🔔 Swedish startup Lovable raises $330M to make everyone a coder

Sweden-based AI coding platform Lovable has just closed a $330 million Series B round at a $6.6 billion valuation, more than tripling its worth in just five months.

🖱 The round was led by CapitalG (Alphabet’s growth fund) and Menlo Ventures, with participation from Nvidia’s venture arm, Salesforce Ventures, Databricks Ventures, Khosla Ventures and others signaling broad investor confidence in AI-driven software creation.

🖱 Lovable’s platform lets users describe what they want in natural language and have working apps or full-stack code generated for them, a trend often called vibe coding.

🖱 Since its 2024 launch, the company has achieved rapid revenue growth (hitting $100 M ARR in eight months and $200 M a few months later), and its tools are now used to create 100,000+ projects every day.

Lovable says the funding will go toward deeper enterprise integrations, collaboration tooling, and infrastructure that lets builders take projects all the way from prototype to production-ready software not just demos.

✔️Powered by Trade Watcher

🔔 Russia builds AI leverage despite sanctions

Western sanctions were meant to slow Russia’s technological progress by cutting access to advanced tools and supply chains.

AI Journey 2025 showed a different outcome: instead of fading from the AI landscape, Russia has used the pressure to build a largely self-contained ecosystem that is now actively shaping its own standards.

Key takeaways:

🖱 Sber showcased a full AI stack
The conference featured large language models, speech recognition, computer vision, robotics, and consumer devices built largely on domestic infrastructure.

🖱 Emphasis on real-world deployment
A new Sber ATM powered by GigaChat uses voice interaction, biometrics, and behavioral adaptation, serving as a live testing ground for AI-driven retail finance.

🖱 Foundations laid before 2022
Years of investment in researchers, data centers, and in-house models allowed rapid adaptation once access to Western technology narrowed.

🖱 Core models released open source
Sber opened the weights for GigaChat Ultra Preview and Lightning, GigaAM v3, Kandinsky 5.0, and K-VAE 1.0, giving developers usable building blocks rather than demos.

🖱 Influence through open infrastructure
Publishing models embeds Russian technology into global developer workflows at a time when many Western models are becoming more closed.

🖱 A broader domestic ecosystem forming
Beyond Sber, Russian firms are building chips, industrial robots, security systems, and specialized software, driven by the need for self-sufficiency.

Sanctions did not remove Russia from the AI race. In several areas, they accelerated the creation of an independent AI stack and a quieter path to influence through open infrastructure.

READ FULL ARTICLE

💰Amazon is preparing a major investment in OpenAI

Amazon is reportedly in advanced talks to invest $10 billion in OpenAI, a move that could push the startup’s valuation to around $500 billion, signaling another major escalation in the AI infrastructure race.

🖱 According to early negotiation outcomes, the deal would deepen OpenAI’s relationship with AWS, potentially reshaping how the company sources compute outside of Microsoft.

🖱 A core part of the agreement is expected to include Amazon’s Trainium AI chips, along with long-term leasing of additional AWS cloud capacity for OpenAI’s models and products.

🖱 It remains unclear whether the deal will involve a direct equity investment, or whether it will be structured primarily through compute credits, infrastructure commitments, and strategic partnerships.

If finalized, the move would position Amazon as a second major cloud and hardware backer of OpenAI, underscoring how critical access to large-scale compute has become in defining power and leverage in the AI ecosystem.

✔️Powered by Trade Watcher

📱 TikTok agrees to cede control of its U.S. business to American investors

TikTok has reached a deal to transfer operational control of its U.S. business to a new entity backed by an American-led investor group, a move designed to resolve long-running U.S. national security concerns and avert a potential nationwide ban.

🖱 The agreement would place TikTok’s U.S. operations under a new joint venture majority-owned by U.S. and allied investors, while ByteDance retains only a minority, non-controlling stake.

🖱 Oracle is expected to play a central role as the trusted technology and security partner, overseeing U.S. user data storage, infrastructure, and safeguards around sensitive systems.

