A late‑2024/2025 wave of large private financings and public‑market pushes has re‑accelerated across major fintechs: Upgrade closed a $165M equity round led by Neuberger Berman that values the company at about $7.3B and management signaled an IPO target within roughly 12–18 months; Revolut’s backers are nearing completion of roughly $3B of new funding/secondary activity that implies a ~$75B valuation; Mexico’s Plata raised $250M at a reported $3.1B valuation as it prepares to convert to a licensed bank (license approved Dec 2024); UK business fintech Tide completed a $120M round and attained unicorn status; and Klarna completed a U.S. IPO in September 2025 that raised roughly $1.37B (priced at $40/share, strong first‑day pops). (en.wikipedia.org)

The cluster of mega‑rounds and IPOs matters because it signals renewed investor appetite for scaled, revenue‑generating fintechs—especially those that can deploy data and AI to improve credit decisioning, pricing and customer service—while also exposing tensions: lofty private valuations increase pressure to show sustainable unit economics and regulatory readiness (notably for bank conversions and BNPL firms), and some incumbents (e.g., Klarna) are publicly reassessing AI‑first cost cuts after service and product tradeoffs emerged. (reuters.com)

The primary companies are Upgrade (Renaud Laplanche’s credit‑focused fintech), Revolut (Nik Storonsky and large private backers), Klarna (CEO Sebastian Siemiatkowski), Plata (Mexico‑based challenger transitioning to a full bank), and Tide (UK SMB fintech); key investors mentioned include Neuberger Berman (led Upgrade’s round), DST/Ribbit/Coatue and other late‑stage backers around Revolut, Kora/Moore/TelevisaUnivision tied to Plata’s round, and TPG/Apax around Tide. Major reportage comes from Bloomberg, Reuters, CNBC and Tech.eu. (en.wikipedia.org)

Several leading AI-infrastructure companies are raising large pre-IPO rounds or courting late-stage private capital as they scale inference and data-center offerings: Cerebras Systems closed a roughly $1.1 billion Series G at an $8.1 billion valuation and filed to pause/withdraw an imminent IPO while saying it still intends to go public later, using the private capital to scale U.S. manufacturing and data centers; FuriosaAI (Seoul) has accelerated bridge and pre-IPO financing (company announcement of a $125M Series C bridge and press reports of a planned ~$300M pre-IPO raise) to mass-produce its RNGD inference chips; and Lambda (a Nvidia‑backed AI cloud operator) has been reported in talks to raise several hundred million at a $4–$5 billion valuation ahead of a possible IPO, while striking large commercial deals (including reports of Nvidia leasing GPU capacity back from Lambda) as it readies for public markets. (cnbc.com)

This wave of pre-IPO funding highlights strong investor appetite for companies that provide inference-optimized chips and AI cloud infrastructure (a segment that supports LLM deployment and real‑time AI services), shifts some exit timing from IPOs to large private financings, and raises regulatory/geopolitical and supply‑chain questions (CFIUS scrutiny over foreign investor ties, scaling U.S. manufacturing, and strategic vendor relationships with Nvidia/TSMC). The financings and strategic deals materially affect future public-market comparables, valuation benchmarks for hardware/cloud players, and competitive dynamics between GPU incumbents and specialized inference-chip challengers. (reuters.com)

Cerebras Systems (CEO Andrew Feldman) and new/lead investors including Fidelity, 1789 Capital, Atreides and others; FuriosaAI (founder/CEO June Paik and Korean institutional backers including Korea Development Bank, Kakao Investment) as a regional inference‑chip challenger; Lambda (cloud operator with prior Nvidia backing and reported talks with new investors and bankers) and Nvidia as both investor/supplier/customer; and other ecosystem players such as TSMC (foundry), G42 (large customer/investor tied to Cerebras), and late-stage public/private investors and banks managing/preparing IPOs. (cnbc.com)

A renewed wave of financing and market activity is reshaping the commercial space sector: Firefly Aerospace completed a high‑profile U.S. IPO in early August 2025 (pricing 19.3M shares at $45 and raising about $868.3M, implying a roughly $6.3B valuation) even as it carries sizable losses and a growing backlog; at the same time major European primes (Airbus, Thales, Leonardo) are actively exploring a large consolidation (‘Project Bromo’) to build a home‑grown satellite/systems champion while the European Space Agency has signalled antitrust and sovereignty concerns; concurrently a new strand of “space AI infrastructure” is emerging — startups like Starcloud (partnering with NVIDIA) are proposing GPU‑powered orbital data centers and in‑orbit inference, and NVIDIA is shipping compact DGX Spark systems (and delivering units to SpaceX) to accelerate on‑orbit and edge AI efforts. (reuters.com)

