July 04, 2026: Nasdaq And S&P At Key Technical Levels

Last week, Dow Jones rose 2%, S&P 500 gained 1.8%, and Nasdaq composite was up 2.1%. Investors rotated out of tech names, Nasdaq ended the week weaker than it started. Electric scooter provider, Lime, listed at a US$1.6 billion valuation anchored by Uber. Internet conglomerate BendingSpoons jumped 11% on its IPO. Quantum leader IQM went public via a US$1.9 billion SPAC. Rocket Lab is acquiring satellite operator Iridium for US$8 billion in cash and stock. Schneider Electric is buying industrial AI firm Cognite for US3.1 billion. SpaceX will join the Nasdaq 100 on July 7, triggering a projected US$4.3 billion in passive inflows and compounding short-seller paper losses, currently at an estimated US$760 million. OpenAI proposed handing a 5% equity stake to the U.S. government to defuse regulatory friction. The White House lifted export controls on Anthropic’s Fable 5 model. Tesla deliveries rebounded 25% to 480,000 vehicles in Q2. Management at Tesla capped individual employee AI spending at US$200 per week to curb soaring token costs. Anthropic is exploring custom chips with Samsung. Meta is planning an AI cloud infrastructure play. SoftBank launched an entity to rent U.S. computing capacity. South Korea unveiled an US$880 billion, 10-year mega tech investment plan. China’s Meituan debuted a 1.6 trillion parameter LLM trained entirely on domestic hardware. Coinbase halved its AI spend by shifting to Chinese open-weight models. Tencent locked in a US$2.94 billion DRAM supply deal with CXMT. Microsoft launched a US$2.5 billion AI consulting unit while also planning thousands of sales and engineering layoffs. Strategy announced a potential US$1.25 billion Bitcoin sale. A Stripe-Visa-Mastercard consortium launched the “Open USD” stablecoin to challenge Circle and Tether. SoFi integrated automated, natural-language investing strategies into its app by acquiring fintech platform Composer. In Canada, Sophic client, Kraken Robotics closed its $615 million acquisition of Covelya Group, updating 2026 guidance after securing $30 million in combined new orders and initiating plans for a TSX listing by year-end.

Sophic Client Kraken Robotics (PNG-TSXV, KRKNF-OTC) announces closing of strategic acquisition of Covelya Group Limited, updated 2026 guidance and appointments to executive team.

Kraken announced the closing of its previously announced acquisition of Covelya Group Limited (“Covelya Group”), for approximately $615 million, subject to closing adjustments (the “Acquisition”). Unless otherwise specified, all dollar amounts in this release are denominated in Canadian dollars. Management Comments: “This acquisition positions Kraken as a global provider of mission-critical, dual-use subsea intelligence solutions,” said Greg Reid, CEO of Kraken Robotics. “Since announcing the transaction, we have received positive feedback from customers who are looking forward to working with our combined engineering teams on integrated subsea technology solutions. We welcome the new employees to the team and look forward to the many benefits this combination can provide. Together, Kraken and Covelya Group bring complementary products, technological capabilities, and customer relationships that we expect will strengthen Kraken’s growth potential and long-term outlook. This positive long-term outlook is further supported by the expected increase in defence budgets globally, including growing investment in autonomous underwater systems.” Strategic Rationale: The Acquisition aligns with Kraken’s strategy to deliver value to customers through a portfolio of dual-use technologies and a culture of innovation. As previously announced, the strategic benefits to Kraken from the Acquisition include the following: Allows for deeper customer relationships in the fast-growing defence and maritime surveillance market. Expands product offering and Kraken’s total addressable market in subsea technology. Adds strategic locations for geographic expansion and improves business diversification. Bolsters technical capabilities with an experienced engineering team and highly advanced facilities. Financial accretion across key metrics, including $10 million of cost synergies within 24 months. New Product Orders And Updated 2026 Financial Guidance: Since reporting its Q1 2026 results on May 28, 2026, Kraken and Covelya have secured additional product orders of approximately $13 million and $17 million, respectively. These awards bring announced orders in 2026 to approximately $110 million for Kraken and $182 million for Covelya Group. Gross profit margins associated with these orders are consistent with historical gross profit margins. Kraken is updating its 2026 guidance to reflect the Acquisition’s July 2, 2026, closing date and the inclusion of Covelya Group, which was excluded from the Company’s prior guidance. As previously outlined at announcement, the Acquisition is expected to be accretive across key financial metrics and is expected to generate low-to-mid double-digit EPS accretion in 2027, after including the full impact of expected cost synergies. The Company continues to maintain a strong balance sheet, with minimal net debt following the drawdown of the New Credit Facility, as defined below, and financial flexibility to fund future growth opportunities. Leadership Team And Organizational Structure: As part of its integration with Covelya Group and its subsidiary companies, Kraken is implementing a new organizational structure and strengthening its leadership team. The new structure will consist of Kraken Group, which will focus on financial and organizational governance, and a clearly delineated Kraken Robotics operating business, whose business units will focus on operational excellence, strategic execution, and financial performance. These changes are designed to combine the strengths and talents of both organizations while creating a more efficient and scalable platform to support long-term growth. Closing Details And Subscription Receipt Conversion: The approximately $615 million purchase price for the Acquisition, prior to closing adjustments, was comprised of approximately $480 million in cash and approximately $135 million through the issuance of 15,882,352 common shares of Kraken (each a “Common Share”) at a deemed issue price of $8.50 per Common Share to Covelya Group’s shareholder. The cash portion of the purchase price was funded from the net proceeds of Kraken’s $402.5 million bought deal public offering of subscription receipts (the “Subscription Receipts”), which closed on March 12, 2026, interest earned on such net proceeds, and borrowings under the New Credit Facility. In connection with the closing of the Acquisition, Kraken also closed amendments to its existing credit facility to create a new committed, secured, non-revolving term credit facility in the amount of $125 million (the “New Credit Facility”), increase its existing revolving credit facility from $35 million to $60 million, and to extend the term of the revolving credit facility to March 2031, among other amendments. The Company drew down the New Credit Facility in full, with the proceeds applied to pay a portion of the cash consideration for the Acquisition. Kraken plans to report its Q2 2026 results in late August 2026 and its Q3 2026 results, which will include Covelya’s contribution, in late November 2026. With the Acquisition now complete, Kraken intends to apply to list its Common Shares on the Toronto Stock Exchange (“TSX”), subject to satisfying applicable TSX listing requirements and receiving TSX approval. The Company expects the process to be completed by year-end 2026 or early 2027. https://tinyurl.com/ykevr7pm