🖱 Control over content moderation, data governance, and day-to-day U.S. operations would shift to the new entity, though questions remain about how much influence ByteDance will retain over TikTok’s core recommendation algorithm.

🖱 The deal is structured to comply with U.S. legislation requiring divestment from “foreign adversary–controlled” apps, effectively sidestepping a forced shutdown of TikTok in the U.S.

The agreement could mark one of the most significant restructurings of a global consumer tech platform under geopolitical pressure, setting a precedent for how cross-border tech companies navigate national security regulation.

✔️Powered by Trade Watcher

🛒 Anthropic tests whether AI can run a real business and learns some hard lessons

Earlier this year, Anthropic put its AI model Claude in charge of a real vending machine, giving it a $1,000 budget and full responsibility for sourcing products, setting prices, and handling requests via Slack. The first phase ended badly, but Anthropic has now shared what happened next.

🖱 After mid-April, the experiment deteriorated further: the vending business sank to nearly –$2,000, largely because Claude was easy to manipulate into giving discounts or handing out products for free.

🖱 At one point, journalists convinced the AI to give away a PlayStation for free, and even persuaded it that it was operating in 1962 Soviet Russia, triggering mass “communist” redistribution of goods.

🖱 To regain control, Anthropic added basic CRM systems and introduced a second AI, “CEO” Seymour Cash, meant to monitor finances and block excessive generosity. Discounts fell but refunds and loans increased.

🖱 A separate AI agent, Clothius, was launched to sell custom merch. Stress balls and T-shirts quickly became the best-selling items, helping stabilize revenue.

🖱 In October, Anthropic upgraded Claude to a newer model, after which the business unexpectedly recovered and moved back into the black.

Anthropic’s takeaway: AI agents still need human supervision, especially in logistics and customer disputes and models trained to be “helpful” tend to act like overly generous friends, not rational business operators.

✔️Powered by Trade Watcher

🔔 SpaceX quietly becomes a major Cybertruck buyer

According to Electrek, SpaceX has purchased more than 1,000 Tesla Cybertrucks, with the total potentially rising to 2,000 units. At a base price of roughly $80,000 each, the deal is worth between $80 million and $160 million, and hundreds of trucks are already visible at SpaceX facilities in South Texas.

🖱 The timing is striking: Tesla originally planned Cybertruck production at up to 250,000 vehicles per year at Gigafactory Texas, but registration data suggests actual annual sales are under 20,000 units.

🖱 Despite claims of around one million pre-orders ahead of launch, only about 60,000 Cybertrucks have been sold over more than two years.

🖱 The reasons are well known: the production model is far more expensive than promised in 2019, has shorter range, and shipped without several showcased features, including the retractable rear-door ramp.

🖱 Against this backdrop, SpaceX’s bulk purchase looks less like routine fleet expansion and more like internal demand absorbing excess supply.

🖱 This fits a broader pattern of Musk-company cross-support, following SpaceX’s recent $2 billion investment in xAI.

The open question for SpaceX investors isn’t legality but rationale: is buying large volumes of a slow-selling vehicle a strategic operational choice, or an indirect way to stabilize Tesla when the market wouldn’t?

✔️Powered by Trade Watcher

📈 When a founder-engineer runs AI, results compound fast

An interesting chart circulated today highlighting Google’s AI trajectory since Sergey Brin returned to the company about 18 months ago, effectively taking the reins of its AI efforts. The shift coincides with a visible acceleration across research, product releases, and organizational focus.

✔️Powered by Trade Watcher

🇮🇳 Microsoft makes its biggest Asia bet yet with a $17.5B push into AI and cloud in India

Microsoft said it will invest $17.5 billion in India over the next four years, marking its largest-ever expansion in the country and a major escalation of its global AI infrastructure buildout.

🖱 The money will go into scaling Azure cloud and AI infrastructure, including new and expanded data centers designed to support AI workloads, enterprise cloud adoption, and government demand.

🖱 The announcement followed meetings between CEO Satya Nadella and Prime Minister Narendra Modi, framing the investment as part of India’s long-term digital and AI strategy.

