An anonymous reader shares a report: Last week South Korea's SK Hynix announced it would partner with Purdue University on a $3.9 billion semiconductor complex here, the largest single corporate investment in state history. Now comes the hard part. SK Hynix must not only build the fabrication plant, or fab, which will package high-bandwidth memory chips used in artificial intelligence, and a connected research-and-development center. It also has to staff them. "We need several hundred engineers to operate our advanced-packaging manufacturing fab -- in physics, chemistry, material science, electronics engineering," Kwak Noh-Jung, chief executive of SK Hynix, said in an interview following last week's announcement.

Staffing a fab is harder in the U.S. than in South Korea, where SK Hynix has contracts with local universities and its own in-house university. Nonetheless, Kwak said, "the final goal is very clear. We need to have very good engineers for our success in U.S." The U.S. is trying to do something unprecedented: reverse a shrinking share in a key manufacturing sector. Between 1990 and 2020, the U.S. share of world chip making shrank to 12% from 37%, while the combined share of Taiwan, South Korea and China grew to 58%. The federal CHIPS program has showered billions of dollars on Intel for fabs in several states, Taiwan Semiconductor Manufacturing Co.in Arizona and GlobalFoundries in New York and Vermont. SK Hynix hopes for support as well.

Subsidies alone won't guarantee a sustainable industry. Fabs need customers, a supply chain and, above all, a skilled, specialized workforce. From 2000 to 2017, U.S. employment in semiconductor manufacturing shrank to 181,000 from 287,000. It has since recovered to about 200,000. Why did the U.S. share of semiconductor production shrink? As in other industries, the U.S. became an expensive place to manufacture. Susan Houseman of the Upjohn Institute, who has studied outsourcing, said this wasn't "primarily a story about offshoring." U.S. companies still lead in chip design: Nvidia in artificial intelligence, Qualcomm in communications and Apple in smartphones. Over time they mostly contracted out fabrication of their chips to foundries such as TSMC who benefited from generous domestic subsidies. The theory behind CHIPS is that, by matching Asia's subsidies, the U.S. can again be competitive in chip making. Nonetheless, there is a chicken-egg problem. Fabs need a ready supply of skilled workers. But without fabs, America's best and brightest have little incentive to pursue careers in the sector.

Intel on Tuesday unveiled its new "Gaudi 3" AI chip that the company claims is over twice as power-efficient and can run AI models one-and-a-half times faster than Nvidia's H100 GPU. "It also comes in different configurations like a bundle of eight Gaudi 3 chips on one motherboard or a card that can slot into existing systems," adds CNBC. From the report: Intel tested the chip on models like Meta's open-source Llama and the Abu Dhabi-backed Falcon. It said Gaudi 3 can help train or deploy models, including Stable Diffusion or OpenAI's Whisper model for speech recognition. Intel says its chips use less power than Nvidia's. Intel said that the new Gaudi 3 chips would be available to customers in the third quarter, and companies including Dell, Hewlett Packard Enterprise, and Supermicro will build systems with the chips. Intel didn't provide a price range for Gaudi 3.

Gaudi 3 is built on a five nanometer process, a relatively recent manufacturing technique, suggesting that the company is using an outside foundry to manufacture the chips. In addition to designing Gaudi 3, Intel also plans to manufacture AI chips, potentially for outside companies, at a new Ohio factory expected to open in 2027 or 2028, CEO Patrick Gelsinger told reporters last month. "We do expect it to be highly competitive" with Nvidia's latest chips, said Das Kamhout, vice president of Xeon software at Intel, on a call with reporters. "From our competitive pricing, our distinctive open integrated network on chip, we're using industry-standard Ethernet. We believe it's a strong offering."

An anonymous reader quotes a report from Ars Technica: Owners of Intel's latest 13th and 14th Gen Core i9 desktop processors have been noticing an increase in game crashes in recent months. It's happening in games like The Finals, Fortnite, and Tekken 8, and has even led Epic Games to issue a support notice to encourage Intel Core i9 13900K and 14900K owners to adjust BIOS settings. Now, Intel says it's investigating the reports. "Intel is aware of problems that occur when executing certain tasks on 13th and 14th generation core processors for desktop PCs, and is analyzing them with major affiliates," says an Intel spokesperson in a statement to ZDNet Korea.

