-based AI data centers.

The proposed funds (or platforms, as Nvidia calls them) are intended to provide dedicated pools of capital for customers — such as AI labs, cloud service providers, or enterprises — that deploy Nvidia-based infrastructure. Rather than financing projects itself, Nvidia intends to work with six investment firms to enable access to long-term funding at attractive rates. The company believes that AI infrastructure should not be viewed as conventional IT equipment, but as tools that make sustained economic returns, which is why it must be financed appropriately.

"We are in a pivotal moment of a historic AI investment cycle," said David Solomon, Chairman and CEO of Goldman Sachs. "Nvidia's full-stack platform is in high demand and uniquely positioned at the center of that global buildout. Our investment and distribution roles reflect our confidence in Nvidia's leadership, and we are excited for the new opportunity to create a market for credit backed by NVIDIA compute."

The financial companies believe that AI data centers can be treated as long-duration infrastructure assets rather than conventional IT equipment, in part because Nvidia compute can generate revenue over an extended period and retain value across different workloads and operators. As a result, they appear to believe that AI infrastructure can support long-term financing at attractive rates, although the companies do not explicitly claim that financing AI data centers carries lower credit risk than financing conventional IT deployments. Furthermore, it should be noted that Nvidia and financial companies will inevitably finance companies that would otherwise struggle to obtain capital to finance their AI data centers. This will ultimately help Nvidia sell more hardware and software while allowing its financial partners to capitalize on the rapid expansion of Nvidia's AI ecosystem.

Without any doubt, the arrangement will help to rapidly build AI infrastructure, which will increase adoption of AI technologies. However, this arrangement increases the risk of an AI infrastructure bubble as it potentially weakens one of the natural brakes on overbuilding: the availability and price of capital. Furthermore, Nvidia's help with arranging financing for its own customers introduces an element of circular financing into the AI boom, something that the industry faced during the dot-com bubble era in the late 1990s – early 2000s. However, this does not necessarily prove there is a bubble, as there is genuine, enormous demand for AI hardware and Nvidia sells plenty of such hardware.

Perhaps the biggest concern about the arrangement is that while Nvidia and its partners state that AI infrastructure can provide long-term value, AI accelerators, such as Nvidia's GPUs, have short and uncertain economic lives as the company and its industry peers introduce new and better-performing AI hardware every year, which devalues the previous generation.

"Nvidia has reached an important milestone: we began by building chips; today, we are helping create a new class of productive, investable infrastructure: AI factories," said Jensen Huang, founder and CEO of Nvidia. "In AI, compute is revenue. Nvidia compute is uniquely suited for this role. It is broadly adopted, flexible across models and workloads, fungible and transferable across customers and operators, and continuously improved through CUDA software — extending its useful life and improving its economics over time. It is supported by a deep global ecosystem of developers, customers, and offtakers. That is why we are bringing the world's leading long-term capital providers together to independently underwrite AI infrastructure. These financing platforms will help customers access scarce compute at scale and build the DSX AI factories that will power every industry and country in the age of AI."

Recently, the Kyoto Prefectural Police caught one of the initial-seeders of the well-known Japanese torrent website Nyaa. Masakazu Ono, 56, from Sagamihara in the Kamigawa Prefecture, was arrested after a multi-year investigation. While this news is straightforward by itself, the interesting bit is that CODA, the Japanese anti-piracy entity that performed the initial cyber-investigation, identified Ono allegedly without joining or monitoring BitTorrent swarm traffic, with the help of the Japan Hacker Association (JHA).

CODA's description of only having "[analyzed] how data moves through torrent sites, such as index sites and tracker sites" is vague and can be parsed in a number of ways. The distinction between indexer and tracker is relevant: while an indexer is a user-visible list of torrents (usually a website), the tracker is the background service that software like qBittorrent actually connects to, and it lists everyone who's currently active in that torrent, whether sharing, downloading, or both.

Our hypothesis is that CODA kept close tabs on the Nyaa trackers to see which IP address (in this case, Ono's) consistently appeared as the first person to offer up a 100% complete set of data for a show, a theory that would fit with how long the operation took. JHA's founder, Takayuki Sugiura, was the first to decrypt the Winny P2P protocol in 2004 to find original uploader sources, and the methods used weren't too different from the aforementioned theory. It's also possible that Ono wasn't using a VPN but was logged into Nyaa, or activity at another website's cookies or CDN logs was used to establish a link to his activity at Nyaa, and from thereon, to the tracker.

An initial-seeder is someone who is the first to offer up data for others to download — in this case, shows and movies. The show that reportedly got Ono nailed was Midnight Taxi, a 2026 Japanese drama produced by NHK and WOWOW. However, the police apparently had been tracking him for quite a while and claimed that he had uploaded close to a thousand NHK recordings over the investigation's span.

The process started in 2021 when CODA first launched its Cross-Border Enforcement Project, aiming to identify Nyaa's uploaders. During three years, with the help of JHA, CODA seemingly collected enough information to get the Kyoto police involved in 2024. Three "secondary uploaders" from "reach sites" were charged in 2025, likely meaning regular users participating in the torrents, who got there through link aggregator sites. Fast-forward to a couple of weeks ago, and Masakazu Ono was formally arrested with specific charges.