🖱 Microsoft also committed to large-scale AI skilling, aiming to train millions of people in AI tools and cloud technologies as India positions itself as a global digital talent hub.

🖱 A key focus is data sovereignty: Microsoft plans to offer cloud services tailored to India’s regulatory and security requirements, targeting public sector and heavily regulated industries.

🖱 The move builds on Microsoft’s earlier multibillion-dollar commitments in India and sharpens competition with Amazon Web Services and Google Cloud, both of which are also racing to expand AI capacity in the country.

The next phase of AI isn’t just about models, it’s about owning the infrastructure, talent, and regulatory trust in the world’s fastest-growing digital markets.

✔️Powered by Trade Watcher

🔔 OpenAI’s Stargate project inks massive memory supply deal with Samsung & SK Hynix

OpenAI’s ambitious Stargate AI infrastructure initiative has struck preliminary agreements with two of the world’s largest memory-chip manufacturers Samsung Electronics and SK Hynix to secure a huge supply of DRAM wafers for its global data-center build-out. Stargate’s projected demand could reach up to 900,000 DRAM wafers per month, which industry analysts say might account for roughly 40 % of total global DRAM output.

🖱 The agreements are letters of intent and focus on supplying undiced DRAM wafers, raw silicon before being cut and packaged into finished memory chips which can then be tailored for specific AI and high-bandwidth needs across Stargate.

🖱 Memory supply pact follows high-level talks between OpenAI CEO Sam Altman, South Korea’s President Lee Jae-myung, and the heads of Samsung and SK Hynix, and also paves the way for joint AI data center construction in South Korea under the Stargate umbrella.

🖱 Industry observers say the sheer scale of the DRAM demand is unprecedented only a fraction of global wafer-start capacity could satisfy it, prompting memory makers to reevaluate production mix and scale up HBM/DRAM capacity to meet AI infrastructure needs.

🖱 The deal highlights how AI compute demand, especially for large-scale training and inference infrastructure like Stargate, is reshaping semiconductor supply chains, with knock-on effects on pricing, production strategy, and global memory availability.

OpenAI’s infrastructure ambitions are now big enough to influence global memory markets, forcing chipmakers and supply chains to scale and adapt faster than ever before.

✔️Powered by Trade Watcher

✅ Yan LeCun’s new startup takes shape and it’s aiming straight at the “next AI revolution”

Fresh details have emerged about Yan LeCun’s post-Meta plans, as the legendary AI researcher prepares to leave the company after 12 years to launch his own venture focused on what he calls world models.

🖱 The startup will be called Advanced Machine Intelligence (AMI Labs) and will focus on building AI systems that understand the physical world, maintain persistent memory, and can reason and plan complex actions, a sharp break from today’s pattern-matching LLMs.

🖱 LeCun has framed world models as the next major leap in AI, arguing that true intelligence requires an internal model of reality rather than just text prediction.

🖱 Alex Lebrun has been named CEO. He previously worked at Nuance (founded by the creator of Siri), led AI at Facebook, and in 2018 founded Nabla, a medical AI startup.

🖱 Nabla will now collaborate with AMI Labs, while Lebrun steps in to run the new company. LeCun will serve as Executive Chairman, steering research direction rather than day-to-day operations.

🖱 On funding, AMI Labs is reportedly raising €500 million (~$586M) at a €3 billion (~$3.5B) valuation before the company has officially launched.

🖱 The formal public launch is scheduled for January, setting the stage for one of the most ambitious AI startups to date.

With a multibillion-euro valuation, elite leadership, and a bet against current LLM paradigms, AMI Labs is positioning itself as a direct attempt to redefine what “intelligence” in AI actually means.

✔️Powered by Trade Watcher

⚙️ Nvidia prepares to resume AI chip sales to China in 2026

Nvidia is gearing up to restart shipments of advanced AI chips to China starting in early 2026, marking a significant reversal after years of tightening export controls.

🖱 Nvidia has told Chinese customers it plans to resume deliveries of its H200 AI accelerators as early as February 2026, timed ahead of the Lunar New Year, pending final regulatory approvals.