The crashes vary in severity depending on the game, with some titles producing an "out of memory" error, others simply exiting out to the desktop, and some locking up a machine entirely. Most of the games affected seem to be based on the Unreal Engine, which could point to a stability issue that Intel needs to address. The only workarounds that seem to improve stability involve manually downclocking or undervolting Intel's processors. Epic Games has suggested changing the SVID behavior to Intel Fail Safe in the BIOS settings of Asus, Gigabyte, or MSI motherboards. Custom PC builders Power GPU recommend reducing the performance core ratio limit, which seems to help with stability in certain games.

An anonymous reader shares a report: Microsoft is getting ready to fully unveil its vision for "AI PCs" next month at an event in Seattle. Sources familiar with Microsoft's plans tell The Verge that Microsoft is confident that a round of new Arm-powered Windows laptops will beat Apple's M3-powered MacBook Air both in CPU performance and AI-accelerated tasks. After years of failed promises from Qualcomm, Microsoft believes the upcoming Snapdragon X Elite processors will finally offer the performance it has been looking for to push Windows on Arm much more aggressively. Microsoft is now betting big on Qualcomm's upcoming Snapdragon X Elite processors, which will ship in a variety of Windows laptops this year and Microsoft's latest consumer-focused Surface hardware.

Microsoft is so confident in these new Qualcomm chips that it's planning a number of demos that will show how these processors will be faster than an M3 MacBook Air for CPU tasks, AI acceleration, and even app emulation. Microsoft claims, in internal documents seen by The Verge, that these new Windows AI PCs will have "faster app emulation than Rosetta 2" -- the application compatibility layer that Apple uses on its Apple Silicon Macs to translate apps compiled for 64-bit Intel processors to Apple's own processors.

App emulation has been a big problem for Windows on Arm over the past decade, but Microsoft did deliver x64 app emulation for Windows 11 more than two years ago. This helps ensure apps can run on Windows on Arm devices when there isn't a native ARM64 version. Native Arm apps are key for improved performance on upcoming Windows on Arm laptops, and Google has just recently released its own ARM64 version of Chrome ready for these upcoming devices.

An anonymous reader writes: The massive GPU clusters needed to train Stability AI's popular text-to-image generation model Stable Diffusion are apparently also at least partially responsible for former CEO Emad Mostaque's downfall -- because he couldn't find a way to pay for them. According to an extensive expose citing company documents and dozens of persons familiar with the matter, it's indicated that the British model builder's extreme infrastructure costs drained its coffers, leaving the biz with just $4 million in reserve by last October. Stability rented its infrastructure from Amazon Web Services, Google Cloud Platform, and GPU-centric cloud operator CoreWeave, at a reported cost of around $99 million a year. That's on top of the $54 million in wages and operating expenses required to keep the AI upstart afloat.

What's more, it appears that a sizable portion of the cloudy resources Stability AI paid for were being given away to anyone outside the startup interested in experimenting with Stability's models. One external researcher cited in the report estimated that a now-cancelled project was provided with at least $2.5 million worth of compute over the span of four months. Stability AI's infrastructure spending was not matched by revenue or fresh funding. The startup was projected to make just $11 million in sales for the 2023 calendar year. Its financials were apparently so bad that it allegedly underpaid its July 2023 bills to AWS by $1 million and had no intention of paying its August bill for $7 million. Google Cloud and CoreWeave were also not paid in full, with debts to the pair reaching $1.6 million as of October, it's reported.

It's not clear whether those bills were ultimately paid, but it's reported that the company -- once valued at a billion dollars -- weighed delaying tax payments to the UK government rather than skimping on its American payroll and risking legal penalties. The failing was pinned on Mostaque's inability to devise and execute a viable business plan. The company also failed to land deals with clients including Canva, NightCafe, Tome, and the Singaporean government, which contemplated a custom model, the report asserts. Stability's financial predicament spiraled, eroding trust among investors, making it difficult for the generative AI darling to raise additional capital, it is claimed. According to the report, Mostaque hoped to bring in a $95 million lifeline at the end of last year, but only managed to bring in $50 million from Intel. Only $20 million of that sum was disbursed, a significant shortfall given that the processor titan has a vested interest in Stability, with the AI biz slated to be a key customer for a supercomputer powered by 4,000 of its Gaudi2 accelerators. The report goes on to mention further fundraising challenges, issues retaining employees, and copyright infringement lawsuits challenging the company's future prospects. The full expose can be read via Forbes (paywalled).