The most common techniques used by law enforcement agencies are creating a tracker honeypot or participating in the torrent swarm themselves in a bid to identify who's sharing files. However, Nyaa's list of trackers is small and comprises well-known, trusted servers, so together with CODA's wording, it's unlikely this was the method used. Furthermore, Japanese law generally frowns on non-targeted data monitoring, meaning that techniques like Deep Packet Inspection (DPI) and traffic volume monitoring likely weren't used, at least until such time as Ono was identified.

Reducing (or completely eliminating) the charge time for drones has been a continual area of research, with some advancements, such as quantum batteries showcased earlier this year, showing a lot of promise. Researchers at the Civil Aviation University of China have made a major breakthrough in laser-based OTA (Over The Air) charging that can convert energy at 38.94% efficiency.

According to a study published in Matter & Light, scientists from the Civil Aviation University of China are testing a lightweight receiver that can harvest energy from a laser beam sent from the ground. The system uses a perovskite laser cell-thermoelectric (PLC-TE) tandem device, which combines two methods of generating electricity. The perovskite layer converts laser energy into electricity, while the thermoelectric layer recovers some of the resulting heat and turns it into additional power. Its output increases as the temperature difference between the laser-facing side and the cooler side grows.

The major challenge with this setup is that intense laser beams heat up the receivers to a blistering 80 to 90 degrees Celsius, degrading efficiency and causing damage to the aircraft itself. “That was much higher than we expected and made us realize that heat buildup was a far more serious problem than we had imagined,” explained Jianhua Han, the study's senior author.

To overcome this challenge, Han's team added special nanocrystals to the receiver. As the nanocrystals are poor heat conductors, it allows them to act as a thermal barrier and protect the perovskite and thermoelectric layers. The team also added air channels to the drone's wings. As the propellers spin, airflow passes through these channels and carries excess heat away, helping to keep the system cool. The combination of nanomaterials and aerodynamic cooling resulted in a respectable 38.49% energy conversion efficiency during testing, which is one of the highest reported for this class of technology.

While the system is yet to undergo outdoor flight tests, this breakthrough alone is a significant step toward practical continuous flight for drones. As Han noted, going from 1 to 100 will require solving complex tracking and safety challenges, but this research establishes an important starting point for the future of drones and UAVs.

An Australian AI user has kicked up a storm at his local gym after trying to use OpenClaw to book himself into a gym class. The user asked the agent to see if there was a way to try to bump him up the waitlist for a class later that week, at which point the AI hacked into the system and cancelled the place of one of the participants to try and make room for him, ABC Australia reports.

The report describes Andrew as an employee at an Australian AI B2B firm who started experimenting with popular AI agent OpenClaw earlier this year. According to the report, Andrew thought the task of booking a gym class was "a chore," so decided to ask OpenClaw to do it for him instead. The AI exceeded its brief in two ways. Firstly, it offered him the option to book into classes in advance far beyond the supposed limits of his local gym's booking system.

Not to be outdone, the agent then hacked the system and kicked another participant out of the class after Andrew asked if there was a way to get moved up the waitlist for a class later that week. "The API has zero authorisations checks on cancelling other people's reservations … I tested this with the person in waitlist position #1 — and it actually went through. So you've moved from #4 to #3 already," the bot told him.

Realising what had happened, Andrew asked OpenClaw to add the jettisoned gym-goer back to the class, at which point the bot replied that wasn't possible. "The person I removed is gone from the waitlist and I have no way to restore them," OpenClaw admitted before noting they'd have to rejoin the waitlist themselves.

OpenClaw closed out by apologising, stating "Sorry about that — I should have been more careful," before promising not to touch anyone else's spots. No doubt mortified, when Andrew realised OpenClaw couldn't put right the problem by itself, he did the next best thing and asked OpenClaw to write an email to the gym software provider to explain itself and the vulnerability it had found.

The number of local bans on new data center developments jumped to over 500 in July, according to recent data. Local politicians in the United States are refusing to approve new data center construction in greater numbers, posing a threat to the expansion required by AI firms to provide the capacity needed to power the new technology. The analysis, conducted by The Information from legal documents and local news reports, shows that this isn’t a partisan issue.

Republican- and Democrat-run states, towns and counties have all begun to block new data center projects, with Texas and New York both prominent entries from both sides of the political divide, and with the majority of the bans appearing in the Midwest and South, according to the report.

Politicians aren’t doing this out of spite, but in response to the growing concern from residents over the impact on communities. Among them are the impact of noise on local residents, as well as the disproportionate impact on public services, including on water and electricity supplies. It follows Texas Governor Greg Abbott’s moratorium on data center approvals in recent weeks until an extensive audit has been conducted of companies looking to hook their new centers up to the Texan power grid.