Quantum tech firm BTQ (BTQ-NASDAQ, BTQ-CBOE) to acquire France’s QPerfect.

Vancouver’s BTQ Technologies (BTQ) is one step closer to acquiring France’s QPerfect after the quantum company received approval for the acquisition from the French foreign direct investment (FDI) authorities. On July 1, BTQ announced it has received regulatory approval to acquire the remainder of the Strasbourg-based quantum computing company, after announcing plans to take over the company last year. FDI approval comes with a laundry list of stipulations from France’s Ministry for the Economy and Finance that the ministry says are “reflect the will to preserve QPerfect’s technological assets” in a sector “critical to the technological sovereignty of France and Europe.” They include maintenance of QPerfect’s French office, the appointment of a French corporate officer, continuing research and development activity, and protecting QPerfect’s intellectual property. BTQ is poised to pay a $30 million price tag for the acquisition, broken down into a cash payment of roughly $3.3 million, and the remainder paid through the issuance of just over 2.1 million common shares of BTQ. Following the thread: BTQ is framing the QPerfect acquisition as a strategic investment aimed at shoring up the company’s capability to build the tech needed to transition from classical computing infrastructure to quantum computing systems. Specifically, BTQ cited three additions to the company’s quantum infrastructure stack, including MIMIQ, QPerfect’s quantum emulator which enables researchers to design, test, and validate quantum algorithms, Digital Twin, QPerfect’s system modelling capability, and the company’s Quantum Logical Unit, a control framework designed to support development of scalable, fault-tolerant quantum systems. Final thought: With the acquisition of QPerfect, BTQ is showing that the momentum of the quantum sector continues to grow as companies look to consolidate, expand, and achieve larger market dominance. For BTQ, which is valued at north of US$700 million, QPerfect’s acquisition won’t necessarily impact BTQ’s bottom line, but it will impact the company’s ability to scale its presence as a quantum computing leader at a time when governments on both sides of the Atlantic are looking at quantum as the next technological frontier. https://tinyurl.com/5hbz8ptf

Dominion Dynamics lands $139 million in Canada’s largest defence-tech Series A.

Six months after closing a major seed round, Ottawa-based defence tech company Dominion Dynamics has secured $139 million (US$100 million) in Series A financing to advance its mission to become Canada’s first line of defence in the Arctic. Dominion’s all-equity, all-primary capital Series A was led by existing backer Georgian, with participation from new investors Valor Equity Partners, Valor Atreides AI Fund, Expeditions, Lakestar, OMERS Ventures, BDC Capital’s Strong North Fund, Deloitte Ventures, Royal Bank of Canada, and JDY Capital. The majority Canadian round also saw follow-on support from British Columbia Investment Management Corporation, Bessemer Venture Partners, Garage Capital, Golden Ventures, and Silent Ventures. https://tinyurl.com/22v57b2j

Stathera raises US$55 million for semiconductor clock tech amid data centre boom.

The Montréal startup, founded in 2020 as a spin-out of Nxtsens Microsystems, announced on Tuesday that it had raised an oversubscribed Series B round led by new investor San Francisco-based Maverick Silicon. Existing investors BDC Capital, Celesta Capital, MediaTek Innovation Fund, TXC Corporation, and Ultratech Capital Partners also participated in the all-equity, all-primary round, bringing its total amount raised to US$75 million. Stathera develops semiconductor timing technology—a kind of clock for computing chips—for various uses, from mobile devices to hyperscale AI data centres. The company said it’ll use the funding to mass-produce its second-generation silicon timing components and open an office in Silicon Valley. https://tinyurl.com/9unmsmfd

US finance app SoFi acquires Composer to add AI investing strategies.

Last week, SoFi revealed Composer by SoFi, a new AI-powered investing platform within its financial app, built on its acquisition of Composer. The new tool will allow SoFi’s 14 million members to turn their investing ideas into full strategies. Using natural language, users can define their strategy, and the Composer tool walks them through building, testing, and automating a rules-based roadmap to invest based on their thesis. https://tinyurl.com/5n6kr96z

Lime shares open up 9%.