Intel on Tuesday disclosed $7 billion in operating losses for its foundry business in 2023, "a steeper loss than the $5.2 billion in operating losses the year before," reports Reuters. "The unit had revenue of $18.9 billion for 2023, down 31% from $27.49 billion the year before." From the report: Intel shares were down 4.3% after the documents were filed with the U.S. Securities and Exchange Commission (SEC). During a presentation for investors, Chief Executive Pat Gelsinger said that 2024 would be the year of worst operating losses for the company's chipmaking business and that it expects to break even on an operating basis by about 2027. Gelsinger said the foundry business was weighed down by bad decisions, including one years ago against using extreme ultraviolet (EUV) machines from Dutch firm ASML. While those machines can cost more than $150 million, they are more cost-effective than earlier chip making tools.

Partially as a result of the missteps, Intel has outsourced about 30% of the total number of wafers to external contract manufacturers such as TSMC, Gelsinger said. It aims to bring that number down to roughly 20%. Intel has now switched over to using EUV tools, which will cover more and more production needs as older machines are phased out. "In the post EUV era, we see that we're very competitive now on price, performance (and) back to leadership," Gelsinger said. "And in the pre-EUV era we carried a lot of costs and (were) uncompetitive." Editor's note: This story has been corrected to change the 2022 revenue figure for Intel Foundry to $27.49 billion, as reflected in the source article. We apologize for the math error.

"After being in development since 2019, the huge NetBSD 10.0 is out today as a wonderful Easter surprise," reports Phoronix: NetBSD 10 provides WireGuard support, support for many newer Arm platforms including for Apple Silicon and newer Raspberry Pi boards, a new Intel Ethernet drive, support for Realtek 2.5GbE network adapters, SMP performance improvements, automatic swap encryption, and an enormous amount of other hardware support improvements that accumulated over the past 4+ years.

Plus there is no shortage of bug fixes and performance optimizations with NetBSD 10. Some tests of NetBSD 10.0 in development back during 2020 showed at that point it was already 12% faster than NetBSD 9.
"A lot of development went into this new release," NetBSD wrote on their blog, saying "This also caused the release announcement to be one of the longest we ever did."

Among the new userspace programs is warp(6), which they describe as a "classic BSD space war game (copyright donated to the NetBSD Foundation by Larry Wall)."

Silicon Valley chiefs are swarming the Capitol to try to sway lawmakers on the dangers of falling behind in the AI race. From a report: In recent weeks, American lawmakers have moved to ban the Chinese-owned app TikTok. President Biden reinforced his commitment to overcome China's rise in tech. And the Chinese government added chips from Intel and AMD to a blacklist of imports. Now, as the tech and economic cold war between the United States and China accelerates, Silicon Valley's leaders are capitalizing on the strife with a lobbying push for their interests in another promising field of technology: artificial intelligence.

On May 1, more than 100 tech chiefs and investors, including Alex Karp, the head of the defense contractor Palantir, and Roelof Botha, the managing partner of the venture capital firm Sequoia Capital, will come to Washington for a daylong conference and private dinner focused on drumming up more hawkishness toward China's progress in A.I. Dozens of lawmakers, including Speaker Mike Johnson, Republican of Louisiana, will also attend the event, the Hill & Valley Forum, which will include fireside chats and keynote discussions with members of a new House A.I. task force.

Tech executives plan to use the event to directly lobby against A.I. regulations that they consider onerous, as well as ask for more government spending on the technology and research to support its development. They also plan to ask to relax immigration restrictions to bring more A.I. experts to the United States. The event highlights an unusual area of agreement between Washington and Silicon Valley, which have long clashed on topics like data privacy, children's online protections and even China.

An anonymous reader shares a report:Intel, Microsoft, Qualcomm, and AMD have all been pushing the idea of an "AI PC" for months now as we head toward more AI-powered features in Windows. While we're still waiting to hear the finer details from Microsoft on its big plans for AI in Windows, Intel has started sharing Microsoft's requirements for OEMs to build an AI PC -- and one of the main ones is that an AI PC must have Microsoft's Copilot key. Microsoft wants its OEM partners to provide a combination of hardware and software for its idea of an AI PC. That includes a system that comes with a Neural Processing Unit (NPU), the latest CPUs and GPUs, and access to Copilot. It will also need to have the new Copilot key that Microsoft announced earlier this year.