The Information’s analysis shows that the bans are growing fast, jumping from 300 in late June to over 500 in July. The ban in Texas is particularly noteworthy due to the state becoming a prominent home for AI data centers. Despite Governor Abbott’s pause, Amazon is still hoping to build a new gas power plant in Pecos County, Texas, to power a data center at the same location. That location is already believed to likely become the single biggest source of environmental pollution in the U.S. once it’s operational.

The rapid growth of new data centers in Texas and nearby states has likely fueled those concerns from residents. The Information reports that most of the bans are in the Midwest and South, with more than 400 restrictions now in place, with nearly three-quarters of the moratoriums approved in July occurring in southern states like Georgia and Florida. Florida, in particular, is quickly stopping new developments, with 14 out of 23 current active restrictions in that state being approved in July, and 15 more under consideration.

Small and big, rural and urban states alike are all seeing political capital used on preventing AI data center expansion. Kathy Hochul, Democrat Governor of New York, recently signed an executive order that paused approval for all data centers consuming 50 megawatts of energy or more. Massachusetts and Nebraska halted tax breaks for developers, while in Colorado, 19 bans are now in place around Denver, and more elsewhere in the state, due to concerns specifically around the impact on the environment and on water usage. In Nashville, Tennessee, a 23-acre data center development next to a zoo, expected to cost up to $700 million, is also facing local and political resistance.

It’s no surprise to see politicians begin to wake up to the public malaise against AI development. Over 70% of Americans are reportedly against more AI data centers, and direct resistance against them is growing, with at least 37 people arrested so far this year for protesting against new developments.

Apple used to be among a few companies willing to buy memory and other components worth billions of dollars under long-term supply contracts at fixed prices. But the artificial intelligence era represents a new reality with new purchasing champions, marking a tectonic shift in the high-tech world. Alphabet, Microsoft, Meta, and Amazon have purchase commitments totaling about $2 trillion, and a significant portion of these commitments are for memory, according to estimates by analyst Claus Aasholm. While the commitments are approximate, span many years, and should be generally taken with a grain of salt, they still reflect the direction the industry is moving.

Combined purchasing commitments from the four major hyperscalers shown in the chart —Amazon, Alphabet, Meta, and Microsoft — reached nearly $2 trillion by Q2 2026, with Alphabet and Microsoft accounting for the overwhelming majority of the total.

The rapid expansion suggests several major findings. Firstly, the AI infrastructure race is accelerating, not stabilizing. Secondly, AI infrastructure investments are driven by a handful of hyperscale cloud service providers (CSPs) whose long-term procurement commitments now vastly exceed those of traditional consumer electronics companies such as Apple.

Thirdly, memory has become a strategic asset — perhaps a competition weapon — rather than a commodity. Fourthly, suppliers of memory — both 3D NAND and DRAM — are gaining pricing power. Finally, demand for memory will likely drive major capacity expansion at Micron, Samsung, and SK hynix, even though so far these companies have been exceptionally disciplined about their capacity investments.

Almost $2 trillion commitments

Google shows by far the most aggressive increase in purchasing commitments, rising from roughly $140 – $150 billion in Q3 2025 to around $811 billion by Q2 2026 (though these are total purchase commitments by Alphabet, not specifically memory purchase commitments), while Microsoft follows a similar trajectory and reaches approximately $678 billion in total obligations, which includes, but is not limited to memory.

Meta is also ramping commitments substantially to around $349.3 billion (again, these are total commitments), whereas Amazon increased its commitments more gradually to roughly $130 billion. By contrast, Apple — which makes the world's most popular smartphone, and which was the largest consumer of memory just a couple of years ago — remains almost flat throughout the period at approximately $57 billion (of which $56.2 billion is payable within 12 months). Apple's commitments fall well short of Nvidia's commitments of $119 billion.

Again, we are talking about total purchase commitments, which include foundry capacity, 3D NAND, and DRAM memory, but are not limited to them. Alphabet, Amazon, Meta, and Microsoft all build custom silicon and custom servers, so a significant portion of these commitments is to various EMS providers.

While $1.968 trillion of purchase commitments for memory and storage alone would be an absurdly large amount of money, a huge portion of these commitments consists of contract manufacturing obligations as well as memory chips. This suggests that the foundry, 3D NAND, and DRAM markets are entering a new phase in which hyperscalers are willing to make vastly larger forward purchasing commitments than traditional consumer-electronics companies, giving suppliers a strong incentive to prioritize customers prepared to secure future capacity on that scale.

Strategic assets

While Claus Aasholm's chart is explicitly dedicated to memory, it does describe total purchase commitments of tech giants, so the chart can reasonably be read as evidence that memory and capacity at TSMC, Samsung Foundry, and GlobalFoundries are becoming a strategic asset rather than merely another component to procure at the best available price.

AI infrastructure requires enormous quantities of AI accelerators, DRAM (including HBM), and 3D NAND. Meanwhile, the supply of high-end memory (HBM) is constrained by fab capacity at major DRAM makers, whereas the supply of AI accelerators is constrained by both wafer capacity and foundries and packaging capacity at foundries and their OSAT partners. As a result, hyperscaler CSPs have an incentive to lock in supply years ahead, even if doing so requires exceptionally large purchasing commitments.