Shares in the company behind Lime electric scooters and bikes began trading on Wednesday, rising nearly 10% in early trading to above US$27. The company, Neutron Holdings, said late Tuesday it had priced its initial public offering at US$25 a share, the midpoint of its expected range, selling US$174 million in new and existing shares. The offering values the company at US$1.6 billion. The IPO of the nine-year-old company is the latest venture-backed offering to test investor appetite after the massive listing from SpaceX earlier in June. It previously raised US$1.5 billion from investors including Andreessen Horowitz, GV, Fidelity and Coatue Management, according to PitchBook. In the run-up to the listing, Lime got a boost of support from existing backer Uber, which moved to become an anchor investor in the IPO. Uber will own about one-fifth of the company after the IPO. https://tinyurl.com/2vneat9p

Bending Spoons stock opens up 11%.

Shares of Italian conglomerate Bending Spoons, which is rolling up aging internet businesses, opened up 11% in its public debut on Wednesday. The company had priced its initial public offering at US$29 a share, just above the preliminary range of US$26 to US$28 a share. The IPO price values the company at US$19.4 billion. Bending Spoons has swallowed up well known firms such as AOL, Eventbrite, Vimeo and Evernote. In the first quarter, those acquisitions more than doubled its revenue to US$601 million, and helped it report an operating profit of US$120 million compared with an operating loss a year earlier. A successful IPO would show that it’s not only AI-related firms which can do well on the public markets. The real test, however, will be where Bending Spoons trades in the coming weeks. https://tinyurl.com/2cf7388p

IQM, Europe’s first public quantum company, admits the future of the tech is uncertain.

IQM, a full-stack quantum company out of Finland, went public on the Nasdaq Thursday via a SPAC merger at a valuation of about US$1.9 billion. But share prices didn’t pop. They spent most the day below the IPO price — a lukewarm welcome. SPAC mergers are often not immediately popular with retail investors these days. But this fizzle was arguably fueled by IQM’s own admission in its prospectus that “large-scale commercial traction of quantum computing technology may never occur.” In fairness, this warning applies to all quantum companies. Yet, that hasn’t stopped the industry, including IQM, from acquiring customers, who use the tech as it is today for tasks like simulations and optimizations. IQM, which sells actual physical computers, as well as a cloud service, has customers like VTT Technical Research Centre of Finland and Leibniz Supercomputing Centre in Germany. “We sell computers into advanced supercomputing centers and data centers, and we sell computing time through the cloud,” its CEO and co-founder Jan Goetz told TechCrunch. Having grown from eight customers in 2024 to 22 in 2025 is a fair motive for celebration in IQM’s circles, especially when two recent customers are from the private sector. But it also suggests that demand won’t scale until the “quantum advantage” — when quantum chips start outperforming classical computers for a larger range of complex and lengthy tasks, unlocking use cases from biotech to fintech, while potentially upending encryption. But no one, not even a company making quantum computers, can say when that might be. This hasn’t stopped investors from doubling down on quantum companies public and private, further encouraged by President Trump’s recent executive orders to accelerate the timeline for quantum. In response, the U.S. Department of Energy (DOE) has committed to deploying “the world’s first fault-tolerant, scientifically relevant quantum computer” by 2028. While this follows similar announcements from France, Germany, and the U.K., Trump’s orders carry extra weight for IQM, which has recently established a quantum tech center in Maryland and deployed a computer at Oak Ridge National Laboratory, which is part of the DOE. “We can benefit directly from it,” Goetz said. https://tinyurl.com/ycxemjtz

OpenAI has discussed size of stake U.S. government could take.

OpenAI has discussed the government taking an equity stake in the company with a range of percentages put forward, according to a person familiar with the talks, who described the discussions as preliminary. The Financial Times earlier reported that OpenAI CEO had proposed a 5% stake. In a previous white paper published in April, OpenAI had floated the idea of a public wealth fund seeded by “diversified, long-term assets” from AI companies, with returns distributed to citizens, though did not provide further specifics. OpenAI’s proposal to the Trump administration has been in line with the structure of the public fund in its white paper, according to the person. Anthropic released a similar proposal in June but has not discussed the government taking a stake with the Trump administration, according to a person familiar with that company. OpenAI’s white paper deviates from the Trump administration’s investments in companies including Intel, where the government bought shares in the company. It’s more similar to a bill introduced by Sen. Bernie Sanders (I-Vt.) that would establish a sovereign wealth fund seeded with a 50% ownership stake in top AI companies. OpenAI is currently working with the White House on the rollout of its newest model, GPT-5.6. The Trump administration ordered OpenAI to do a staggered release of the model. https://tinyurl.com/jfk364m

Comcast to split into two companies.

Comcast Corp. is splitting into two companies, spinning off its NBCUniversal and European Sky operations into a separate company from the core cable TV-broadband business, the company announced Monday. The split comes as Comcast stock is trading at its lowest point since 2014 and as merger activity in the film and TV market has intensified. Comcast stock jumped 22% in pre-market trading in response to the news. Word of the split comes just six months after Comcast spun off its cable channels, housed in NBCUniversal, into a company called Versant. The two restructurings return Comcast to its position before its 2011 acquisition of NBCUniversal, which turned it into a diversified entertainment conglomerate. But in recent years, cord-cutting and the rise of streaming services like Netflix has transformed Comcast’s business, shrinking its cable TV business and eroding its traditional TV channel operation. https://tinyurl.com/3d5hnrrv

Schneider Electric agrees to buy industrial AI firm Cognite for US$3.1 billion.