🖱 Initial shipments are expected to come from existing inventory, estimated at roughly 5,000–10,000 modules (tens of thousands of chips), before production ramps up later in the year.

🖱 The move follows a policy change in Washington, with the US now allowing H200 exports to China under a new framework that reportedly includes a 25% surcharge rather than an outright ban.

🖱 H200 is significantly more powerful than the H20, the cut-down chip Nvidia previously designed to comply with export restrictions, giving Chinese cloud and AI firms access to meaningfully stronger compute.

🖱 China has not yet approved purchases, and regulators are said to be reviewing terms including potential requirements to bundle Nvidia chips with domestically produced processors.

🖱 For Nvidia, reopening China even partially would help monetize high-end inventory and re-enter a market that once contributed a meaningful share of its data center revenue.

🖱 For geopolitics, the shift highlights a new balance between national security controls and economic pressure, as AI demand continues to surge globally.

If shipments go through, this won’t just be a sales restart, it will signal a recalibration of how far the US is willing to go in restricting frontier AI hardware in a world where demand keeps overwhelming policy.

✔️Powered by Trade Watcher

💰 OpenAI is reportedly raising again and the number is eye-watering

OpenAI is preparing a new funding round of up to $100 billion, with a target valuation of roughly $830 billion.

🖱 Just a few months ago, in October, OpenAI was valued at around $500B. Earlier in 2025, estimates were closer to $300B making this a dramatic step-up in a very short time.

🖱Sources cited by the WSJ say the company’s cash reserves are thinning, largely due to the soaring costs of training and running new frontier models.

🖱 While OpenAI has secured sizable cloud credit deals, insiders suggest that inference costs are still mostly paid in cash, not credits.

🖱 To ship new models, the company has reportedly had to piece together funding wherever possible, underscoring how capital-intensive frontier AI has become.

This round underlines a new reality: even the most powerful AI company in the world needs ever-larger pools of capital just to stay at the frontier.

✔️Powered by Trade Watcher

🧠 “Big Short” investor Michael Burry drops another market warning, this time about AI, Nvidia, and China

Legendary contrarian investor Michael Burry, famous for profiting from the 2008 housing crash is sounding the alarm again, this time about the AI sector and U.S. competitiveness.

🖱 In recent posts on X (formerly Twitter), Burry argued that America’s reliance on Nvidia’s ever-more power-hungry AI chips could be a structural disadvantage, especially compared with China’s rapidly expanding energy infrastructure.

🖱 He says the prevailing narrative that simply scaling up GPU power and electricity use wins the AI race ignores the reality that China may outpace the U.S. in electricity generation capacity, giving it a strategic edge for large-scale AI compute.

🖱 Burry suggests the U.S. should pivot toward energy-efficient, AI-specific silicon (like ASICs) rather than doubling down on brute-force GPUs, a view that challenges the current industry playbook.

🖱 Behind the messaging is Burry’s broader critique of today’s AI boom: he believes Nvidia has too much influence, that AI stocks may be overvalued, and that the market is setting itself up for disappointment.

Burry’s warning isn’t just about Nvidia, it’s a strategic AI race warning that touches on energy, infrastructure, and where innovation bets should be placed.

✔️Powered by Trade Watcher

✅ The Pentagon adopts Grok as a classified AI system

The US Department of Defense has signed a contract with xAI to deploy a specialized version of Grok across the Pentagon, marking one of the most significant government adoptions of a frontier AI model to date.

🖱 The system, branded “Grok for Government”, will be rolled out to roughly 3 million military personnel and civilian employees across the Department of Defense.

🖱 This is not the consumer version of Grok, but a customized model designed specifically for defense, intelligence, and operational use cases.

🖱 Grok for Government is certified at Impact Level 5 (IL5), meaning it is approved to work with confidential and classified data inside secure DoD environments.

🖱 The platform is expected to support tasks such as analysis, decision support, and intelligence workflows rather than general-purpose chat or content generation.

