The FT ran “US tech stocks hit by concerns over future of AI boom” after Altman’s AI bubble comments triggered moderate chaos. Nvidia reports earnings on Wednesday, which will test sentiment again. Meanwhile, it was confirmed that the US government under Trump will take a 10% stake in Intel. Once the embodiment of Moore’s Law, Intel has stumbled on smartphones, lithography, and AI. Revenues fell from $80B in 2021 to $50B in 2024, losing $20B last year. A state equity stake signals creeping nationalisation of strategic tech assets – the US is building a quasi-public model for critical compute, encouraging other nations to consider whether sovereign compute is optional or essential.
IPOs / Publics
Netskope (cloud data & cybersecurity) filed for Nasdaq IPO. H1 revenue $328.5M (+31% YoY). Still loss-making on GAAP basis but operating cashflow positive. Backers: Lightspeed, Iconiq, Accel.
Hong Kong overtook the US as Chinese startups’ listing venue – 46 IPOs raising $16.5B this year vs. just $741M in the US.
London IPOs slumped to a three-decade low – 88 companies delisted in 2024; Shein paused UK IPO.
Klarna (fintech) raised €1.4B from Santander, reigniting IPO speculation.
Apple in talks with Google about using Gemini to revamp Siri – potentially outsourcing core AI after years of failed internal efforts. Meanwhile, xAI filed an antitrust suit against Apple and OpenAI, alleging collusion via exclusive iPhone integrations. A ruling is expected this year to halt Google’s ~$20B annual payments to Apple for default search – it would be amusing if this were simply replaced by a $20B chatbot-default deal.
Meta froze hiring in its AI group after spending hundreds of millions to poach researchers. Division reorganised into four teams (research, superintelligence, products, infra). Considering closed-source models, signed a $10B six-year Google Cloud deal, and licensed Midjourney models. Meanwhile, OpenAI subpoenaed Meta for documents linked to Elon Musk’s unsolicited $97B OpenAI takeover bid.
Microsoft AI chief Mustafa Suleyman criticised “AI welfare” research – the study of whether advanced AI systems may one day deserve rights – as premature and dangerous. Rivals (Anthropic, Google, OpenAI) lean into this idea.
Google announced Android app store changes to comply with the EU’s Digital Markets Act (DMA) – easier off-app payments, non-store downloads, reduced fees. Also revealed energy-use data: Gemini query = 0.24Wh (≈9s of TV), a 33x efficiency gain in 12 months.
Big Dogs
Anthropic (AI model lab) raising $10B (Iconiq, TPG, Lightspeed, Spark, Menlo, Qatar & Singapore funds). Valuation to jump from $61.5B earlier this year to ~$100B+. Revenue run-rate tripled from $1B to $3B in five months.
OpenAI (AI model lab) hit $10B run-rate in June, then $1B July revenue alone (+20% in 2 months).
Databricks (data & AI infra) in talks to raise at $100B valuation (Thrive, a16z). Raised ~$20B since 2013. Recently acquired Tecton (ML data ops).
Armis Security (cyber for unmanaged/IoT/OT/critical infra) in late-stage talks at $5B valuation.
Field AI (general-purpose robotics intelligence) raised $405M across two oversubscribed rounds.
Stark (German unmanned weapons systems) raised $62M led by Sequoia at $500M valuation. Sequoia had also led an unannounced $20M Series A last year – its first European defence bet. Joined by 8VC, Thiel Capital, NATO Innovation Fund, In-Q-Tel, Project A.
Midas (consumer trading platform) raised $80M Series B (QED, IFC, Spark, Portage). Users 3.5M, up from 2M in 2024. Scalability outside its core market remains unclear.
Nuro (AI-first self-driving system) raised $203M Series E at $6B valuation (Uber, Nvidia, Fidelity, Tiger). Highlights rising self-driving chatter.
Lambda (GPU cloud infra) in talks with Nvidia to raise at >$4B valuation. IPO also possible.
Crusoe (AI cloud powered by flared gas) raising $1B at $10B valuation.
Hades (Munich deeptech drilling for geothermal/rare earths) raised €5.5M pre-seed (Visionaries Tomorrow, Project A). Fits European industrial resilience thesis.
Venture Capital Insights
Sam Altman on the AI funding boom: “When bubbles happen, smart people get overexcited about a kernel of truth. Are we in a phase where investors, as a whole, are overly excited about AI? Yes. Is AI the most important thing to happen in a very long time? Also, Yes.”
Venture Geopolitics
Intel – Trump confirmed the US will take a 10% equity stake, likely via CHIPS Act grants. Revenues fell from $80B in 2021 to $50B in 2024; it lost $20B last year. Shares rose 7%. Signals quasi-nationalisation of strategic compute.
US data centres – construction spend (excluding compute) has doubled retail and is set to surpass office spend. Infra boom shifts capital away from traditional sectors.
China compute gap – open-model labs like DeepSeek constrained without Nvidia chips; reliant on Huawei. Smuggling cases rising. Innovation bottlenecked.
Chip standoff – US considering Nvidia H20/B20A sales to China; Beijing may retaliate by banning foreign inference chips! Highlights fragility of global supply chain.
India AI push – launched $1.25B IndiaAI Mission: six foundation models, 18 applied projects. Sarvam AI training 70B-parameter Indian-language LLM on 4,096 H100s. Hyperscale data centres in India cost half US/EU (~$5M each). AI funding up 50% YoY. India shifting from “global back office” to AI power play.
UK dropped demand for Apple iCloud backdoor after talks with US – a win for privacy advocates.
UK Space Agency to be merged into DSIT to cut costs.
Strategic Sectors
Stablecoins
Goldman Sachs research on the stablecoin market: $271B, projected to grow 40% CAGR to 2027. USDC expected +$77B. Drivers: expanding crypto ecosystem, demand for USD abroad, regulatory clarity via GENIUS Act.
Payments TAM ~$240T annually. Today stablecoins mainly for crypto trading and EM dollar access.
Visa/Mastercard disruption limited near-term, but could help scale adoption.
Banks risk deposit flight, but stablecoins under GENIUS Act can’t pay interest and lack FDIC insurance.
Tokenisation of real-world assets ($25B today) still nascent, but stablecoins are natural settlement rails.
B2B payments remain slow and fraud-prone. Stablecoins present credible modernisation path.
Sifted notes the UK risks falling behind here – FCA regime expected in 2026 creates a cliff-edge: stablecoins start under FCA supervision, but if deemed systemic, shift to Bank of England oversight bans reserve-yield models. Scaling becomes unattractive and innovation stifled.
Crypto
Google bought 8% of TeraWulf (US HPC & Bitcoin mining infra).
S&P Dow Jones Indices in talks to launch tokenised benchmarks.
Trump family media group struck multibillion-dollar deal to buy Crypto.com’s CRO token.
AI
AI infra spend forecast to hit $520B by 2028, requiring $1T revenues at 50% margins. Hyperscalers made just $45B in 2023. Revenue curve may not meet GPU depreciation timelines.
LLM usage – 40% of firms subscribe, but >90% of employees use them. Bottom-up adoption dominates.
China tokens – usage up 300x in 18 months to 30T. China testing digital rails at population scale.
Google Gemini – inference energy use cut 33x, carbon emissions 44% in 12 months. Sustainability narrative improving.
Autonomous vehicles
Baidu Apollo Go logged 14M rides vs. Waymo 10M. Cars cost 1/6 of Waymo’s – China may scale autonomy faster.
And a titbit…
US households 75+ saw net worth +81% (2007–22), only age group to gain. Intergenerational wealth gaps continue to widen.