Tech Executive and Analyst. Founder of FourWeekMBA, The Business Engineer & The AI Supercycle

London, England
Joined September 2015
This completely changed how I see the AI industry. It blew my mind. I could hardly believe it!

5
7
85
1,190,693
Google's TPUs are coming to Oracle's cloud from 2027, the first hyperscaler outside Google to offer them, per a joint release. Oracle also announced Cerebras inference for OCI. In under a minute: $GOOGL @Google @Oracle #Semiconductors #AIInfrastructure
62
"Genie 4 is our first real-time world model to generate audio natively." - Google DeepMind Worlds now carry ambient sound and dialogue, and Gemini 3.8 Flash plans and checks each step as the scene renders. Research preview for trusted testers. #AI
63
“Our co-founders have each pledged to dedicate 80% of their personal equity in Anthropic to philanthropic causes” - Anthropic's S-1 It also puts 846.8M shares, 5% of the fully diluted count, in a fund Anthropic owns: 3.5x the biggest co-founder stake (our math). #AI
31
Three companies. Three IPO filings. $3.5 trillion in combined target valuation. The AI economy's first real test. 🧵
1
28
OpenAI: $852B. Targeting Q4 2026. Advantage: ChatGPT's consumer scale, frontier model leadership. Risk: nonprofit restructuring, Microsoft building competing models, $6.6B in employee stock already traded on secondary.
1
7
The real question: can public markets tell the difference between usage growth and durable profits? AI inference costs money per token. Every API call, every agent action = GPU time. Unlike software, marginal cost doesn't approach zero. Full analysis: fourweekmba.com/ai-ipo-race-…
12
Neocloud Argentum AI is in talks for $5B of private credit for AI chips, Reuters reports, citing two people familiar. The debt is meant for chips leased to customers without investment-grade ratings, the same week SpaceX secured $40B. #Semiconductors #PrivateCredit
33
Apple is putting an assistant built on its own models and OpenAI into Pages, Keynote and Numbers. It is also opening the three apps to outside AI agents through MCP. Both come with its $12.99-a-month Creator Studio in November. #AIAgents #AI
34
Anduril and Qualcomm unveiled BlueChip: a laptop designed and assembled in the US, with more than 90% of prototype component cost from US and allied suppliers. Palmer Luckey estimates it will cost about 20% more than typical high-end PCs. Orders from 2027. $QCOM #AI
69
American Tower's CoreSite data centres grew revenue 17.4% to $313M in Q3, while its US & Canada tower segment fell 3.3%, the company said today. It raised its 2026 property revenue outlook by $55M at the midpoint, citing Data Centers outperformance.
23
AMD says it will invest $1B in Meta's stock offering, a day after Nvidia named $3B. Its Helios racks are installed and being tested at Meta. Two chip suppliers have now named $4B of the up to $5B they indicated. #Semiconductors #AI
40
Corning's Optical Communications sales rose 44% to $2.38B in Q3, and its Gen AI sales doubled, the company said today. CEO Wendell Weeks says that is "outpacing GPU unit growth." The rest of Corning grew more slowly or shrank. #Semiconductors #AI
36
Hubbell's Electrical Solutions sales: $559M to $845M in Q3 (+22% organic). CEO Gerben Bakker says the growth was "led by strong datacenter markets". Hubbell raised its 2026 outlook, mainly on that segment. #VentureCapital #AIInfrastructure
32
General Compute: $60M valuation in May, $800M now, per TechCrunch. The inference-chip neocloud raised a $100M Series A off a $100M-ARR customer deal, in the week Musk predicted 200+ neoclouds face "guppy-level death." #VentureCapital #AIInfrastructure
33
Nace AI says coding agents using its new Drex for Search took 62% fewer steps and used 56% fewer input tokens across 529 business-file questions, in its own tests. Accuracy rose from 63.1% to 67.1%. Two of ten setups it tested scored lower. #AIAgents #AI
27
UPS: ~$1.8B of program benefits in nine months from automation, sort consolidation and a smaller workforce. It raised its 2026 revenue outlook to ~$92.6B. CEO Carol Tomé: "the most automated and efficient network in our history." #VentureCapital #AI
28