Tomasz Tunguz

SandHill has indexed 837 posts from Tomasz Tunguz, most recently in 17 Nov 2026.

Where they publish

Tomasz Tunguz ↗

What they write about

AI (Horizontal) · AI Infrastructure · Enterprise Software · Developer Tools · Semiconductors · Venture Capital & Startups

Recent posts
  • The Lakehouse for AI↗ 17 Nov 2026
    LanceDB, a multimodal lakehouse for AI training and inference, raised $57.5M Series B on 10x YoY revenue growth. Major AI labs (Microsoft, Adobe, NVIDIA, Runway, Exa) are consolidating fragmented data infrastructure into Lance-based systems handling 100B+ rows and petabyte-scale unstructured data.
  • Strategic Decision or Structural Advantage?↗ 16 Nov 2026
    Databricks telemetry from 10,000+ enterprises reveals Anthropic and OpenAI carving out distinct market segments: Anthropic dominates frontier models (94% SOTA share), while OpenAI captures the low-end (81%) via aggressive Luna pricing cuts. The question: deliberate market segmentation or structural cost advantage?
  • No Shirt, No Shoes, No Service for Bots↗ 13 Nov 2026
    Amazon, United, and Dell ban AI shopping agents to protect high-margin ancillary revenue (ads, fees, warranties) that now exceeds their base product profits; Shopify welcomes agents because its business model relies on transaction-processing cuts regardless of buyer type.
  • When the Central Bank of AI Changes Chairs↗ 12 Nov 2026
    Jensen Huang's resignation as Nvidia CEO threatens the company's unprecedented role as simultaneous chip supplier, venture investor, and financier across 200+ AI startups—potentially slowing AI buildouts, weakening open-source funding, and accelerating customer adoption of custom silicon.
  • Who Captures the Most Margin in AI?↗ 11 Nov 2026
    The AI value chain distributes gross profit across five layers—power, data centers, GPU operators, inference vendors, and applications—with GPUs capturing 28% of margin. Rising compute costs and vertical integration will reshape competition and determine which layers maintain pricing power.
  • We Are All Junior Talent Now↗ 10 Nov 2026
    As AI agents become capable of complex reasoning and problem-solving, the traditional junior/senior distinction in talent becomes obsolete; companies must rethink how humans and AI collaborate, repositioning workers as managers of AI agents rather than task executors, with emotional intelligence and integrity as the remaining human differentiator.
  • When Does the US Become a Net Importer of Tokens?↗ 9 Nov 2026
    US AI token production will lag demand by 2028–2030, forcing America to become a net importer of compute—reshaping startup infrastructure, capital deployment, and regulatory priorities as supply constraints push pricing and policy to the fore.
  • A $90 Key into Your Company's Secrets↗ 6 Nov 2026
    The cost of AI-powered corporate network breaches has plummeted 99% in five months, falling from $5-10K to $90, while defenders must outspend attackers across speed, scale, and breadth—reshaping the offensive-defensive economics of enterprise security.
  • Watts in Store for 2027 : The AI Ceiling↗ 4 Nov 2026
    AI compute capacity will hit a ceiling around 80GW by 2027, creating structural GPU price pressure even as new capacity comes online. This scarcity will reshape AI economics, favoring inference pricing, algorithmic innovation, software efficiency, and off-the-shelf solutions over custom builds.
  • The Financial Engineering Essential for AI in 2027↗ 3 Nov 2026
    GPU costs for AI have doubled in six months, forcing frontier labs to pursue "financial engineering"—locking enterprise customers into multi-year committed contracts at fixed rates to match the duration of their underlying compute leases and preserve unit economics.
  • The Four Questions of AI↗ 2 Nov 2026
    The AI stack divides into four interdependent layers—applications, harnesses, models, and data centers—each facing a critical question about moats, ownership, scaling economics, and infrastructure constraints. These layers form a coupled system where moves in one cascade unpredictably through the others.
  • Crossing the Trillion Dollar Chasm↗ 29 Oct 2026
