February 21, 2026: NASDAQ Up (Finally), All Eyes On Nvidia Earnings

Last week, Dow Jones rose 0.25%, S&P 500 was up 1.1%, Nasdaq composite gained 1.5%. Software stocks had another rough day Friday, driven by fresh terminal value concerns driven by Anthropic’s new cybersecurity tools, which sent CrowdStrike and Okta tumbling 8-9%. Nvidia’s earning report next week will provide investors the next readthrough for the AI trade. ServiceNow CEO Bill McDermott is attempting to stem a 25% YTD stock slide by halting personal stock sales and initiating a US$5 billion buyback. It remains to be seen whether investors will interpret this move to support the thesis that AI is a tailwind, not a headwind for traditional per-seat SaaS models. AMD is mirroring the Nvidia playbook, backstopping a US$300 million loan for Crusoe to support chip sales. Meta’s move deepening ties with Nvidia via Vera Rubin chips signals another move towards co-designed proprietary models. OpenAI expects to burn US$17 billion this year alone, maintaining a 20% revenue share with Microsoft until 2032. Anthropic’s cloud payouts are projected to rise to US$6.4 billion next year. The battle for the “AI Edge” is moving to hardware. Apple is accelerating work on AI-powered glasses and pendants. OpenAI (via its Io acquisition) and Meta prepare to flood the market with smart speakers and watches by 2026/2027. Even the mobility sector is seeing CapEx shifts, with Uber earmarking US$100 million for robotaxi charging hubs to challenge Waymo. Macro tailwinds emerged as the US Supreme Court struck down Trump tariffs, providing a relief rally for Shopify and Amazon. On the bearish side, credit cracks could be forming in the “Nvidia-backed” ecosystem as CoreWeave shares stumbled 12% on reports that lenders are growing wary of below-investment-grade AI infrastructure debt. Blue Owl (a major investor in data center projects) gated redemptions from its non-traded business development company. In Canada, MDA Space is pivoting toward terrestrial defense with its 49North subsidiary, following Ottawa’s $6.6 billion industrial strategy.

MDA Space (MDA-TSX) makes defence-industry push with launch of 49North.

Space hardware manufacturer MDA Space has come down to Earth to formally declare itself a player in Canada’s growing defence industry. The emergence of 49North follows the launch of Canada’s new, $6.6 billion Defence Industrial Strategy this week. The Brampton-based company has launched 49North, a wholly owned subsidiary based in Ottawa exclusively focused on bolstering Canada’s terrestrial national defence priorities. 49North will develop Canadian defence solutions for land, air, maritime, and joint operations, including multi-domain C4ISR integration, according to MDA Space. https://tinyurl.com/3j6635rs

Chip startup Taalas raises US$169 million to help build AI chips to take on Nvidia.

Toronto-based chip startup Taalas said on Thursday it had raised US$169 million and has developed a chip capable of running artificial intelligence applications faster and more cheaply than conventional approaches. Taalas has raised a total of US$219 million from investors such as Quiet Capital, Fidelity and Pierre Lamond, a chip industry venture capitalist. Taalas’ announcement arrives weeks after Nvidia’s deal to license intellectual property from chip startup Groq for US$20 billion. https://tinyurl.com/58zvc8v3

Mecademic secures $21 million to bring its tiny industrial robots to the world.

Montréal-based Mecademic has secured new funding from a trio of government agencies to go big, while keeping its robots small. Mecademic’s robots are compact enough to fit in the palm of your hand. The company announced on Thursday that it raised a $21-million strategic funding round led by Investissement Québec (IQ), with participation from Export Development Canada (EDC) and the Business Development Bank of Canada (BDC).. https://tinyurl.com/yc662j49

Robinhood’s US$1 billion fund pitches pre-IPO stock as next craze.

Robinhood Markets Inc. is seeking to raise US$1 billion in an initial public offering of a closed-end fund designed to give US retail investors access to private companies, as the brokerage associated with meme stocks chases the latest frenzy. The Menlo Park, California-based firm is offering 5 million shares of Robinhood Ventures Fund I, and the fund itself will sell 35 million shares, priced at US$25 apiece, according to a statement Tuesday. Investors on Robinhood’s platform can place requests for IPO shares starting from Feb. 17, ahead of trading on Feb. 26, according to a marketing presentation. https://tinyurl.com/4h4cd5sb

Meta and Nvidia sign strategic partnership.