This requirement means that some laptops, like Asus' new ROG Zephyrus, have already shipped with Intel's new Core Ultra chips and aren't technically AI PCs in the eyes of Microsoft's strict requirements because they don't have a Copilot key. But they're still AI PCs in Intel's eyes. "Our joint aligned definition, Intel and Microsoft, we've aligned on Core Ultra, Copilot, and Copilot key," explains Todd Lewellen, head of the PC ecosystem at Intel, in a press briefing with The Verge. "From an Intel perspective our AI PC has Core Ultra and it has an integrated NPU because it is unlocking all kinds of new capabilities and functions in the AI space. We have great alignment with Microsoft, but there are going to be some systems out there that may not have the physical key on it but it does have our integrated NPU."

Court filings unsealed last week allege Meta created an internal effort to spy on Snapchat in a secret initiative called "Project Ghostbusters." Gizmodo: Meta did so through Onavo, a Virtual Private Network (VPN) service the company offered between 2016 and 2019 that, ultimately, wasn't private at all. "Whenever someone asks a question about Snapchat, the answer is usually that because their traffic is encrypted we have no analytics about them," said Mark Zuckerberg in an email to three Facebook executives in 2016, unsealed in Meta's antitrust case on Saturday. "It seems important to figure out a new way to get reliable analytics about them... You should figure out how to do this."

Thus, Project Ghostbusters was born. It's Meta's in-house wiretapping tool to spy on data analytics from Snapchat starting in 2016, later used on YouTube and Amazon. This involved creating "kits" that can be installed on iOS and Android devices, to intercept traffic for certain apps, according to the filings. This was described as a "man-in-the-middle" approach to get data on Facebook's rivals, but users of Onavo were the "men in the middle."

Meta's Onavo unit has a history of using invasive techniques to collect data on Facebook's users. Meta acquired Onavo from an Israeli firm over 10 years ago, promising users private networking, as most VPNs do. However, the service was reportedly used to spy on rival social media apps through tens of millions of people who downloaded Onavo. It gave Facebook valuable intel about competitors, and this week's court filings seem to confirm that. A team of senior executives and roughly 41 lawyers worked on Project Ghostbusters, according to court filings. The group was heavily concerned with whether to continue the program in the face of press scrutiny. Facebook ultimately shut down Onavo in 2019 after Apple booted the VPN from its app store.

Tobias Mann reports via The Register: The land surrounding a nuclear power plant might not sound like prime real estate, but as more bit barns seek to trim costs, it's poised to become a rather hot commodity. All datacenters are energy-hungry but with more watt-greedy AI workloads on the horizon, nuclear power has fresh appeal, especially for hyperscalers. Such a shift in power also does wonders for greenwashing narratives around net-zero operations. While not technically renewable, nuclear power does have the benefit of being carbon-free, not to mention historically reliable -- with a few notable exceptions of course. All of these are purported benefits cited by startup NE Edge, which has been fighting for more than a year to be able to build a pair of AI datacenters adjacent to a 2GW Millstone nuclear power plant in Waterford, Connecticut.

According to the Hartford Courant, NE Energy has secured $1.6 billion to construct the switching station and bit barns, which will span 1.2 million square feet in total. NE Energy will reportedly spend an equivalent sum on between 25,000 and 35,000 servers. Considering the price of GPU systems from Nvidia, AMD, and Intel, we suspect that those figures probably refer to the number of GPUs. We've asked NE Edge for more information. NE Energy has faced local challenges getting the project approved because residents are concerned the project would end up increasing the cost of electricity. The facilities will reportedly consume as much as 13 percent of the plant's output. The project's president Thomas Quinn attempted to quell concerns, arguing that by connecting directly to the plants, NE Energy will be able to negotiate prices that make building such a power hungry facility viable in Connecticut. NE Energy has also committed to paying a 12.08 percent premium to the town on top of what it pays Dominion for power, along with other payments said to total more than $1 billion over the next 30 years. But after initially denying the sale of land to NE Edge back in January over a lack of information regarding the datacenter project, it's reported that the town council has yet to tell the company what information it is after.

China has introduced new guidelines that will mean US microprocessors from Intel and AMD are phased out of government PCs and servers [Editor's note: the link may be paywalled; non-paywalled source], as Beijing ramps up a campaign to replace foreign technology with homegrown solutions. From a report: The stricter government procurement guidance also seeks to sideline Microsoft's Windows operating system and foreign-made database software in favour of domestic options. It runs alongside a parallel localisation drive under way in state-owned enterprises. The latest purchasing rules represent China's most significant step yet to build up domestic substitutes for foreign technology and echo moves in the US as tensions increase between the two countries. Washington has imposed sanctions on a growing number of Chinese companies on national security grounds, legislated to encourage more tech to be produced in the US and blocked exports of advanced chips and related tools to China.