Schneider Electric, a French energy management and power equipment company, said Tuesday that it agreed to buy Cognite Holding B.V, a Norwegian startup that develops AI agents, apps and models to solve complex industrial data problems. The transaction, an all-cash deal valued at US$3.1 billion, should close in the coming quarters. Schneider provides electrical infrastructure, such as advanced cooling technology and high voltage electricity, used in AI data centers. In May, it announced a €75 billion (roughly US$87.5 billion) commitment to build 5 gigawatts of AI data center capacity in France, partnering with SoftBank Group. Cognite was founded in 2016 and sells its software to companies for industries such as oil, gas, shipping, manufacturing, power and utilities. It raised US$225 million from investors including Aramco, TCV, Accel, TRK Group and the California Energy Commission. It was valued at US$1.53 billion as of February 2022, according to PitchBook. Last year it generated more than US$170 million in revenue. The deal is part of a larger trend of engineering giants incorporating AI. Siemens, the automation and software market’s most aggressive competitor, in March 2025 closed its US$10 billion acquisition of Altair Engineering, whose AI software runs simulations and analyzes data for industrial giants. In January Siemens bought Canopus AI to weave machine learning into semiconductor and hardware manufacturing. Competitors Rockwell and Honeywell are also actively working to redefine themselves as AI companies. https://tinyurl.com/ybyk5yfh

Rocket Lab announces deal to buy satellite company Iridium.

Space launch company Rocket Lab is acquiring satellite company Iridium Communications for US$8 billion in cash and stock, a deal that will make it easier for Rocket Lab to compete with SpaceX and its Starlink satellite network. The deal will combine Rocket Lab’s launch vehicles and satellite manufacturing expertise with Iridium’s satellite network and 2.55 million global subscribers. In its press release, Rocket Lab said it plans to continue to scale Iridium’s satellite network and explore new markets. Rocket Lab has made a number of smaller acquisitions recently, including of robotics company Motiv in May and laser optical communications company Mynaric in April. Shares of Iridium jumped more than 23%, while Rocket Lab’s stock climbed as high as 14% on Monday morning. https://tinyurl.com/5eh47zvk

SpaceX set to join Nasdaq 100, paving way for wave of passive buying.

SpaceX will be added to the tech-heavy Nasdaq 100 index on July 7, exchange operator Nasdaq confirmed on Friday, paving the way for a surge in passive investments in ‌Elon Musk’s rocket and AI giant. Inclusion in the index typically boosts the stock price, as exchange-traded funds looking to replicate the index’s performance buy shares of the newly included firm. To make it more attractive for companies seeking U.S. listings, Nasdaq, along with other index providers FTSE Russell and MSCI, relaxed ⁠its entry requirements including profitability, the number of days after a company goes public and the number of shares available for trading. SpaceX, which made its Nasdaq debut on June 12, has swung between sharp losses and small profits over the past three years. Last year, the company reported a net loss of US$4.9 billion. Large Language Model (LLM) makers OpenAI and Anthropic are also expected to file for their initial public offerings this year or next year and likely target valuations of more ‌than US$1 trillion. Investors ⁠buy mutual funds and ETFs, such as Invesco’s QQQ and QQQM, that track the Nasdaq 100, to get broader exposure. J.P. Morgan estimated that SpaceX’s inclusion in the Nasdaq 100 could draw US$4.3 billion in passive inflows. S&P Global said this month that it was not changing the requirements for SpaceX to enter its ⁠major indices, including Wall Street’s benchmark S&P 500 index, and will wait for at least 12 months before even considering it. https://tinyurl.com/dwxpykhy

Short sellers boost bets against SpaceX and it’s already costing them.

Short sellers are betting SpaceX’s will resume its post-debut decline with nearly a third of its tradable shares now sold short — even as those wagers have already cost them nearly three-quarters of a billion dollars in paper losses. The sizeable short position could inject further volatility into the stock, with every US$1 SpaceX share price swing translating to roughly US$200 million in gains or losses for shorts, Ortex estimates. Short sellers, who sell ⁠borrowed shares in the hope of buying them back at a profit when the stock slips, were emboldened after SpaceX shares’ initial burst of strength gave way to weakness and the share price slipped as much as 23% in the days following its June 12 market debut. Short interest now stands at 196 million shares, about 31% of the free float, through Tuesday, up from some 83 million shares, or 13% of the free float, a week ago, Ortex data showed. “(The rise in short bets) is extraordinary for a stock that has been public less than a month,” said Ortex co-founder Peter Hillerberg. SpaceX’s more than US$2 trillion valuation makes it a target for short sellers ‌skeptical ⁠of its rich price tag, but strong retail and institutional interest and Musk’s history of public battles against short sellers make that a risky proposition. SpaceX did not immediately respond to a request for comment. SpaceX shorts are sitting on mark-to-market losses of about US$760 million since the IPO, Ortex estimates. When the stock bottomed near US$153 last week they were up around US$2.5 billion on paper, ⁠but the rebound in SpaceX shares since has wiped all of that out, Ortex data showed. “SpaceX has been a roller coaster for the short sellers,” Hillerberg said. The cost to borrow SpaceX shares, a gauge of demand to short a stock relative to ⁠the supply of shares available to lend, remains relatively cheap at about 1%, Ortex data showed. https://tinyurl.com/u8prv7wy

Tesla vehicle sales rebounded 25% in second quarter.