Meta Platforms and Nvidia on Tuesday announced a multi-year strategic partnership, in which Meta will use millions of Nvidia chips across its own data centers and through cloud providers. Terms of the deal were not disclosed. Meta, a longtime Nvidia customer, already puts Nvidia GPUs in its own data centers and rents them from cloud providers such as Oracle. But the new partnership, which involves multiple Nvidia technologies, including its forthcoming Vera Rubin chips and in-house networking technology, indicates that the companies will be working more closely together. The companies also said their engineering teams will work together to “co-design” Meta’s AI models. Meta has explored using rival AI chips. The Information reported in November that Meta was in talks with Google about spending billions of dollars to use TPUs in Meta’s data centers in 2027, as well as to rent Google chips from Google Cloud this year. It’s not clear whether the new deal will affect the TPU talks. A spokesperson for Meta said they had “nothing to share” and did not respond to a question about whether the talks were still ongoing. Nvidia said Meta would continue using Nvidia’s Grace central processing units and consider a “large-scale deployment” of the forthcoming Vera CPUs in 2027. (CPUs are traditional data center chips used to run non-AI workloads). https://tinyurl.com/e2886439

OpenAI pays 20% of total revenue to Microsoft Until 2032.

OpenAI can shift some of its revenue-sharing payments to Microsoft to later years under the terms of the pair’s recently renegotiated partnership last fall. This should reduce the impact of the payments on OpenAI’s cash flow. Last fall, OpenAI and Microsoft agreed to continue an arrangement in which OpenAI pays 20% of its revenue to Microsoft till 2032 in exchange for allowing OpenAI to work with other compute providers without giving Microsoft the option first, along with other conditions. As part of their original partnership agreement, Microsoft was entitled to 20% of the startup’s revenue through 2030. OpenAI can make some of those payments in years later than when it generates the revenue. OpenAI expected to lose US$9 billion in cash last year and US$17 billion in cash this year. The company doesn’t expect to become cash flow positive till 2030, when it expects to generate US$38 billion in cash. OpenAI projects paying more than US$13 billion in total in revenue share, mostly to Microsoft, this year and next year. https://tinyurl.com/yz2nre6p

Anthropic projects rising payouts to cloud providers: Financial Disclosures.

Anthropic paid about US$1.3 million in revenue share in 2024 to cloud providers for reselling Anthropic’s AI models. That figure is expected to rise to about US$360 million last year, US$1.9 billion this year and US$6.4 billion next year. That’s an increase from Anthropic’s past summer projections, when it expected to generate US$1.6 billion this year and about US$4.4 billion next year. By another measure, Anthropic shares about 50% of its gross profits to Amazon for any sales of Anthropic’s models Amazon sells. Google typically takes a cut between 20% and 30% of net revenue, after subtracting infrastructure costs, from resale of its partners’ software. It’s not clear what that portion is for Microsoft, which became Anthropic’s cloud provider in November. https://tinyurl.com/2wf6s389

AMD to backstop US$300 million Crusoe loan, following Nvidia playbook.

Chipmaker AMD is pursuing the same growth-driving strategy that has boosted sales at rival Nvidia: lending its financial support to upstart cloud providers buying its chips. AMD will effectively guarantee a US$300 million loan for data center and cloud startup Crusoe to purchase AMD’s AI chips and install them in a data center in Ohio. The loan from Goldman Sachs will be backed by the chips and related equipment, said people with knowledge of the deal. The deal, the first known example of a borrower using AMD chips as debt collateral, shows how the 56-year-old semiconductor company is turning to new tactics to boost sales and compete with Nvidia’s commanding AI chip business. AMD CEO Lisa Su has said she wants to reach tens of billions of dollars in annual sales of AI chips by next year and grab at least a tenth of the broader market. AMD backstopped the Crusoe deal by agreeing to rent its own chips from Crusoe if the startup can’t find customers that want to use them, such as AI developers, said the people with knowledge of the deal. AMD’s backstop helped Crusoe, which expects to burn up to US$4 billion a year into the next decade, secure a roughly 6% interest rate for the loan, a far lower rate than it otherwise would have gotten. https://tinyurl.com/ysa6nnab

Palo Alto Networks buys Koi.