This convergence — big public financings, potential industrial consolidation in Europe, and nascent AI‑in‑space infrastructure — matters because it reshapes capital flows, competitive dynamics and regulatory scrutiny in aerospace: sizable IPOs (and attendant liquidity) make more capital available for launches, lunar/value‑chain plays and AI payloads; European consolidation aims to create scale to rival U.S./Chinese vertically integrated incumbents but raises monopoly and procurement questions for ESA/Brussels; and putting GPUs in orbit (or deploying compact supercomputers at launch/launch sites) could unlock low‑latency, privacy‑sensitive inference for Earth observation and defense use cases while also creating new cost/energy and launch‑risk tradeoffs investors and insurers must price. (reuters.com)

The principal corporate and institutional actors are Firefly Aerospace (newly public, underwritten by banks including Goldman Sachs, J.P. Morgan, Jefferies and Wells Fargo), Airbus, Thales and Leonardo (exploring Project Bromo), the European Space Agency and EU competition authorities (raising monopoly/sovereignty concerns), SpaceX (launch and constellation leader and partner/customer for in‑space AI hardware), NVIDIA (providing GPUs and the new DGX Spark platform), startups like Starcloud (orbital data‑centre concept) and a constellation of investors, underwriters and defence contractors backing launches and lunar missions. (reuters.com)

Hugging Face has added a developer-focused workflow to its Spaces hosting platform by launching "Spaces Dev Mode" (announced May 21, 2024) which lets developers connect via VS Code or SSH into a running Space, override the Docker image, and restart the app as a subprocess for much faster iteration; separately, Hugging Face has long supported hosting models and datasets in Spaces using Streamlit (blog post Oct 5, 2021) and the current Hub docs clarify SDK and Docker usage (including Streamlit deployment via the Docker SDK). (huggingface.co)

For AI in general and finance specifically this matters because Dev Mode materially reduces development cycle time for model demos and prototypes (no full Docker rebuild on every change), enabling quicker experimentation with financial models and interactive dataset exploration while enterprise-grade controls (PRO/Team/Enterprise tiers) and Docker-based deployment requirements shape how firms can integrate Spaces into secure, auditable ML workflows; however, Dev Mode’s beta status, persistence behavior (changes are not auto-persisted), and Docker/SSH requirements introduce operational and governance considerations for regulated financial environments. (huggingface.co)

The primary organization is Hugging Face (Spaces product team and Hub/docs authors); key technologies and partners involved include Streamlit (used to build interactive UIs in Spaces), VS Code/SSH (the developer connection surface), and cloud/enterprise customers (PRO/Team/Enterprise subscribers and financial firms evaluating Spaces for model hosting). Community contributors and user feedback (Hugging Face forums) have already influenced the feature’s stabilization. (huggingface.co)

Between July and September 2025 Google Cloud has rolled out and showcased a suite of enterprise AI and infrastructure capabilities aimed at regulated industries — notably financial services: Google announced Compliance Manager (preview, Aug 20, 2025) to centralize cloud compliance and auditing, published Cloud HSM as an encryption key service for Google Workspace client‑side encryption (available Aug 18, 2025 in the U.S.) to give organizations hardware‑backed key control, and highlighted large financial‑services deployments such as Wells Fargo’s expanded use of Google Agentspace/agentic AI (announced Aug 5, 2025) to automate research, post‑trade inquiries, contract review and customer workflows; complementary partner use cases (Pluto7 + Agentspace for supply‑chain planning) and cross‑sector R&D (Google + NASA testing an AI clinical assistant for space missions) show the same agentic platform and Vertex AI models being applied across mission‑critical domains, while Google’s subsea networking investments (detailed Sep 2025) underline the infrastructure strategy that supports low‑latency, global data flows. (cloud.google.com)