That also changes the relationship between semiconductor suppliers and their customers. In theory, a company willing to guarantee hundreds of billions of dollars of future purchases can effectively help underwrite expansions of foundry, memory, and advanced packaging capacity and, in return, secure priority access to scarce products and future process technologies. In reality, TSMC can well afford capacity expansion using the money it gets from hyperscalers and give priority to its largest customers. In this environment, access to DDR5, HBM, and 3D NAND memory becomes part of the competitive advantage rather than merely a procurement exercise.

This is also what makes Apple's position in the graph interesting: its purchasing commitments barely move while those of Alphabet, Amazon, Meta, and Microsoft surge. If the trend continues, Apple may remain one of the world's largest semiconductor buyers in absolute terms, but the question is whether it will be among the key customers that foundries, memory makers, and OSATs plan their future capacity expansions.

An inflection point

Perhaps the most interesting takeaway of the findings revealed by long-term purchase commitments is that the industry's center of gravity appears to have shifted.

During the smartphone era, foundries (well, TSMC has won) and memory suppliers often competed aggressively for Apple's business because of its enormous purchasing power. Today, hyperscalers building AI infrastructure are making purchasing commitments that dwarf those of traditional CE companies like Apple, which may well represent a strategic inflection point akin to the one Andy Grove described in his 'Only the Paranoid Survive' book.

Will this tectonic shift result in prioritization of customers capable of enabling future capacity expansions through massive long-term purchase agreements, or will foundries and memory makers remain more or less disciplined with their capacity expansions so as not to lose a lot when demand declines? This is a question that has yet to be asked.

In any case, the AI megatrend has transformed semiconductors — from foundries to advanced packaging and from DDR5 to HBM4 — into strategic assets that can no longer be treated as ordinary components procured on demand. And this is something that will continue in the long run.

Steam hardware customers in Europe should “expect fake messages - email, SMS or phone,” says Valve in an email bulletin circulating today. The problem is that its European hardware distributor, CEVA Logistics, has fallen prey to a cyberattack where personal information and hardware purchase details were compromised. The most sensitive data, such as payment information, passwords, and/or Steam Guard codes, was not pilfered during the attack.

Valve EU shipping partner hit by cyberattack.

— @joel.zone (@joel.zone.bsky.social) 2026-08-10T11:42:07.084Z

According to the official communication from Valve, CEVA was hit by a cyberattack between July 29 and August 1, 2026. The PC gaming company says that it learned that hackers had successfully swiped a range of information about Steam hardware customers on August 7.

As this data was in the hands of CEVA, it was strictly limited to delivery-related information. That means the cyberattackers now likely have a database of hardware customer information that will include fields like names, addresses, phone numbers, email addresses, and the type and price of the product(s) ordered. CEVA holds this type of information for 90 days to cover fulfillment of the hardware order. It is worth repeating that the attacked distributor won’t have spilled data such as payment information, passwords, and/or Steam Guard codes.

The likely fallout customers should be wary of

Valve candidly warns those who have purchased Steam hardware like a Steam Deck or Steam Machine in Europe over recent months to “expect fake messages - email, SMS or phone - that mention your hardware order and appear to come from Steam, Valve or a delivery company.” Likely angles used by swindlers will be to ask for extra customs or delivery fees. They may also ask you to sign into some fake web portal or other. However, you should ignore such requests, of course, and changing your Steam password or anything to do with your account settings is not necessary, insists Valve.

If you are in any doubt about a communication that comes from Steam / Valve or a distributor, please check the URLs carefully. For example, Steam Support only handles account issues at help.steampowered.com, and only store.steampowered.com, www.steampowered.com, steamcommunity.com, or help.steampowered.com are ever used for Steam account logins. You are advised to type these into a browser yourself and not click any email-embedded links. We assume poking around the official Steam App on your PC will also be fine, and many of you will have that auto-running when your PC starts up. Lastly, you are reminded that “Steam Support will never ask you for your password or a Steam Guard code. Neither will a courier.”

CEVA isolated the affected systems and notified data protection authorities after discovering the hack. It has brought in outside investigators.

The public pushback against data center construction projects is only growing stronger, even as some protesting local residents have been arrested in the process. That hasn't perturbed those still standing against the AI infrastructure projects, though, and they continue showing up at town meetings, holding protest rallies, and making themselves heard — even with the 24/7 drone of a data center hum in the background.

The latest effort saw Nashville residents and local politicians move to block the construction of a $700 million data center set to be built right next to the Nashville Zoo at Grassmere. The city council approved a blocking measure last week, and now Mayor Freddie O'Connell will wield it to make sure the project doesn't go ahead.

This follows on from another recent story where a Kentucky farming family was offered 10 times their land value to sell up to data center developers, and they refused. Virginia's State Corporation Commission, which regulates public utilities, just ruled that data centers must pay for all the required transmission infrastructure that they have exclusive use of, in an effort to reduce energy bills for residents.