Tesla’s vehicle deliveries rebounded by 25% to 480,000 in the second quarter, following a sales slump in 2025, the company said on Thursday. Out of the deliveries, 468,000 were Model 3 or Model Y vehicles, while 12,000 were other models including the Cybertruck, Model X and Model S. Tesla discontinued the Model X and Model S during the second quarter. Tesla’s energy storage business also reported deploying 13.5 gigawatt hours worth of batteries in the second quarter, a 41% year-over-year jump. Makers of electric vehicles and plug-in hybrids have benefitted from surging gas prices this year due to the Iran War, even as other U.S. automakers beyond Tesla have been refocusing on combustion vehicles. Tesla is due to report financial results for the second quarter on July 22. Despite the rebound in deliveries, Tesla shares were down 7% on Thursday. https://tinyurl.com/yck89t57

Anthropic in talks with Samsung to manufacture custom AI chip.

Anthropic has begun early-stage work on its own AI chip and held talks with Samsung Electronics as a potential manufacturing partner, following the lead of rival OpenAI in trying to get more control over the costly computing systems behind its models, according to three people with direct knowledge of the project. If the Claude maker goes ahead with the chip, it would be a relative newcomer in developing its own AI server chips, compared to others. Google and Amazon Web Services have both been successful at developing their own chips over many years, while Meta Platforms and Microsoft have also developed their own chips. OpenAI tapped Broadcom to design its own chip in 2024 and last month unveiled the first product of that partnership, Jalapeño, an inference chip built to run large-language models more efficiently. Anthropic is still at the stage of figuring out what it wants the processor to do, how powerful it should be and how it would fit into a server or clusters of servers, according to the three people. It has held discussions with multiple chip design firms but it has yet to move into detailed design, testing and manufacturing work. Such processors are hard to design because engineers must balance speed, power consumption, memory, networking and cooling all at once, and manufacturing reliably at high volume is even harder. https://tinyurl.com/35ncz739

Meta said to plan AI cloud business push.

Meta Platforms is developing plans for a cloud infrastructure business, Bloomberg reported, putting it in direct competition with cloud firms such as Google Cloud, Amazon Web Services, and Microsoft Azure. Meta has spent heavily to expand its AI data centers in recent years, to give it more capacity for its own AI development and operations. But CEO Mark Zuckerberg has signaled in recent months that he could sell some excess capacity to other firms. “It’s definitely on the table,” Zuckerberg said at Meta’s annual shareholder meeting in May. Elon Musk’s SpaceX has taken similar steps, striking deals with firms such as Anthropic and Google to rent out spare capacity originally developed for SpaceX’s own AI unit. Launching a cloud business would help Meta generate revenue that could offset some of the substantial costs associated with its AI data center expansion. The Bloomberg report said Meta is building a unit to sell excess computing capacity to external customers, the report said, citing unnamed sources. Meta is also considering offering access to AI models hosted on its own infrastructure, alongside raw computing capacity. https://tinyurl.com/mvktzmv3

SoftBank plans to rent AI computing capacity to U.S. companies.

Softbank Group Corp. and its telecom unit SoftBank Corp. said Thursday they will start renting AI computing resources to U.S. companies, starting in April. The capacity will come through a new neocloud venture, SB Neo, Inc., with plans to scale to 10 gigawatts to meet the growing demand for AI data centers in the U.S. SoftBank also plans to build gigawatt-scale AI data centers in Japan “as soon as preparations are in place,” Junichi Miyakawa, President and CEO of SoftBank Corp said in a statement. SoftBank is borrowing heavily to invest across the entire physical AI supply chain: chips, server manufacturing, data centers and the energy required to power them. SoftBank’s new computing venture puts it in direct competition with cloud providers like CoreWeave or Nebius Group, which are racing to secure gigawatts of electricity, physical data center real estate and high-end AI chips. https://tinyurl.com/mpfrtt4p

South Korea to invest US$880 billion into chips, robotics and AI over 10 years.

South Korea’s government on Monday announced an 1,350 trillion won (US$880 billion) investment plan into semiconductors, robotics and AI over the next decade, in response to the surging demand for memory chips and the AI infrastructure boom. Samsung Group and SK Group, parent companies of Samsung Electronics and SK Hynix, respectively, two of the world’s leading memory chipmakers, will invest 800 trillion won (US$518 billion) to build four memory chip fabrication plants in the country’s southwest, according to Yonhap News Agency. The government will streamline permits and construction procedures, and invest in infrastructure including electricity and industrial water supplies, in order to bring forward the completion schedule for the new fabs to the mid-2030s, Yonhap reported. The semiconductor investments are part of the country’s “three mega projects” initiative, which also include robotics and data centers. Industrial minister Kim Jung-kwan said South Korea must accelerate mass production of humanoid robotics, in light of rising competition from China. In the long run, the country aims to lift its share of the global humanoid market to 20% from just 1% last year. South Korea is also planning an initial investment of 550 trillion won (US$356 billion) to build 8.4 gigawatts of AI data centers by 2029, and gradually expanding another 10 GW by 2035. https://tinyurl.com/mt8y3sym

Tesla caps employee AI spend at US$200 per week after adoption push.