This matters for finance because the combination of (1) governance tooling (Compliance Manager) and cloud HSM key control materially reduces compliance, audit and data‑sovereignty friction for regulated banks and trading firms; (2) agentic/Agentspace deployments (Wells Fargo and partners) promise operational scaling of research, post‑trade and client workflows while raising questions about model reliability and supervision; and (3) Google’s continued investments in backbone infrastructure (subsea cables, data centers) and Vertex/agent tooling create an integrated stack that can deliver both the latency and security assurances that large financial institutions require for production AI in trading, treasury, risk, and customer channels. (cloud.google.com)

Primary players are Google Cloud (product teams releasing Compliance Manager, Cloud HSM, Agentspace/Gemini Enterprise and Vertex AI), strategic enterprise customers such as Wells Fargo (enterprise agentic deployments), partners like Pluto7 (supply‑chain planning integrations), and public‑sector collaborators including NASA (CMO‑DA medical assistant trials) — with oversight/interest from regulators and compliance frameworks (FedRAMP, PCI‑DSS, DISA IL5, etc.) and infrastructure stakeholders tied to subsea cable projects (national governments, telecom operators). (cloud.google.com)

Demand for AI capabilities — from data centers and compute to talent, enterprise AI stacks, and cybersecurity — has meaningfully accelerated M&A and private‑equity (PE) activity in banking and software: global deal value surged in 2025 (driven by megadeals and strategic AI plays), large and megadeals are materially up year‑to‑date, and both strategic acquirers and PE sponsors are prioritizing software and data‑rich assets with AI upside (examples include PE processes for software platforms and multiple AI‑themed megadeals). (reuters.com)

This matters because AI is reshaping capital allocation and valuation benchmarks — boards and PE LPs are pushing exits, banks are rebuilding M&A capacity to capture advisory fees, and buyers are paying premiums for recurring‑revenue, data‑rich software businesses and AI infrastructure; the result is a faster deal cycle, higher aggregate megadeal value, and increased use of private credit and bespoke financing for AI projects. The trend raises integration, regulatory and valuation‑risk questions even as it sustains a strong advisory and underwriting market for banks and boutiques. (pwc.com)

Key players include investment banks (JPMorgan, Morgan Stanley, Goldman Sachs) expanding M&A coverage and advising PE exits; PE sponsors and buyers (TA Associates, GI Partners, Harvest Partners, Blackstone, TPG, KKR, Apollo) targeting software and AI‑enablement assets; consulting and Big‑Four deal teams (PwC) producing sector analysis and playbooks; and tech/cyber/AI infrastructure firms and hyperscalers (NVIDIA, OpenAI and large cloud/data‑center participants) that both drive demand and act as strategic acquirers or financing anchors. Specific corporate examples cited in reporting include JPMorgan’s leadership moves amid a deal rebound and MRI Software’s PE owners exploring a sale or IPO. (investing.com)

Over the last several months the AI-for-finance stack has tightened around three converging trends: (1) agentic workspaces and orchestration platforms that let teams build multi‑step LLM agents connected to live data and enterprise apps (AWS Quick Suite / Amazon Q and MCP tooling), (2) Retrieval‑Augmented Generation (RAG) and graph/embedding innovations that make retrieval both cheaper and more accurate (spaCy-related RAG/embedding work and graph‑RAG patterns), and (3) MLOps/developer tooling improvements that push model lifecycle operations into data platforms used by finance teams (BigQuery ML UI improvements, containerized DLC workflows via Amazon Q/MCP). These moves are concrete — AWS launched Quick Suite as an agentic AI workspace (Oct 16, 2025) and is integrating Amazon Q + MCP into developer flows, Google Cloud published Keeta’s Spanner‑backed 11.1M TPS stress test for a payments/blockchain workload (Aug 14, 2025), and practical docs/orchestration projects (OrKa) and spaCy teams published YAML‑first agent orchestration and graph/RAG/embedding work to make retrieval + pipelines reproducible and testable. (infoq.com)

Why it matters: finance workloads demand high throughput, auditable decision trails, low-latency access to private data, and strict governance — and the current wave stitches RAG, agent orchestration, and MLOps tooling to meet exactly those needs. Enterprises get (a) end‑to‑end agent workspaces that can execute cross‑system workflows while keeping data in‑tenant and auditable, (b) more efficient retrieval (graph‑RAG, multi‑hash embeddings) that reduces token costs and improves grounding for compliance‑sensitive outputs, and (c) tighter integration of model build/deploy flows into data platforms (BigQuery ML UI, containerized DLCs) so finance teams can iterate models with SQL/containers instead of bespoke infra. The result is faster productization of LLM features (e.g., automated reconciliation, real‑time cash‑flow Q&A, agentic workflows that act on ledgers) — but it also concentrates risk around governance, reproducibility, and vendor lock‑in. (machinelearningmastery.com)