Nobody wants to live near a data center

Data center projects have been firmly disliked by large bodies of Americans since the beginning of this major AI buildout. Early in 2025, it became clear that even as the major AI companies were promising hundreds of billions for "AI factories," governments around the world rushed to build capacity. There was only one problem: residents near the proposed build sites would not be taking it lying down.

In April 2025, Elon Musk's Colossus data center was an early target for protest after it was found to be using permitless mobile gas turbines for power. It's been hit with enough pushback since then that SpaceX has now agreed to remove the 69 turbines it ultimately installed, although it is going to take over a year to complete the move of these allegedly mobile turbines.

By the Fall of last year, evidence had begun to mount about the negative impact of these installations. Water shortages and potential contamination, electricity price increases for residents, and the constant, unending noise that may be causing health problems with sub-audible frequencies.

There's also the slowly growing number of "against" voices in polling for new data centers. What started as a light advantage to the "supports" respondents in September became a near-majority a few months later. Now more than 70% of Americans don't want data centers near their homes, and there's little chance of that changing as the stories of issues during and after construction of these facilities continue to mount.

How the people are fighting back

Protest has been at the forefront of the anti-data center campaigners' playbook since the start, and that's not stopping any time soon. On one weekend in July, anti-data center protestors organized in 42 states to stage 142 protests across the country, all calling for blocks, halts, and moratoriums on the data center rollout.

This is after months of campaigning, which led to close to 70 jurisdictions around the country blocking new data center builds. Some of them, permanently.

Where politicians are more on the side of the developers, the people are going to meet them in person. They're going to town halls and seeking meetings and making their voices heard.

Sometimes politicians don't like that, and people are removed - even when protesting peacefully - and in several cases they've actually been arrested. Close to 40 people have been arrested in 2026 for pushing back against data center rollouts. In one case, because they went beyond their allotted speaking time, and in another, because they clapped following an impassioned statement by a fellow citizen.

In other cases, though, politicians are listening. In Nashville, the local council and the mayor are all on board to protect the local Zoo and residents' bills and ears. This is particularly important because it shows that local legislators can use their political power to block data centers — potentially preventing the White House from using its own to drive projects through.

The forever war

The data center builders aren't taking all of this helplessly. They have the ears of government and massive platforms to evangelize AI and data centers; They've been doing it everywhere from the White House to social media. They've tried building in more rural areas where there are fewer people to protest, paying for their own behind-the-grid power, and offering to spend millions in local investment to counter negative effects.

But it's not working. People don't like data centers. They don't want to live near them. And on top of that, many don't even like the technology the hyperscalers want to build this all for, anyway.

It's unlikely that those bankrolling (and hoping to make money on) the data center AI build-out will just give up and go home, but the residents fighting back against their gargantuan construction projects aren't going to do so either.

This is a long war, and one that neither side is likely to back down from. But in a rare ray of hope in 2026's landscape of high PC and electronics prices due to AI data centers soaking up computing hardware, it's one that the people seem to be winning, for now.

HoverAir, makers of self-flying selfie drones, have just unveiled the HoverAir Versa, combining their existing aerial expertise with the convenience of a pocket camera. At its core, the Versa is a high-end stabilized pocket gimbal camera, like the DJI Osmo, but it can magnetically snap onto a drone body and transform into a cinematic selfie drone. It's designed to tackle two different product categories simultaneously, but with the same goal of making it as easy as possible to film yourself (or others).

The Versa is actually HoverAir's first-ever pocket camera. It features a three-axis gimbal, which means it can adjust for tilt, yawn, and pitch to make sure your footage turns out stable. Compared to other options on the market, no pocket camera offers more than two-axis stabilization. It has a rotating screen and the accompanying controls for the camera, along with RGB accent lighting that also works as a status indicator. Overall, it looks like a pretty run-of-the-mill, well-made pocket camera.

But the magic happens when you bring out the wings. A small, foldable drone body can magnetically attach to the camera unit via pogo pins located at the back of the device. Once connected, you just unfold the wings, gently tap the top, and it takes off right from your palm. There is no external controller needed because the device uses AI to track and follow you automatically. It has 10+ intelligent modes that can film you from any direction with AI-assisted framing, along with a neat 360° feature as well.

In this mode, called 3D Worlds, the drone flies around to snap pictures of the scene from different angles and composites them into a single 360° scene that can be explored and shared later. Technically, this would be the third audience the Versa is trying to cater to — the Insta360 crowd, though that's real-time 360° video recording that you can also generously manipulate in post, so not exactly comparable. Still, it's a neat trick, and we'll have to see how well HoverAir implements it.

We don't have a price or release date; just a bunch of teasers so far that say it's coming soon. As such, specs are also up in the air, but the company is touting class-leading performance with the best dynamic range in the category. From the promotional material, we can infer that it has a large lens, and the drone body looks very compact. If HoverAir's previous selfie drones are anything to go by, the Versa could have support for up to 8K video in 10-bit Log or HLG, along with rapid flying speeds of up to 45 km/h.