Tesla told employees last month it would impose a US$200 per week limit for staff’s AI spending beginning July 6, according to an internal memo, a sign that even companies committed to using the technology to transform their operations and products are having to watch their costs. Over the past few months, software engineers at Tesla were often consuming thousands of dollars’ worth of tokens each week, according to two people familiar with the usage. Workers will need a sign-off to spend above the new limit, the memo said, though the tally excludes beta versions of xAI products, the two people said. What’s going on inside Tesla mirrors the whiplash playing out across many businesses as companies like Meta Platforms, Uber and Walmart move from pushing employees to go all-in on using AI to limiting their spending. How quickly Tesla’s arc played out is striking, though, as it initially lagged behind some tech and other corporate giants in its steps to formalize workers’ AI usage and policies. https://tinyurl.com/y4thhrdv

Microsoft commits US$2.5 billion to new applied AI consulting effort.

Microsoft on Thursday announced the creation of a new business unit geared towards helping large customers set up customized AI applications, commercial CEO Judson Althoff announced. Microsoft is spending US$2.5 billion and assigning 6,000 employees to the effort, called Microsoft Frontier Company, Althoff said. The new unit comes as large companies have groused that it can be tricky to configure AI software to their own needs and generate meaningful returns. Even sophisticated firms like Uber and Starbucks have recently complained about struggling to get meaningful returns from their AI spending. The new Microsoft unit mirrors similar efforts by other AI companies like Palantir, Google, Meta, and OpenAI to set up dedicated teams of engineers, consultants, and salespeople to help customers figure out how to configure AI applications for their specific business needs. Those teams, commonly referred to as forward deployed engineers, offer technical support and aim to convince large customers to spend more heavily on AI software. Amazon, Microsoft’s biggest cloud competitor, last week announced the creation of a similar forward deployed engineering effort and said it was committing US$1 billion to the new unit. Althoff said Thursday that Microsoft’s new effort will aim to be even more expansive than those of its competitors. Microsoft sales president Rodrigo Kede Lima, who previously led its Asia sales operation, will head up the effort, Althoff said. https://tinyurl.com/bdhywu45

Microsoft plans thousands of layoffs in sales, engineering, Xbox.

Microsoft is planning to lay off thousands of employees across the company in the coming week, including in sales and engineering units, according to someone briefed on the plans. The layoffs will also include job cuts across Microsoft’s Xbox gaming organization that have been previously reported. Microsoft typically conducts layoffs after the close of its fiscal year, which ended on Tuesday. The company last year laid off more than 9,000 employees in July after cutting 6,000 employees just a few months prior. This year’s layoffs are expected to be significantly smaller, said the person briefed on them, in part because Microsoft already offered a voluntary buyout program for employees near retirement age earlier this year. But salespeople who earn commissions were not eligible for the buyouts, this person added. The cuts come as Microsoft aims to balance costs, including its unprecedented spending on data centers needed to train and run AI applications for itself and its customers. Even with its massive layoffs last year, headcount stayed flat between 2024 and 2025 at around 220,000 employees. However, Microsoft earlier this year froze hiring in sales and other departments, and CFO Amy Hood said in April that its headcount slightly declined from the same period a year earlier and is expected to decline slightly in the year ahead. https://tinyurl.com/ynr3d6fw

Google put limits on Meta’s use of Gemini due to capacity constraints.

Google put limits on Meta’s use of its Gemini AI models a few months ago, saying it couldn’t provide all the capacity that the social media giant wanted, the Financial Times reported. Google restricted other clients as well and has since signed a deal to rent cloud computing capacity from Elon Musk’s SpaceX, the FT said. The report said Meta, had responded to the restrictions by pushing employees to be more careful with their token usage, as The Information has previously reported. Google’s actions show how a shortage of computing capacity is sending ripples across the entire tech sector. https://tinyurl.com/5acmescp

Coinbase cuts AI spending in half as usage rises, Armstrong says.

Coinbase has cut its AI spending “nearly in half” even as it increases the number of tokens it uses, by using various measures to control costs. These measures include defaulting to open-weight models from Chinese firms, Coinbase CEO Brian Armstrong said in an X post on Friday night. Armstrong’s post highlights both the growing usage among U.S. firms of open-weight models coming out of China to control AI spending even as the cost of cutting edge U.S. services rises. Armstrong said Coinbase had experimented with GLM 5.2, a just-released model from Z.ai, and Kimi 2.7, from Beijing-based Moonshot. Armstrong listed a series of other measures Coinbase has taken, including making sure engineers know ho wmany tokens they’re using. “Our engineers can use as many tokens as they want, from whatever model they want, but we’ve made usage visible—and the more you spend on AI, the more impact we expect.” https://tinyurl.com/4tdxk9pw

Anthropic runs pre-clinical drug trials, launches Sonnet 5 model.