Who’s involved: major cloud vendors (AWS pushing Quick Suite + Amazon Q and MCP and DLC integrations; Google Cloud enabling low‑latency analytics and operational DB scale with Spanner and BigQuery), developer tool / NLP platform teams (Explosion / spaCy working on graph‑RAG, multi‑hash embeddings and developer UIs), open‑source agent frameworks (OrKa and local agent orchestration projects), niche/high‑performance fintechs (Keeta Network demonstrating Spanner‑backed 11.1M TPS), and independent publications / tooling communities (MachineLearningMastery, InfoQ, community GitHub/Reddit posts) shaping practitioner adoption and debate. (infoq.com)

A surge of institutional activity around corporate crypto treasuries and stablecoin funding is unfolding: Tether is reported to be marketing a very large private raise—seeking roughly $15–$20 billion for about a 3% stake, which Bloomberg/Reuters say could imply a valuation as high as ~$500 billion—while separately corporate ‘treasury’ purchases of bitcoin have cooled sharply (a ~76% drop in monthly buying from early summer levels), and several listed companies are explicitly launching or expanding crypto-treasury strategies (e.g., Harrison Global’s bitcoin purchase term sheet; SharpLink Gaming’s $76.5M premium-priced direct offering to add ether to its treasury). (reuters.com)

The developments matter because they show two simultaneous forces reshaping crypto and finance: (1) very large private capital flows into market infrastructure and stablecoin issuers (Tether’s raise would be one of the largest private financings ever) with potential macro implications for liquidity and reserve flows (Tether already holds large Treasury positions and says it will expand into sectors like AI), and (2) a re‑rating of corporate treasury activity — some corporates/treasury firms rushed to accumulate crypto in 2025 but buying has since dropped, exposing concentration/liquidity risks and raising questions about whether corporate treasuries are stabilizers or amplifiers in crypto markets. These moves also intersect with AI finance narratives (Tether and others highlighting AI as an investment area), meaning private-capital, stablecoins, corporate treasuries and AI strategies are increasingly linked in macro and regulatory debates. (reuters.com)

Key named players: Tether (Paolo Ardoino / USAT initiative / Cantor Fitzgerald advising), potential investors reported in discussions such as SoftBank and Ark Investment Management, data/analytics providers like CryptoQuant (whose metrics underpin the Bloomberg buy-drop claim), listed corporate-treasury actors and treasury-focused issuers including SharpLink Gaming (SBET) and BitMine Immersion Technologies (largest public ether treasury), plus smaller public issuers/treasury buyers including Harrison Global (partnering with White Lion Capital), and broader market participants such as MicroStrategy/other public bitcoin treasuries and major media/data outlets (Reuters, Bloomberg, CoinDesk) reporting these trends. (reuters.com)

Agentic and conversational AI are moving from demos into production across finance and HR: cloud providers (AWS) have launched agentic workspaces that index enterprise data and execute multi-step workflows (Amazon Quick Suite, GA Oct 9, 2025), startups are deploying conversational hiring agents (Jack & Jill raised a $20M seed and reports ~50k users in London) and employee "digital twins" (Viven surfaced from stealth with a $35M seed) to automate sourcing, interviewing, onboarding and knowledge-sharing, and identity/security vendors and researchers are publishing agentic-identity and secure-delegation patterns to prevent fraud and preserve auditability in loan approvals and AML processes. (aws.amazon.com)

This matters because agentic agents + conversational interfaces can dramatically shorten finance and HR workflows (from reconciling invoices and account reconciliation to initial candidate screening and interview scheduling), but they also change the control plane: enterprises now face choices about governance, identity delegation, audit trails, data minimization, vendor lock‑in and headcount/role design — with potential efficiency gains (customer examples and projected time savings cited by AWS and early customers) balanced by regulatory, privacy and security risk that industry vendors and academics are explicitly addressing. (aboutamazon.com)