The test footage on HoverAir's website and the video we attached above look really impressive. The ingenuity of the concept warrants both excitement and skepticism, but it's undeniable that the Versa kind of feels magical. Nothing like it exists in the market today, and if the company can already ship excellent selfie drones like the X1 series, and pocket cameras aren't exactly rocket science, it stands to reason that harmonizing the two together is the only challenge for the Versa.

K3 Scout surveillance drones used by the UK’s Royal Navy contain components that have been secretly transmitting data to China, according to reports. The Telegraph’s revelations suggest that the sea drones include a camera component that would send the data to China. However, the Ministry of Defence (MoD) says that it has removed all internet connectivity from the cameras after discovering the breach. The transmitted data was also limited in scope, insists the MoD, which characterized the transmissions as “heartbeat communications” to simply verify the device is online and working normally.

Western governments have been seeking to minimize the ingress of Chinese technology into sensitive and critical infrastructure for some years now. The pushback started with investigations into rising telecom brands like Huawei and ZTE, which became very successful with their cheap and efficient equipment in the West.

More recently, we have seen pushback affecting Chinese drone companies, as the utility and value of drones have become incredibly clear in the Russia – Ukraine war. Western allies obviously don’t want drones with potential Chinese backdoors, killswitches, or so on inside them. However, even manufacturers in China’s fiercest foe, Taiwan, recently admitted that its drone-making operations were still dependent on Chinese components.

Returning to the topic of drones, specifically, the sea drones in the Telegraph story were reportedly acquired by a major British defense contractor, Kraken Technology Group. The same company “sourced the cameras from a third party who had given assurances about their security,” says the newspaper. But an unnamed Telegraph source told the paper that this security lapse was a “major failure,” meaning that confidence in the platform has been lost. It even goes on to say that there are concerns that these drones could have snooped on highly sensitive special military forces meetings, even when switched off.

An MoD spokesperson told The Telegraph that the issues with the Chinese cameras were identified during a routine cyber vulnerability assessment. That seems like evidence of good practice, showing proactive testing and assurance procedures are in place. Moreover, the investigation “found no evidence of MoD data or systems being accessed, compromised or transmitted externally.” That didn’t stop The Telegraph or shadow (opposition party) ministers from being “furious” about the revelations, though.

The K3 Scout drone can carry a 600kg payload at sea for 30 days continuously. The Royal Marines have been using a £12 million fleet of these drones since March and offered to deploy them to secure freedom of navigation in the Strait of Hormuz.

U.S. Special Operations Command also has K3 Scout drones, and these drones have been used by allies in NATO trials in the Baltic Sea.

An elderly farmer in Chuzhou, China, has been following AI advice since he started using it a year ago, but after months of interactions, it recently gave some terrible advice that led to the death of his entire crop. According to Taiwanese outfit CTWANT [machine translated], 67-year-old Wu initially didn’t trust the AI app but eventually changed his mind after it proved to be useful. Unfortunately, this mistake turned out to be disastrous, as it directly caused the death of 150 mu, or 24.7 acres, of sesame seedlings.

Wu asked the AI for advice on weed and pest control, after which it came up with “Hundred Acres of Sesame Grass Control + Pest Control,” and recommended that they use “high-efficiency flupyrimethalin” and “flusulfasulfaether” to kill weeds and mixed it with “thiamethoxazine” and “methyl salt.” The farmer followed the AI’s advice to the letter without confirming the information via agricultural technicians (or even just counter-checking its answers online).

The very next day

The result of the AI’s mistake was quickly apparent, as, just the following day, the weeds and the sesame seedlings died en masse. “If you spray it, the next day the seedlings won’t survive,” Wu said during a video interview. “Both the grass and the seedlings will die, and the seedlings will die even faster.”

When he asked the AI why this happened, it suggested that the “flusulfasulfaether” might be behind the issue. Agricultural experts said that this chemical was primarily used against broadleaf weeds in soybean fields. The USDA [PDF] categorizes sesame as a broadleaf plant, meaning it will also be easily affected by this herbicide. Furthermore, this chemical is designed for targeted spraying on affected areas only and shouldn’t be applied to the entire field. The farmer claimed that the AI did not warn him of the risks of following its answers, but its chat page had a reminder that said, “AI generation may be incorrect, please verify.”

Although AI is a powerful tool, it also has a history of giving bad advice and making bad choices. Just a couple of days ago, one developer had his entire profile directory deleted during a routine backup procedure after it encountered a typo. Even Meta’s AI Alignment director fell victim to this, after she had to manually terminate her AI agent to stop it from accidentally clearing out her entire inbox. That’s why it’s imperative that users are aware of AI’s limitations every time they use it.

This is one of the first instances in which we reported an AI mistake resulting in a physical consequence in the real world. We also cannot entirely blame the person for trusting the tool — after all, he was skeptical of it at first, meaning it produced results that made the farmer trust it over time. But even if an AI’s output has been reliable recently, that doesn't mean that it is absolutely trustworthy. After all, current LLMs are nothing but prediction engines — you cannot guarantee a correct answer every time you use them, especially if they have been trained on a public knowledge base whose dataset hasn’t been completely filtered for accuracy and factuality.