Palo Alto Networks on Tuesday announced plans to acquire Israeli cybersecurity startup Koi, a deal meant to strengthen Palo Alto’s cybersecurity tools amid new threats posed by the advancement of agentic AI. Palo Alto didn’t give a price but Calcalist reported the company is paying US$400 million. Koi sells a tool meant to guard against AI agents containing malware. Palo Alto will integrate the tool into its security platforms. “AI agents and tools are the ultimate insiders. They have full access to your systems and data, but operate entirely outside the view of traditional security controls,” said Palo Alto’s chief technology officer, Lee Klarich, in a statement. “By acquiring Koi, we will be closing this gap and setting a new standard for endpoint security.” The news marks the latest in a series of acquisitions by the US$131 billion company, including a blockbuster, US$25 billion deal to purchase another Israeli startup, CyberArk, in July of 2025 as well as smaller-scale purchases of firms Protect AI and Talon. Palo Alto shares dropped following the announcement as investors await the company’s second quarter earnings results slated for release after the bell. The stock has fallen victim to a broad decline in software equities as Wall Street questions the impact of AI on the sector. https://tinyurl.com/25v9cepd

ServiceNow CEO stops selling stock in bid to calm AI-skittish investors.

ServiceNow CEO Bill McDermott and other senior executives from the software provider have agreed to cancel pre-scheduled stock sales, and McDermott plans to purchase US$3 million in ServiceNow stock later this month, according to a regulatory filing. The moves could reassure investors that ServiceNow is on solid ground despite a broad enterprise software market selloff that has seen its shares drop more than 25% since the start of the year. They come a few weeks after ServiceNow took another step to shore up investor confidence by announcing a plan to buy back US$5 billion in stock. McDermott is one of the first enterprise software CEOs to cancel pre-scheduled stock trades this year, raising the question of whether other industry counterparts might follow suit. Some investors fear that AI could decimate the traditional per-seat software licensing model and let companies slash software budgets by building their own features and applications. https://tinyurl.com/e2yrasuf

CoreWeave stock is tumbling amid fresh data-center financing concerns.

CoreWeave stock fell as much as 12% on Friday. The AI infrastructure stock has battled high volatility lately, though it’s up about 19% year to date. CoreWeave’s decline on Friday comes after a report from Business Insider that said Blue Owl Capital was unable to arrange debt financing for a data center it’s co-developing that is expected to be occupied by CoreWeave. Business Insider’s Dan Geiger reports that a debt arranger for major data center deals said that the project, located in Pennsylvania, failed to attract interest because lenders are reluctant to add exposure to AI firms with below-investment-grade credit. CoreWeave debt is rated B+ by S&P Global Ratings, several notches below investment grade. Despite being backed by Nvidia, CoreWeave has faced skepticism from finance pros who have wondered about the sustainability of the business model of renting out compute to AI customers. The company’s high debt-to-equity ratio and its AI-dependent business model have been red flags for some investors, particularly as AI enthusiasm has waned somewhat in the last six months. Famed short-seller Jim Chanos has also raised concerns about CoreWeave, highlighting problems he sees with its financials and profitability, and warning investors against owning data center stocks. Blue Owl, meanwhile, was already in the spotlight this week for reportedly halting redepmtions on one of its private credit funds offered to retail investors. https://tinyurl.com/3fwesr9s

Anthropic’s new security tool sends cyber stocks reeling.

Anthropic’s unveiling of Claude Code Security, a new cybersecurity tool, sent cybersecurity stocks plunging—the latest part of the software sector to be hit by anxieties about AI disruption. Anthropic’s new tool detects vulnerabilities in software to help security teams defend against hackers using AI. Claude Code Security reads code the way a human security researcher would to “find and fix issues that traditional tools often miss,” Anthropic said. Once vulnerabilities are detected, it suggests software “patches” to fix them. Security software stocks CrowdStrike and Okta tumbled roughly 8% and 9%, respectively. Cloudflare shares lost over 7%, Zscaler shed 5%, and SailPoint sank almost 9%. Up until Friday, cybersecurity stocks had been less affected by the software stock selloff: An exchange traded fund tracking cybersecurity stocks, the First Trust NASDAQ Cybersecurity ETF, has lost about 11% over the past six months, while overall, software stocks in the broader S&P 500 have dropped 24%. A tech lead for Cloudflare in a post on X dismissed the idea that Claude Code Security is a threat to existing firms, however, snubbing “investors who apparently think all forms of ‘security’ are fungible.” https://tinyurl.com/2ffkht78

Amazon, Shopify shares rise after supreme court strikes down Trump tariffs.