Anthropic released Claude Science on Tuesday, a new desktop application to assist biologists with research by writing code, literature reviews and research reports, drawing on databases of biology research and biology-focused AI models. Anthropic has also decided to start running its own drug programs in pre-clinical stages, which are the early parts of drug discovery, focusing on neglected diseases. “These are areas that are outside the scope of what the traditional pharma and biotech landscape might consider attractive targets, but nonetheless have real burden associated with them,” said Eric Kauderer-Abrams, Athropic’s head of life sciences, at an event in San Francisco announcing the product. At the event, Anthropic staff demonstrated using Claude Science to research new drug candidates for a rare disease, and to migrate code at a therapeutics company from one programming language to another. Anthropic CEO Dario Amodei cautioned that as models become better at biology research, that could also enable them to help with “potentially concerning applications” such as creating biological weapons, similar to how the coding capabilities of Anthropic’s latest models have raised concerns about cyberattacks. Separately on Tuesday, Anthropic also released Claude Sonnet 5, the latest generation of its mid-size model, which it said is better at making plans and using browsers than the previous generations. https://tinyurl.com/2827dm86

Chinese AI matches Mythos in cybersecurity, report says.

Security researchers say that Chinese AI systems have matched Anthropic’s Mythos in cybersecurity capabilities, the Wall Street Journal reported, intensifying the competitive threat posed by China’s AI sector to the U.S. The researchers found that the new GLM-2 model from Z.ai can find bugs as well as Mythos, the Journal report said. GLM-2 has become widely used since its release, including by U.S. companies. China’s advance in AI comes as the Trump administration has blocked Anthropic and OpenAI from widely releasing their most up to date models, out of concerns about the potential impact of the models. https://tinyurl.com/awea94mz

Meituan releases new AI model trained entirely on Chinese chips.

Chinese food delivery giant Meituan unveiled its new open-source large language model that it says has been trained entirely on Chinese chips. Meituan, which developed its first LLM last year, said its new model, LongCat-2.0, which has 1.6 trillion parameters, is the first model of this scale to complete the full training as well as inference on domestic chip clusters. Meituan didn’t disclose which Chinese chip suppliers it is working with. Meituan’s move is the latest example of how China is ramping up its effort to reduce its dependence on American AI chips amid Washington’s export controls. Beijing is promoting the use of domestic alternatives to Nvidia, as the country tries to become more self-sufficient in critical areas of technology. In January, Chinese AI developer Zhipu released an open-source AI image model that was trained entirely on Huawei chips. The size of the Zhipu model, which focuses on image generation, is much smaller than Meituan’s new LLM. DeepSeek is also stepping up its efforts to run its AI models on Huawei chips, even though the company continues to train its models on advanced Nvidia chips. https://tinyurl.com/yym5mmev

Firmus And Nvidia partner on massive data center project in Indonesia.

Firmus, an Asia-Pacific neocloud, said on Sunday that it’s building a new data center in Batam, Indonesia, with at least 170,000 of Nvidia’s advanced server chips: a mix of Grace Blackwell and Vera Rubin GPUs and CPUs. In a deal that appears to be the first of its kind, Nvidia is set to earn a revenue share based on chip usage and provide credit support for the project. The project is part of a deepening partnership between Firmus and Nvidia. Founded in 2019, Firmus is a neocloud headquartered in both Australia and Singapore, building data centers in the Asia-Pacific region. The company last raised US$505 million from Coatue, Nvidia and others at a US$5.5 billion valuation in April. It also raised US$10 billion in debt in a Blackstone-led deal in February. Nvidia and Firmus have been working together since 2021, initially to codesign Firmus’ proprietary cooling technology, called the HyperCube, and work together on how to make Nvidia’s chips run more efficiently, according to Firmus cofounder and co-CEO Tim Rosenfield. Rosenfield said Nvidia was eager to support the Indonesia project in part because Firmus is adopting Nvidia’s full stack of hardware and specifications. But, perhaps more importantly, the data center’s customers are exclusively newer companies built from the ground up with AI, termed “AI natives,” and not hyperscalers like Microsoft, Amazon and Google. Rosenfield declined to name specific customers but these customers could include life sciences outfits, robotics firms, AI software companies like Lovable or Cognition and AI model makers like OpenAI and Anthropic. He said capacity sold out within a couple of weeks and is a mix of longer- and shorter-term contracts. The capacity is slated to start coming online in the first quarter of 2027. https://tinyurl.com/35b2a5z7

SpaceX showed investors prototype of Elon Musk’s new AI device.

Elon Musk’s SpaceX has developed a prototype for a handset-like device designed to reshape how humans interact with artificial intelligence that SpaceX has shown investors recently. The rocket and AI company showed the prototype, which features a sleek design that is slimmer than an iPhone, to some investors and other stakeholders ahead of the company’s mega initial public offering, according to people familiar with the matter. The prototype was designed to run on a proprietary operating system and integrate AI technology from SpaceX’s xAI, some of the people said. The device would use a Qualcomm Snapdragon chipset, they said. SpaceX told some investors that the project was at an early stage. The design could change and it is unclear whether such a device will be made. The device prototype SpaceX recently showed investors draws on the philosophy behind Musk’s “everything app,” a concept he championed when acquiring social-media platform Twitter, now X, in 2022, some of the people familiar with the matter said. Often referred to as “super apps,” these software programs are popular in Asia and incorporate services that Americans usually access by downloading separate apps. https://tinyurl.com/4sy68bk8

U.S. eases export curbs on Anthropic’s fable model.