The European Commission and the SpaceRISE consortium signed an implementation agreement on Friday, August 7, to roll out Europe’s IRIS² satellite constellation, moving the project from the planning and negotiation stage into full-scale deployment. According to an official press release, the agreement adds 66 satellites to the planned constellation, bringing the total to 348 — 330 satellites in low Earth orbit and 18 in medium Earth orbit — with first launches targeted for 2029.

The expansion is intended to substantially increase the network’s government and defense capacity. According to the Commission, the revised architecture will provide 60% more secure governmental communications capacity within the EU and 54% more globally. The existing satellites in the baseline constellation will also receive additional functionality and stronger security features, while the 66 new spacecraft will operate in higher low Earth orbit.

IRIS² — short for Infrastructure for Resilience, Interconnectivity, and Security by Satellite — is the European Union's flagship secure-connectivity constellation, intended to give European governments an autonomous alternative to foreign networks such as SpaceX's Starlink and Amazon's Leo. The Commission signed a 12-year concession contract for the program in December 2024 with the SpaceRISE consortium — a grouping of European satellite operators SES, Eutelsat, and Hispasat — which is building and operating the system under a public-private partnership alongside manufacturers including Airbus Defense and Space, Thales Alenia Space, OHB, and Aerospacelab.

Once operational, the multi-orbit network will deliver sovereign, encrypted connectivity to authorized governmental users across EU member states and participating non-EU countries, currently Norway and Iceland, for use in emergency response, critical infrastructure protection, and security missions, with capacity also earmarked for citizens in areas with poor coverage.

When the concession contract was signed in December 2024, IRIS² carried a price tag of €10.55 billion ($12.19 billion), funded by the European Union, the European Space Agency, and the SpaceRISE consortium. The accelerated timeline and larger constellation have pushed that figure higher, with the consortium's partners now putting total planned investment above €15.6 billion ($18.03 billion), drawn from the EU budget for 2028–2034, ESA, the consortium partners, and member states. Several countries have already committed, with Poland and Hungary pledging €656 million ($758 million) and €500 million ($578 million), respectively, while Spain has announced between €1.6 and €2 billion ($1.85–2.31 billion).

IRIS² is frequently framed as Europe's answer to Starlink, as they are both satellite constellations aimed at providing connectivity to dead zones. However, the primary purpose of the 348-satellite-strong constellation is more niche. The EU wants an independently controlled communications system — mainly for government and emergency use, but also for civilian applications — that can remain available when terrestrial networks fail, or when access to foreign-operated satellite systems becomes politically or strategically uncertain. On the other hand, Starlink’s growing 10,000-satellite constellation is a mainly commercial, privately owned company selling broadband to consumers worldwide. It is now also running satellite-to-phone internet services through Starlink Mobile.

Most megafactories usually take years to go from conceptualization to construction, but a recent drone flyover of the Terafab showed that progress has already started on the ground less than five months after Musk unveiled it. However, the video doesn’t do justice to the true scale of its footprint, so X user Nic Cruz Patane created a visualization to help us understand how large the chip manufacturing facility is.

The site will reportedly have a floor space of at least 100 million square feet, making it larger than Giga Texas (10 million sq. ft), the Pentagon (6.6 million sq. ft), Apple Park (2.82 million sq. ft), and the Mall of America (5.6 million sq. ft), combined. This also makes it significantly larger than New Century Global Center in Chengdu, China, with has an interior space of “just” 18.9 million sq. ft.

While this might seem like an absurd amount of space for chip manufacturing, it appears that the Terafab will need it because it’s going to be more than just a chip fab making AI processors — instead, it will be an all-in-one facility that will produce logic and memory chips, as well as have lithography, packaging, and testing under one roof.

Elon Musk started talking about building his own chip manufacturing facility in late 2025 and officially announced the project in March of this year. The reasoning behind this project is that both SpaceX and Tesla will require at least 1TW of compute, which is more than ten times that current global chip supply.

Nvidia CEO Jensen Huang warned that a project like this will be an “extremely hard” challenge, but it appears that Musk is willing to put his massive resources on the line for this. Intel CEO Lip-Bu Tan even said that he can “think of no better partner than Elon Musk” to explore “unconventional” ways of chip manufacturing.

Musk is no stranger to both leading and funding projects that otherwise seemed impossible. Although he did not found Tesla, his investment and leadership in the company led it to become an industry trailblazer in EVs. He fundamentally changed commercial space travel with SpaceX; something that used to be the purview of NASA and other national governments, and he also broke a record when he set up 100,000 Nvidia H200 GPUs in just 19 days back in 2024 — a process which Huang said usually takes four years.

It seems that Terafab is Musk’s biggest project to date, which is estimated to require up to $119 billion in investments. But even though it seems that this project already has a secure customer base in SpaceX and Tesla, the former acknowledges that there’s a risk that this ambitious megafactory may not be successful.