An anonymous reader shares a report: Nvidia earned its $2.2 trillion market cap by producing AI chips that have become the lifeblood powering the new era of generative AI developers from startups to Microsoft, OpenAI and Google parent Alphabet. Almost as important to its hardware is the company's nearly 20 years' worth of computer code, which helps make competition with the company nearly impossible. More than 4 million global developers rely on Nvidia's CUDA software platform to build AI and other apps. Now a coalition of tech companies that includes Qualcomm, Google and Intel, plans to loosen Nvidia's chokehold by going after the chip giant's secret weapon: the software that keeps developers tied to Nvidia chips.

They are part of an expanding group of financiers and companies hacking away at Nvidia's dominance in AI. "We're actually showing developers how you migrate out from an Nvidia platform," Vinesh Sukumar, Qualcomm's head of AI and machine learning, said in an interview with Reuters. Starting with a piece of technology developed by Intel called OneAPI, the UXL Foundation, a consortium of tech companies, plans to build a suite of software and tools that will be able to power multiple types of AI accelerator chips, executives involved with the group told Reuters. The open-source project aims to make computer code run on any machine, regardless of what chip and hardware powers it.

"It's about specifically - in the context of machine learning frameworks - how do we create an open ecosystem, and promote productivity and choice in hardware," Google's director and chief technologist of high-performance computing, Bill Hugo, told Reuters in an interview. Google is one of the founding members of UXL and helps determine the technical direction of the project, Hugo said. UXL's technical steering committee is preparing to nail down technical specifications in the first half of this year. Engineers plan to refine the technical details to a "mature" state by the end of the year, executives said. These executives stressed the need to build a solid foundation to include contributions from multiple companies that can also be deployed on any chip or hardware.

An anonymous reader shared this report from The Register: Chinese spies exploited a couple of critical-severity bugs in F5 and ConnectWise equipment earlier this year to sell access to compromised U.S. defense organizations, UK government agencies, and hundreds of other entities, according to Mandiant.

The Google-owned threat hunters said they assess, "with moderate confidence," that a crew they track as UNC5174 was behind the exploitation of CVE-2023-46747, a 9.8-out-of-10-CVSS-rated remote code execution bug in the F5 BIG-IP Traffic Management User Interface, and CVE-2024-1709, a path traversal flaw in ConnectWise ScreenConnect that scored a perfect 10 out of 10 CVSS severity rating.

UNC5174 uses the online persona Uteus, and has bragged about its links to China's Ministry of State Security (MSS) — boasts that may well be true. The gang focuses on gaining initial access into victim organizations and then reselling access to valuable targets... Just last month, Mandiant noticed the same combination of tools, believed to be unique to this particular Chinese gang, being used to exploit the ConnectWise flaw and compromise "hundreds" or entities, mostly in the U.S. and Canada. Also between October 2023 and February 2024, UNC5174 exploited CVE-2023-22518 in Atlassian Confluence, CVE-2022-0185 in Linux kernels, and CVE-2022-3052, a Zyxel Firewall OS command injection vulnerability, according to Mandiant.

These campaigns included "extensive reconnaissance, web application fuzzing, and aggressive scanning for vulnerabilities on internet-facing systems belonging to prominent universities in the U.S., Oceania, and Hong Kong regions," the threat intel team noted.
More details from The Record. "One of the strangest things the researchers found was that UNC5174 would create backdoors into compromised systems and then patch the vulnerability they used to break in. Mandiant said it believes this was an 'attempt to limit subsequent exploitation of the system by additional unrelated threat actors attempting to access the appliance.'"

"There is a new side channel attack against Apple 'M' series CPUs that does not appear to be fixable without a major performance hit," writes Slashdot reader EncryptedSoldier. SecurityWeek reports: A team of researchers representing several universities in the United States has disclosed the details of a new side-channel attack method that can be used to extract secret encryption keys from systems powered by Apple CPUs. The attack method, dubbed GoFetch, has been described as a microarchitectural side-channel attack that allows the extraction of secret keys from constant-time cryptographic implementations. These types of attacks require local access to the targeted system. The attack targets a hardware optimization named data memory-dependent prefetcher (DMP), which attempts to prefetch addresses found in the contents of program memory to improve performance.