The White House lifted export restrictions on Anthropic’s Fable 5, according to a U.S. official. In a post on X, Anthropic said the U.S. government notified it that it has lifted export controls on its Fable 5 model, along with Mythos 5. The company said it will begin restoring customer access to the models on Wednesday. The government’s move, first reported by Politico, was confirmed by a U.S. official. On June 12, the federal government forced Anthropic to withdraw Mythos 5 and Fable 5, a version of Mythos with extra cyberattack safeguards, from the market over national security concerns. The government partially eased the restrictions on Mythos on Friday. To address government concerns, Anthropic implemented a new safeguard to Fable that targets and blocks behavior described in a reported bypass of its protections that triggered the administration’s move, a person close to the company said. The person added that researchers from the Commerce Department’s Center for AI Standards and Innovation have validated both the prior and new safeguards. The lifting of the restrictions comes as the administration balances frontier-model security risks against pressure to keep pace with China’s AI development. OpenAI faced similar treatment last week, agreeing to a limited rollout of its GPT-5.6 model despite CEO Sam Altman’s objections. Industry figures hope a forthcoming executive order will replace ad hoc decisions with a predictable evaluation process. https://tinyurl.com/ycyjc4nb

OpenAI proposes 5% stake to Trump administration to ease Washington pressure: Report.

OpenAI has proposed handing the U.S. government a 5% stake in the company, the Financial Times reported Thursday, as the artificial intelligence startup seeks to defuse mounting political pressure in Washington. A 5% holding would be worth roughly US$42.6 billion, after the AI lab closed a record-breaking funding round in March at a post-money valuation of US$852 billion. OpenAI CEO Sam Altman argued that giving the public a financial interest in the company is the best way to share the upside of AI, the FT reported, citing two people familiar with the talks. Altman suggested a stake of that size in early discussions with the Trump administration, as part of a broader arrangement under which Washington would hold 5% of each of the leading U.S. AI developers via a government vehicle, according to the report. The proposed arrangement envisions other U.S. AI companies, such as Anthropic, Google and Meta, ceding similar stakes to the government through a sovereign wealth fund vehicle, the FT said. It is not clear whether any of these groups would agree to OpenAI’s proposal. https://tinyurl.com/m9mzu8zj

Strategy says it may sell up to US$1.25 billion of Bitcoin.

Michael Saylor’s Strategy on Monday said it may sell up to US$1.25 billion of bitcoin to bolster its cash holdings as part of its measures to turn around the stock, a major pivot from its long-held philosophy of buying and holding bitcoin. The company also plans to buy back up to US$1 billion of its common stock and up to US$1 billion of its preferred shares. It said it has about US$2.55 billion in cash, which will cover about 17.4 months of its dividend and interest expense payments. Crypto investors have been closely monitoring Strategy due to concerns that any major sales of its massive bitcoin holdings – now US$51 billion – could put further pressure on the broader crypto market. Strategy’s stock jumped 13% on Monday after the announcement. It has plunged 40% so far this year, sharper than 31% drop in the price of bitcoin. https://tinyurl.com/ycxck48y

Stripe, Visa, Mastercard join consortium launching new stablecoin.

Stripe, Visa, Mastercard, Coinbase and BlackRock are joining a partnership with over 140 firms to launch a new stablecoin, called Open USD, which will aim to challenge the dominance of Circle and Tether. Circle’s stock fell 14% on Tuesday after the announcement. The new stablecoin initiative, Open Standard, is led by Zach Abrams, the CEO of Stripe-owned stablecoin firm Bridge, who now also serves as the founding CEO of Open Standard. Unlike Circle and Tether, Open USD will share all of the earnings from its stablecoin reserves with distribution partners of the stablecoins, minus a “small management fee” to cover its operational costs. Tether doesn’t share its revenue with users, and Circle shares a portion of its revenue with partners like Coinbase, which take a cut before paying some rewards back to users. Google, BNY, Shopify, and several foreign banks have also signed up to use Open USD. The stablecoin will launch later this year on blockchains including Coinbase’s Base, Ethereum, Solana, and Tempo. https://tinyurl.com/2jn2kce2

Tencent get’s US$3 billion worth of memory chip supply from China’s CXMT.

China’s ChangXin Memory Technologies has signed a long-term deal to supply Tencent Holdings with more than 20 billion yuan (US$2.94 billion) of server DRAM chips, Reuters reported, citing three people familiar with the matter. DRAM, or dynamic random-access memory, helps servers quickly access data needed to run software and AI workloads. The agreement came as CXMT prepares for its initial public offering, which is expected to raise 29.5 billion yuan on Shanghai’s tech-focused Star Market. The deal comes as the AI boom strains global memory supply. Memory makers are shifting more capacity toward high-bandwidth memory, or HBM, a premium form of DRAM used in AI data centers, tightening supplies of standard server DRAM. The deal reflects a broader shift among Chinese tech giants toward local memory suppliers. Alibaba, ByteDance and Tencent have turned to CXMT and Yangtze Memory Technologies to ease the shortage after Samsung Electronics and SK Hynix prioritized HBM production for U.S. tech companies. https://tinyurl.com/35cvdnw2