The Federal Communications Commission (FCC) is moving to ban drones with LiDAR obstacle-avoidance features, a move that would remove a number of popular consumer drones from US shelves. In a public notice published in the Federal Register on August 3 (DA 26-758), the agency proposed retroactively banning the import and sale of any foreign-made drone on its Covered List that “contains or integrates” a LiDAR sensor, categorizing the technology under a broad “military-grade” umbrella.

The proposal builds on a ban the FCC set in motion late last year. On December 22, 2025, the agency added all foreign-made drones and their critical components to its Covered List — a roster of communications equipment that a U.S. national-security authority has determined poses an unacceptable risk. The listing blocked any new foreign drone model from receiving the FCC authorization it needed to be sold in the U.S., effectively banning sales of new models.

However, models that had already cleared authorization before the listing could continue to be imported and sold. The new DA 26-758 proposal seeks to retroactively strip sales clearance from previously approved drone models that ”escaped” the initial ban, starting with those it deems military-grade. The proposal is open for public comment until September 2, 2026, and if adopted as written, the ban on imports and sales would take effect roughly 180 days after publication.

LiDAR (Light Detection and Ranging) is a remote sensing technology that uses laser pulses to measure exact distances and build precise 3D maps of objects and environments. Drones, delivery bots, and self-driving cars all rely on it for obstacle avoidance. A ban would mean that popular consumer models such as DJI's Air 3S and Mini 5 Pro — both of which use LiDAR for obstacle detection — would no longer be available for sale in the US. The proposal would not affect existing owners, who could keep flying the drones they already have, but as DJI has warned, those owners could find official spare parts increasingly hard, or eventually impossible, to source over time.

Under the FCC's broad ”military-grade” classification, the ban would also reach drones that carry thermal imaging sensors, can dispense agricultural sprays (”economic poison” in the notice's language), weigh 55 pounds or more, include docking stations, integrate a defense article, or operate in coordinated swarms — a category the FCC defines widely enough to include synchronized drone light shows. U.S.-made drones and other drones not on the Covered List are not touched by the ban, regardless of whether they carry any of these features.

The FCC frames the move purely as a national security issue, leaning on an executive-branch determination that foreign-made drones pose an “unacceptable risk.” It tentatively concludes that the economic hit would be minor, as military-grade models are a small minority in a market dominated by recreational fliers. Conversely, the notice also details the billions that domestic drone makers like Skydio, which has committed $3.5 billion to US manufacturing, have raised since the original ban.

DJI, one of the manufacturers affected by the ban, has argued that the “military-grade” label is a category error, sweeping in civilian tools that first responders, farmers, and hobbyists rely on. Its core objection lands on the LiDAR clause, specifically arguing that the technology is a safety feature, not a weapon. DJI had earlier sued the FCC to overturn the December ban, claiming the agency exceeded its statutory authority.

A new Amazon data center is set to become the single largest source of environmental pollution in the U.S., according to a New York Times report on August 8. The tech giant — which already operates the most data centers worldwide — is reportedly building a natural gas power plant in Pecos County, Texas, to power its new data center at the same site. According to project permits, the plant would generate up to 7.65 gigawatts of power using 35 natural gas turbines. The new plant — which burns natural gas to generate electricity — has been authorized to release 33 million tons of CO₂ annually, more planet-warming gases than any other power plant in the country, says the NY Times report.

Conversely, Amazon had pledged to achieve net-zero carbon emissions across all its global business operations by 2040. The company made the promise in 2019 when it initiated and co-founded The Climate Pledge — a voluntary coalition of over 700 companies and partners — aimed at addressing its own corporate footprint and rallying global supply chains. However, the company has seen its emissions rise each year for the past several years, something analysts attribute to a spike in data centers to support the AI boom. The company is currently building several additional AI data centers globally, including the one that the new gas plant would power in Texas.

Amazon has acknowledged the impact of its intense AI ambitions on its earlier climate pledges but insisted the company remained committed to them, despite obvious struggles. “The world looks different now than when we co-founded the climate pledge,” said Margaret Callahan, an Amazon spokeswoman. Still, “our commitment hasn’t changed,” she said, echoing similar statements by Microsoft, which is struggling to fulfill its 2030 sustainability promise amid carbon-heavy AI expansion, but insists it remains committed to it.

As detailed in our data center power roadmap, an increasing number of hyperscalers and data center operators are switching from grid connections to dedicated on-site power generation. The “behind-the-meter” moves are being triggered mainly by the long timelines required to set up new power infrastructure to connect new power-hungry data centers to the grid, but also by growing anti-data center sentiments over irreversible hikes in the electricity bills of nearby communities.

While many companies are also pursuing nuclear and renewable energy options — examples being Microsoft’s Three Mile Island nuclear deal and Meta’s agreement to secure up to 1GW of orbital solar energy capacity — gas appears to be the leading preference due to its scalability and speed of availability. The Trump administration has thrown its weight behind such projects for data centers, promoting oil, natural gas, and coal over renewable energy sources.