Major cloud/platform players (AWS — Quick Suite and broader agentic AI org), fast-growing startups (Jack & Jill in conversational recruiting; Viven for employee digital twins), specialist identity/agent-security vendors (Maverics / agentic-identity work cited in industry posts), traditional financial institutions and consultancies integrating agents into trading, syndication and KYC workflows (Goldman, JPMorgan, Citi referenced in coverage), and investors backing enterprise agent plays (Creandum led Jack & Jill; Khosla, Foundation Capital led Viven). (aws.amazon.com)

A cluster of high‑profile financing moves and IPO filings this week highlights how geography is shaping listing decisions: China’s autonomous‑driving AI supplier Momenta is weighing moving a planned U.S. IPO to Hong Kong with a possible 2026 Hong Kong debut as U.S.–China tensions and U.S. regulatory uncertainty bite; at the same time big private‑equity‑backed and AI/infrastructure plays are pursuing U.S. listings (Fermi, a newly formed data‑center REIT targeting Nasdaq and a ~$13.16 billion valuation) while major fintechs and digital banks pursue local or regional venues (Walmart‑backed PhonePe filed a confidential DRHP in India to raise about $1.5 billion; Nordic digital bank NOBA listed in Stockholm and jumped ~27% on debut). (reuters.com)

The pattern shows capital flows and strategic listing choices being driven by geopolitics (Chinese issuers shifting toward HK after tightened U.S. scrutiny), sector dynamics (AI and cloud/data‑center demand pushing infrastructure issuers to U.S. markets), and strong regional capital pools (India and Europe hosting large fintech and bank listings). The result: issuance is becoming more venue‑specific by industry and nationality, influencing valuations, investor composition, and timing of large private‑equity exits (e.g., Medline eyeing a ~$5 billion U.S. IPO to monetize buyout owners). This matters for global market share between exchanges (HK vs NY), regulatory oversight, and where tech/AI value is crystallized. (reuters.com)

Principal companies and organizations in the current wave include Momenta (backed by Toyota, Bosch, Mercedes‑Benz and Hyundai) and Hong Kong Exchanges & Clearing (HKEX); US and international banks and underwriters (UBS, Evercore, Cantor, Mizuho, Goldman Sachs, Morgan Stanley); Fermi (Rick Perry‑linked data‑center REIT) targeting Nasdaq; fintech/digital‑bank players such as PhonePe (Walmart, Tiger Global, Microsoft investors), NOBA (Nordic Capital, Sampo) and regulators/exchanges such as SEBI (India), Nasdaq (US) and Stockholm exchange. Private equity owners and sovereign investors tied to Medline (Blackstone, Carlyle, Hellman & Friedman) and large book‑runners are also key. (reuters.com)

Three related developments show corporate treasury, payments and banking are being reshaped by AI, tokenisation and crypto: Ant International (Ant Group’s overseas payments arm) is deploying AI time‑series FX models and blockchain tools to run large-scale treasury & cross‑border liquidity services — a Bloomberg profile (via Techmeme) reports the unit handled about $1.1 trillion of cash flow in 2024; Mexico’s fintech Plata, which received regulatory approval for a banking license in December 2024, has continued a rapid funding and bank‑buildout trajectory (reports in October 2025 say it raised $250M at a ~$3.1B valuation as it prepares to begin full bank operations); and smaller public companies such as Harrison Global announced (Oct 17, 2025) a term sheet with White Lion Capital to launch a Bitcoin Treasury Strategy, signalling corporates are experimenting with crypto allocations on their balance sheets. (techmeme.com)

These moves matter because they illustrate three converging trends: (1) large payments platforms are moving upstream into corporate cash management and FX/treasury via AI and blockchain (potentially displacing banks’ corporate treasury revenues); (2) fast‑scaling fintechs in emerging markets (example: Plata) are converting consumer‑facing scale into regulated bank operations, reshaping local corporate and retail deposit/lending markets; and (3) corporates experimenting with crypto treasury strategies (Harrison/White Lion) highlight growing institutional interest in digital assets for diversification — each raises regulatory, accounting and risk‑management implications for treasurers, regulators and investors. (fintechmagazine.com)