The researchers have found a way to use specially crafted cryptographic operation inputs that allow them to infer secret keys, guessing them bits at a time by monitoring the behavior of the DMP. They managed to demonstrate end-to-end key extraction attacks against several crypto implementations, including OpenSSL Diffie-Hellman Key Exchange, Go RSA, and the post-quantum CRYSTALS-Kyber and CRYSTALS-Dilithium. The researchers have conducted successful GoFetch attacks against systems powered by Apple M1 processors, and they have found evidence that the attack could also work against M2 and M3 processors. They have also tested an Intel processor that uses DMP, but found that it's 'more robust' against such attacks.

The experts said Apple is investigating the issue, but fully addressing it does not seem trivial. The researchers have proposed several countermeasures, but they involve hardware changes that are not easy to implement or mitigations that can have a significant impact on performance. Apple told SecurityWeek that it thanks the researchers for their collaboration as this work advances the company's understanding of these types of threats. The tech giant also shared a link to a developer page that outlines one of the mitigations mentioned by the researchers. The researchers have published a paper (PDF) detailing their work.

Ars Technica's Dan Goodin also reported on the vulnerability.

A newly discovered vulnerability baked into Apple's M-series of chips allows attackers to extract secret keys from Macs when they perform widely used cryptographic operations, academic researchers have revealed in a paper published Thursday. From a report: The flaw -- a side channel allowing end-to-end key extractions when Apple chips run implementations of widely used cryptographic protocols -- can't be patched directly because it stems from the microarchitectural design of the silicon itself. Instead, it can only be mitigated by building defenses into third-party cryptographic software that could drastically degrade M-series performance when executing cryptographic operations, particularly on the earlier M1 and M2 generations. The vulnerability can be exploited when the targeted cryptographic operation and the malicious application with normal user system privileges run on the same CPU cluster.

The threat resides in the chips' data memory-dependent prefetcher, a hardware optimization that predicts the memory addresses of data that running code is likely to access in the near future. By loading the contents into the CPU cache before it's actually needed, the DMP, as the feature is abbreviated, reduces latency between the main memory and the CPU, a common bottleneck in modern computing. DMPs are a relatively new phenomenon found only in M-series chips and Intel's 13th-generation Raptor Lake microarchitecture, although older forms of prefetchers have been common for years. Security experts have long known that classical prefetchers open a side channel that malicious processes can probe to obtain secret key material from cryptographic operations. This vulnerability is the result of the prefetchers making predictions based on previous access patterns, which can create changes in state that attackers can exploit to leak information. In response, cryptographic engineers have devised constant-time programming, an approach that ensures that all operations take the same amount of time to complete, regardless of their operands. It does this by keeping code free of secret-dependent memory accesses or structures.

Microsoft has announced two new Surface devices, the Surface Pro 10 for Business and Surface Laptop 6 for Business, both featuring Intel's latest Core Ultra processors, a dedicated Neural Processing Unit (NPU), and a new Copilot key for AI-powered features in Windows 11.

The devices, which will start shipping to commercial customers on April 9th, have been designed exclusively for businesses and will not be sold directly to consumers. The Surface Pro 10 for Business, starting at $1,199, offers a choice between Core Ultra 5 135U and Core Ultra 7 165U options, with up to 64GB of RAM and a 256GB Gen4 SSD. It also features an improved 13-inch display with an antireflective coating and a 1440p front-facing camera with a 114-degree field of view.

The Surface Laptop 6 for Business, also starting at $1,199, is powered by Intel's Core Ultra H-series chips and is available with up to 64GB of RAM and a 1TB Gen4 SSD. The 15-inch model includes two USB-C Thunderbolt 4 ports, while the 13.5-inch model features a single USB-C Thunderbolt 4 port. Both devices have an optional smart card reader and are Microsoft's most easily serviceable Surface devices to date.

Further reading: Microsoft's official blog.

After securing billions in federal grants and loans, Reuters reports that the company is "planning a $100-billion spending spree across four U.S. states" to build and expand its chip manufacturing factories. From the report: The centerpiece of Intel's five-year spending plan is turning empty fields near Columbus, Ohio, into what CEO Pat Gelsinger described to reporters on Tuesday as "the largest AI chip manufacturing site in the world," starting as soon as 2027. Intel's plan will also involve revamping sites in New Mexico and Oregon and expanding operations in Arizona, where longtime rival Taiwan Semiconductor Manufacturing Co is also building a massive factory that it hopes will receive funding from President Joe Biden's push to bring advanced semiconductor manufacturing back to the United States. [...]