Shares of e-commerce stocks like Amazon and Shopify rose Friday after the Supreme Court struck down tariffs imposed by the Trump administration. Shopify shares were up more than 5% midday, and Amazon rose around 2%. Investors had been worried that new tariffs would result in higher prices for consumers, potentially hurting sales, or weigh on online sellers’ margins. Roughly two thirds of the goods sold on Amazon come from outside sellers, many based in China or other parts of Asia. Both Amazon and Shopify had previously downplayed the impact of the tariffs on their businesses to investors last year. In August, Shopify said it hadn’t seen any impact on shopper behavior, though some merchants raised their prices or added additional shopper fees to help cover the costs of the tariffs. https://tinyurl.com/2z4n97tz

Apple ramps up work on glasses, pendant, and camera AirPods for AI era.

Apple Inc. is accelerating development of three new wearable devices as part of a shift toward artificial intelligence-powered hardware, a category also being pursued by OpenAI and Meta Platforms Inc. The company is ramping up work on smart glasses, a pendant that can be pinned to a shirt or worn as a necklace, and AirPods with expanded AI capabilities, according to people with knowledge of the plans. All three devices are being built around the Siri digital assistant, which will rely on visual context to carry out actions. Each of the products, which will be linked to Apple’s iPhone, depends on a camera system with varying capabilities, said the people, who asked not to be identified because the plans haven’t been announced. In an all-hands meeting with employees earlier this month, Chief Executive Officer Tim Cook hinted that the company would be pushing hard into AI devices, saying Apple is working on new “categories of products” that are enabled by artificial intelligence. “We’re extremely excited about that.” Cook added that the company was investing in new technology. “The world is changing fast,” he said. While iPhone sales remain robust, Apple is playing catch-up in AI. Revamping Siri has been a key challenge: Upgrades to the voice assistant have been plagued by development snags, delaying their rollout. https://tinyurl.com/4ctxpwxs

OpenAI plans AI-powered speaker, priced at US$200 to US$300.

OpenAI is developing an AI-powered smart speaker with a camera that will be able to take in information about its users and their surroundings. The speaker, which will likely cost between US$200 and US$300, will also allow people to buy things by identifying them with a facial recognition feature similar to Apple’s Face ID. The speaker is one of a number of potential designs, including smart glasses and a smart lamp, that the company has come up with since acquiring device startup Io last May for close to US$6.5 billion. Io was started by OpenAI CEO Sam Altman and former Apple design chief Jony Ive, who have been discussing building an AI-powered personal device since at least September 2023. OpenAI, which has more than 200 staffers working on the family of AI-powered devices, doesn’t expect to ship any device to customers until at least February 2027 at the earliest, according to a court filing. The device it will ship, though, will differ from today’s most popular devices like phones because it will be able to observe users through video and nudge them toward actions it believes will help them achieve their goals. In developing a device, OpenAI will be plunging into an increasingly crowded hardware market. Apple is reportedly planning a host of AI devices, including an AI wearable pin and AirPods with enhanced sensors, while Meta Platforms and Google are either already selling or planning to sell smart glasses with AI capabilities. https://tinyurl.com/y4yjrk7w

Meta to challenge Apple with its first smartwatch — and it’s reportedly launching this year.

Meta is planning to release the brand’s first smartwatch sometime in 2026, according to an insider report. Allegedly code-named Malibu 2, the smartwatch will reportedly offer health-tracking tools like those already found on the best smartwatches, along with a built-in Meta AI assistant. Successfully launching a Meta smartwatch would potentially put the tech giant in direct competition with other big wearable makers, including Apple, Garmin, Google, Samsung, and Oura. Apparently, Meta first began work on a wrist-based wearable back in 2021, but pulled the plug roughly a year later due to cuts in its hardware department. Insiders say that the previously-shelved project featured an onboard camera, with three variations of the watch to choose from. Can we expect the same from Project Malibu? It’s difficult to say. However, reports of cameras coming to smartwatches are far from new. Meta is obviously no stranger to the wearables market, with multiple AR glasses models currently on offer, including the popular Meta Ray-Ban Displays. You could even go as far as to say that Meta makes some of the best smart glasses in 2026. https://tinyurl.com/3kyeuh9y

Uber will spend US$100 million to build robotaxi charging stations.