Ant International / Ant Group (Alipay+, WorldFirst, Embedded Finance teams and executives such as Ant International leadership) is the tech/payment incumbent pushing AI + blockchain into treasury; Plata (Mexico fintech, backed by investors including Kora and Moore Strategic Ventures, founders/management behind its consumer credit product) is the notable Latin American neo‑bank‑builder; Harrison Global and White Lion Capital are examples of public companies / digital‑asset managers executing bitcoin treasury deals; regulators and banks (e.g., Hong Kong/Singapore licensing regimes, local Mexican regulators) and investors (Kora, Moore Strategic Ventures, TelevisaUnivision in the recent round) are central stakeholders. (scmp.com)

Over 2024–2025 the enterprise productivity stack has seen rapid embedding of generative and agentic AI across tools (Google Workspace/Gemini, Squarespace Blueprint, bespoke agent platforms) and new startups building employee "digital twins" (Viven). Large financial firms are piloting or scaling agentic assistants (Wells Fargo announced an expanded Google Cloud Agentspace partnership on Aug 5, 2025) while SaaS design tools (Squarespace Blueprint, reviewed Oct 17, 2025) and startups (Viven’s public launch/seed round in mid‑October 2025) promise to automate routine work and preserve institutional knowledge — but organizations report uneven productivity gains and persistent implementation barriers (data quality, privacy/compliance, governance, skills, hallucinations and “workslop”).

This matters for finance because banks and finance teams stand to gain measured efficiency (faster document triage, contract search, customer-service automation, knowledge continuity), competitive differentiation, and lower operating costs — yet they also face heightened regulatory, audit and data‑governance risk. The mix of large platform pushes (Google Workspace AI & Agentspace) and point solutions (employee digital twins, design automators) creates pressure on CFOs/CISOs to define policies, remediate data gaps, retrain staff, and quantify ROI before wide rollout; early evidence shows adoption is widespread but measurable productivity upside is far from guaranteed.

Major platform players: Google (Workspace/Gemini/Agentspace), Google Cloud; enterprise adopters: Wells Fargo; design-tool vendor: Squarespace (Blueprint AI); digital‑twin startup: Viven (co‑founders Ashutosh Garg & Varun Kacholia) and investors (Khosla Ventures et al.); analyst and industry voices: Gartner, Thomson Reuters/industry surveys, Microsoft (shadow‑AI studies) and consulting firms that document implementation challenges.

Over the past few weeks the European space and tech-listing landscape has seen three linked developments: (1) the European Space Agency publicly warned about the risk of excessive market concentration as Airbus, Leonardo and Thales are reported to be discussing a roughly €10 billion joint venture to combine satellite businesses to better compete with rivals such as SpaceX’s Starlink (reported Sep 26, 2025). (reuters.com) (2) China’s autonomous‑driving AI firm Momenta is reported to be weighing a move of its planned IPO from New York to Hong Kong (discussed as a possible 2026 HK listing) after a prior Chinese regulator approval to list in the U.S. lapsed in June. (investing.com) (3) U.S. listing risk for smaller tech/networking firms is highlighted by Cambium Networks receiving a Nasdaq staff determination on Oct 10, 2025 that its ordinary shares are subject to delisting after failing to meet the $1.00 minimum bid price and related filing requirements. (sec.gov)

Taken together these items illustrate a cross‑market dynamic where industry consolidation (space/satcom), geopolitical/regulatory pressure on listing venues (Chinese AI/autonomy firms), and strict exchange listing rules (Nasdaq enforcement) are reshaping market structure and capital access for technology and AI‑adjacent companies. The ESA’s public antitrust signal can influence EU merger reviews and procurement policy for satellites, Momenta’s listing choice affects where AI hardware/software firms raise capital and regulatory exposure, and delisting actions like Cambium’s warn investors about execution and governance risks that can quickly remove liquidity from smaller tech names. (reuters.com)

Major industrials Airbus, Leonardo and Thales (the three European full‑system satellite builders) and the European Space Agency (Rolf Densing’s comments) are central to the consolidation thread; China’s Momenta (backers include Toyota, Bosch and reported participant investors such as Mercedes‑Benz/Hyundai in pre‑IPO rounds), Hong Kong Exchanges & Clearing (HKEX) and the China Securities Regulatory Commission (CSRC) are central to the listing‑venue shift; Nasdaq and Cambium Networks are central to the listing‑risk example. Competitors and market reference points include SpaceX/Starlink. (reuters.com)