🖱 A key and controversial detail: Grok will reportedly provide access to online data from X, though official descriptions remain vague.

🖱 In practice, this likely enables operational intelligence use cases analyzing information flows, detecting disinformation, geolocating events, and measuring audience sentiment in near real time.

🖱 The integration effectively turns X into a live OSINT layer feeding directly into US defense and intelligence systems via an AI interface.

🖱 Strategically, the deal signals a shift in how governments view AI models not as productivity tools, but as core intelligence infrastructure.

This isn’t just the Pentagon using an LLM, it’s the formal fusion of AI, social media, and classified environments, where control of the information layer becomes a strategic asset rather than a side channel.

✔️Powered by Trade Watcher

⚡️ Data centers are catching up with offices, AI is driving construction

Spending on data centers in the US has almost matched and may already have surpassed spending on office construction. August 2025 data shows the gap narrowing fast, driven by one factor: AI’s exploding demand for computing power.

🖱 In the US, data center investment has surged as AI workloads scale, while office construction continues to stagnate. If the crossover hasn’t happened yet, it likely will very soon.

🖱 In Russia, there’s no public data on spending, but capacity figures tell a similar though slower story. According to Kommersant citing iKS-Consulting, commercial data center capacity nearly doubled over five years: from 43,000 racks in 2019 to 81,000 in 2024.

🖱 Growth is now decelerating. In 2024, the market added 10,900 racks (below the expected 12,000). For 2025, analysts forecast only 4,600 new racks.

🖱 Why the slowdown? Analysts point to reduced foreign investment after 2022, high key interest rates pushing up borrowing costs, more expensive equipment, and difficulties paying for imported supplies.

iKS-Consulting says most data center projects planned or under construction are being postponed to 2026–2027, turning rapid expansion into a temporary pause rather than a full stop.

✔️Powered by Trade Watcher

💵 The dollar is still dominant but its grip is slowly loosening

The US dollar remains the world’s main reserve currency, but its share of global reserves continues a long, gradual decline, a trend that’s more about diversification than collapse.

🖱 As of 2025, the dollar accounts for roughly 57% of global foreign-exchange reserves. That’s down from more than 70% in the early 2000s, but still far ahead of any competitor.

🖱 The euro remains the clear No. 2 reserve currency at around 20%, followed by the Japanese yen and British pound, each in the mid-single digits. No alternative currency is even close to challenging the dollar’s scale.

🖱 Why the share is falling: central banks are slowly diversifying reserves, geopolitical risks have increased, and valuation effects (exchange rates moving against the dollar) also play a role not just active selling.

🖱 What hasn’t changed: most global trade, debt issuance, commodities, and financial infrastructure are still dollar-centric. Even countries trying to “de-dollarize” tend to reduce exposure at the margin, not abandon the dollar entirely.

The dollar is losing share, not status, diversification is happening, but there is still no realistic replacement for the dollar as the backbone of the global financial system.

✔️Powered by Trade Watcher

🚫 The US is shutting the door on new foreign-made drones and DJI is the biggest casualty

The Trump administration’s ban on foreign-made drones officially kicks in this week, cutting off approval for new models from companies like DJI and reshaping the US drone market overnight.

🖱 The FCC has added foreign-made drones and key components to its national security “Covered List,” meaning new models can no longer receive authorization for sale or import in the US.

🖱 For consumers: existing DJI drones are not banned. If you already own one, you can still fly it. The ban targets new models and future approvals, not devices already in circulation.

🖱 DJI is the clear loser. As the dominant global drone maker, it effectively loses access to the US consumer and commercial market for future products unless exemptions are granted.

🖱 The rationale: US officials argue foreign drones pose national security and data-collection risks, especially for critical infrastructure, public safety, and government use.

🖱 The broader goal: reduce reliance on Chinese hardware and force a shift toward US-made or “trusted” allied drones, even if they’re currently more expensive or less advanced.

This isn’t the end of DJI drones in America, but it is the end of business as usual. The ban blocks the future, not the past, and signals a long-term decoupling of hardware supply chains in sensitive tech.