Uber Technologies Inc. is planning to spend more than US$100 million to build fast-charging, autonomous-vehicle charging stations in the US, the latest move to establish itself as a critical player in the robotaxi industry. The company said in a statement on Wednesday that it will focus on building the new high-capacity charging hubs starting in the San Francisco Bay Area, Los Angeles and Dallas — markets where it also plans to launch public robotaxi services with technology partners to compete with Alphabet Inc.’s Waymo. The amount will cover site development costs, equipment, grid connection and associated capital expenditure for developing the charging infrastructure, a spokesperson said. https://tinyurl.com/337dsees

SpaceX joins pentagon drone technology competition.

SpaceX and its AI unit xAI are entering a Pentagon competition to develop autonomous drone technology, part of a Defense Department effort to advance voice-controlled unmanned systems. Bloomberg reported that SpaceX is one of a small number of companies picked to compete in a challenge launched in January to create “advanced swarming technology,” which can coordinate and control fleets of drones with minimal human programming. The challenge will award a US$100 million prize to the winner. The company is being overseen by the Pentagon’s Defense Innovation Unit and Defense Autonomous Warfare Group. SpaceX is a longtime defense contractor, providing rocket launches for Pentagon satellites. Military drones would represent a new direction for the company, however. In the past, SpaceX CEO Elon Musk has expressed concerns about making new AI-powered weapons. https://tinyurl.com/2s3z6c26

ByteDance launches new LLM with better visual understanding.

ByteDance has released its new generation of large language models, Doubao Seed 2.0, as the Chinese tech giant tries to compete at the highest level with U.S. rivals in all types of AI models, from LLMs to video. ByteDance, which owns TikTok, said Seed 2.0’s strong understanding of visual data enables it to analyze a mixture of complex documents, tables, charts, graphics and video content. The company also said Seed 2.0 Pro, the most powerful version of its new-generation models, has achieved top-tier level performance in benchmarks for mathematical reasoning. In addition to Seed 2.0, ByteDance over the past week has unveiled a new video generation model, Seedance 2.0, and a new image generation model, Seedream 5.0. The Seedance 2.0 video model went viral on U.S. social media platforms last week because of its strong capabilities and controversial videos featuring celebrities and copyrighted materials. Unlike many other Chinese AI developers that open-source their models, ByteDance focuses on proprietary models and offers them mainly through its own apps and cloud platforms. https://tinyurl.com/nvrsvvuk

Alibaba launches New LLM as China’s AI battle heats up.

Alibaba Group on Monday unveiled Qwen3.5, the new generation of its large language models, adding to the recent flood of new AI model releases from Chinese companies ahead of the Lunar New Year, China’s biggest holiday. Alibaba, a major global competitor in open-source AI models, said Qwen3.5-Plus, the first version of Qwen3.5, delivers strong performance in reasoning, coding, AI agent capabilities and ability to process multiple types of data including images, audio and video. Compared to Alibaba’s previous models, the new model can handle tasks faster while reducing computational costs, according to the company. Alibaba is hoping that the release of the new-generation model will boost the popularity of its Qwen AI chatbot app, which is now powered by Qwen3.5-Plus. Alibaba has integrated the app with its e-commerce, online travel, mapping and payment platforms in an effort to build an AI agent that helps consumers with tasks like shopping and travel booking. Alibaba is competing fiercely with ByteDance as the two Chinese tech titans fight to promote their chatbot apps during the holidays. ByteDance, whose Doubao chatbot is China’s most popular AI app by users, has also just launched its new-generation LLM, Seed 2.0, along with new AI image and video models. https://tinyurl.com/393rhxap

Perplexity ends Ads experiment.

Perplexity is no longer offering ads, an executive told The Financial Times. The AI search startup is pulling back from this line of business as rival OpenAI started showing its users ads in ChatGPT conversations earlier this month. The company said it was worried ads would undermine users’ trust in their platform, with an executive saying “the challenge with ads is that a user would just start doubting everything.” Anthropic also questioned the value of ads in chatbots with a Super Bowl ad that painted chatbot ads as distracting, ending with the tagline “Ads are coming to AI. But not to Claude.” https://tinyurl.com/m7h3pvtt