Gelsinger said about 30% of the $100-billion plan will be spent on construction costs such as labor, piping and concrete. The remaining will go towards buying chipmaking tools from firms such as ASML, Tokyo Electron, Applied Materials and KLA, among others. Those tools will help bring the Ohio site online by 2027 or 2028, though Gelsinger warned the timeline could slip if the chip market takes a dive. Beyond grants and loans, Intel plans to make most of the purchases from its existing cash flows.

"It will still take three to five years for Intel to become a serious player in the foundry market" for cutting-edge chips, said Kinngai Chan, an analyst at Summit Insights. However, he warned more investment would be needed before Intel could overtake TSMC, adding that the Taiwanese firm could remain the leader for "some time to come." Gelsinger has previously said a second round of U.S. funding for chip factories would likely be needed to re-establish the U.S. as a leader in semiconductor manufacturing, which he reiterated on Tuesday. "It took us three-plus decades to lose this industry. It's not going to come back in three to five years of CHIPS Act" funding, said Gelsinger, who referred to the low-interest-rate funding as "smart capital."

The White House said Wednesday Intel has been awarded up to $8.5 billion in CHIPS Act funding, as the Biden administration ramps up its effort to bring semiconductor manufacturing to U.S. soil. From a report: Intel could receive an additional $11 billion in loans from the CHIPS and Science Act, which was passed in 2022. The awards will be announced by President Joe Biden in Arizona on Wednesday. The money will help "leading-edge semiconductors made in the United States" keep "America in the driver's seat of innovation," U.S. Secretary of Commerce Gina Raimondo said on a call with reporters. Intel and the White House said their agreement is nonbinding and preliminary and could change.

Intel has long been a stalwart of the U.S. semiconductor industry, developing chips that power many of the world's PCs and data center servers. However, the company has been eclipsed in revenue by Nvidia, which leads in artificial intelligence chips, and has been surpassed in market cap by rival AMD and mobile phone chipmaker Qualcomm.

Last month, the only known surviving copy of 32-bit OS/2 from Microsoft was purchased for $650. "Now, two of the internet's experts in getting early PC operating systems running today have managed to fire it up, and you can see the results," reports The Register. From the report: Why such interest in this nearly third-of-a-century old, unreleased OS? Because this is the way the PC industry very nearly went. This SDK came out in June 1990, just one month after Windows 3.0. If 32-bit OS/2 had launched as planned, Windows 3 would have been the last version before it was absorbed into OS/2 and disappeared. There would never have been any 32-bit versions: no Windows NT, no Windows 95; no Explorer, no Start menu or taskbars. That, in turn, might well have killed off Apple as well. No iPod, no iPhone, no fondleslabs. Twenty-first century computers would be unimaginably different. The surprise here is that we can see a glimpse of this world that never happened. The discovery of this pre-release OS shows how very nearly ready it was in 1990. IBM didn't release its solo version until April 1992, the same month as Windows 3.1 -- but now, we can see it was nearly ready two years earlier.

That's why Michal Necasek of the OS/2 Museum called his look The Future That Never Was. He uncovered a couple of significant bugs, but more impressively, he found workarounds for both, and got both features working fine. OS/2 2 could run multiple DOS VMs at once, but in the preview, they wouldn't open -- due to use of an undocumented instruction which Intel did implement in the Pentium MMX and later processors. Secondly, the bundled network client wouldn't install -- but removing a single file got that working fine. That alone is a significant difference between Microsoft's OS/2 2.0 and IBM's version: Big Blue didn't include networking until Warp Connect 3 in 1995.

His verdict: "The 6.78 build of OS/2 2.0 feels surprisingly stable and complete. The cover letter that came with the SDK stressed that Microsoft developers had been using the OS/2 pre-release for day-to-day work." Over at Virtually Fun, Neozeed also took an actual look at Microsoft OS/2 2.0, carefully recreating that screenshot from PC Magazine in May 1990. He even managed to get some Windows 2 programs running, although this preview release did not yet have a Windows subsystem. On his Internet Archive page, he has disk images and downloadable virtual machines so that you can run this yourself under VMware or 86Box.