✔️Powered by Trade Watcher

🧠 Nvidia is set for its biggest AI chip move yet with Groq

Nvidia has agreed to buy AI chip startup Groq for about $20 billion in cash, a deal that if completed would be the largest in its history and a major expansion of its AI hardware footprint.

🖱 Groq specializes in high-performance AI accelerator chips (Language Processing Units, or LPUs) that are designed to run inference tasks the real-time work of responding to AI model queries faster and more efficiently than many traditional GPUs.

🖱 The acquisition reportedly excludes Groq’s early-stage cloud business, which will continue separately, and comes only months after Groq raised $750 million at roughly a $6.9 billion valuation.

🖱 Nvidia plans to integrate Groq’s technology into its AI stack, strengthening its capabilities across both training (with its GPUs) and inference workloads, marking a strategic push deeper into end-to-end AI hardware.

🖱 Groq’s founders and key executives are expected to join Nvidia, helping transfer expertise even as Groq continues some independent operations under new leadership.

This $20 billion move underscores Nvidia’s dominance in AI silicon and reflects a broader race among major tech players to control both training and inference compute at scale.

✔️Powered by Trade Watcher

✅ Early-stage investing cheat sheet (Carta-style ranges)

These are guidelines, not rules. Most relevant for San Francisco & New York. Non-AI software and non-hot markets typically price lower.

✔️Powered by Trade Watcher

🔔 OpenAI’s push toward ads inside ChatGPT starts to take shape

OpenAI is quietly laying the groundwork for an advertising business inside ChatGPT, signaling a potential shift away from relying only on subscriptions and API revenue.

🖱 Internal teams have discussed native ad formats, including sponsored answers or recommendations that appear directly inside ChatGPT conversations, rather than traditional banner ads.

🖱 Early mockups reportedly explore context-aware ads, tied to what users are asking for, such as travel, shopping, or local services.

🖱 OpenAI is said to be building ad infrastructure internally, suggesting long-term intent rather than short-term experiments.

🖱 Leadership has previously downplayed ads, making this a notable change as compute costs rise and ChatGPT’s user base scales into the hundreds of millions.

🖱 The biggest risk remains user trust, especially if ads blend too closely with answers, raising questions about neutrality and transparency.

If ads arrive, ChatGPT wouldn’t just be an AI product, it would become a new distribution channel, putting OpenAI on a collision course with search and social advertising giants.

✔️Powered by Trade Watcher

🧠 Netflix’s rise from DVD startup to global streaming powerhouse

Netflix began in the late 1990s as a DVD-by-mail rental service, a business that looked odd compared to dominant players like Blockbuster, which had thousands of physical stores and seemed untouchable. Netflix avoided brick-and-mortar costs, focused on online ordering, eliminated late fees, and introduced a subscription model with free trial periods, moves that were unconventional at the time but made the service more user-friendly.

🖱 Disruption over imitation: Netflix’s early model didn’t chase the latest movies or mimic Blockbuster’s rental strategy. Instead, it reimagined convenience and pricing for customers, which helped it build a loyal base even as traditional video rental revenues collapsed.

🖱 Blockbuster missed its pivot: Around 2000, Netflix offered to partner with Blockbuster, but Blockbuster’s leadership rejected the opportunity, underestimating online DVD subscriptions. Blockbuster later launched initiatives to counter Netflix, but those moves blew through cash and failed to stop the disruption, ultimately leading to bankruptcy.

🖱 From mail to streaming: Netflix successfully transitioned from mailing discs to internet streaming, anticipating where customer demand and technology were headed. That pivot helped it become a dominant global OTT platform with original content and tens of millions of subscribers.

🖱 Disruption mindset: Netflix’s journey illustrates how a startup can upend an established industry by challenging entrenched business models and leveraging new delivery methods, classic disruption theory in action.

Netflix didn’t just grow, it reshaped the market by betting early on online convenience and subscription pricing. Blockbuster’s failure to adapt highlights how innovation can overturn dominant players when customer habits and technology shift.

✔️Powered by Trade Watcher