    Amazon, Google, and SpaceX are racing to build $1 trillion AI compute engines, but must solve the 'trillion dollar chasm': monetizing new AI products (agentic shopping, longer search queries, satellite compute) fast enough to fund the $200B+ annual capex before legacy cash cows (Search ads, Retail ads, classical workloads) erode.
  • Beyond Automation at OpenAI↗ 28 Oct 2026
    OpenAI's Chief People Officer discusses how AI capability doubling every six months reframes HR strategy around amplifying human judgment rather than pure efficiency, and how leaders must develop emotional intelligence as AI commoditizes technical intelligence.
  • Anthropic S-1 Analysis: How 7 Key Metrics Stack Up↗ 26 Oct 2026
    Anthropic's S-1 filing reveals 11.9–14-fold annual revenue growth, 900% net revenue retention, and a path to operating profitability—but offset by $637B in compute commitments against $4.6B in 2025 revenue, raising valuation and execution risk questions.
  • The Third Product-Market Fit in AI↗ 26 Oct 2026
    AI's first two product-market fits were coding and customer support; security is emerging as the third, with AI automating SOC triage, threat analysis, penetration testing, and vulnerability ranking across a $175B market entering a high-token-consumption phase.
  • How Large is the AI Circle?↗ 23 Oct 2026
    AI infrastructure investment is reaching revenue payback twice as fast as the telecom internet build-out did, reaching 10% revenue-to-capex ratios in four years versus ten. The risk is circularity—whether venture-backed AI companies spending on compute are just recycling capital—but venture-backed firms now represent a sixth of US employment, suggesting a genuinely large addressable market beyond startup-to-startup spending.
  • Closing Eight Figure Deals in Two Weeks↗ 22 Oct 2026
    Bardia Shahali, who scaled sales at Fireworks AI and now leads Granola, reveals how AI infrastructure and application sales differ: infrastructure deals close in weeks at $10M+ scale with intense competition, while apps move faster with higher rep productivity. Fast-changing product roadmaps force quarterly (not annual) compensation resets and risk under-staffing outbound efforts.
  • The GPU Revenue Ladder↗ 21 Oct 2026
    GPU inference economics show healthy unit economics at scale—SpaceX achieves $2.44/card-hour realized revenue with <1-year paybacks—but face pressure from token price deflation outpacing hardware costs and emerging on-device inference reducing cloud demand.
  • What if They're All Clouds?↗ 18 Oct 2026
    Anthropic's 1,374% YoY revenue growth, largest data center commitments ($518B), and pricing power in AI models position it as a cloud infrastructure player comparable to AWS, Azure, and GCP—suggesting it could reach Azure's scale in ~3 years at current growth rates.
  • Price War in the Chip Aisle↗ 13 Oct 2026
    Custom silicon from Google, Microsoft, and OpenAI is driving a 3.8x cost spread among frontier AI models despite near-identical intelligence levels, signaling that chip efficiency advantages directly translate to pricing power and margin sustainability as token costs deflate.
  • The Theory 2026 AI Survey Results↗ 13 Oct 2026
    Theory VC's 2026 AI survey of 455 senior technical builders reveals security has become the top barrier to AI adoption (43%), nearly half of teams now route queries across models for cost optimization, and Cloudflare dominates agent sandbox hosting at 62%—signaling rapid enterprise sophistication.
  • Sales in the Age of AI↗ 12 Oct 2026
    Bardia Shahali, VP of Sales at Granola, discusses how AI-era sales teams operate differently—smaller headcount, faster ramps, larger deals, and organic product spread—reshaping hiring, compensation, and GTM structure versus traditional SaaS.
  • A Change in AI Strategy↗ 7 Oct 2026
    AI model commoditization—with prices down 41% and token usage up 50%—demands a strategic pivot: labs should partner and resell competing models, control distribution through user interfaces, and capture routing data to train cheaper future models.
  • How to Automate Inbound↗ 6 Oct 2026
    Vercel's COO reveals how they built an inbound sales agent that now handles qualification and outreach with minimal human oversight—starting with a 125-line prompt, evolving to 1,000 lines, then distilling to 14 deterministic rules plus model-driven judgment, running the entire function for ~$1k/year in inference.
  • Inference Is the Most Important Market in Software↗ 5 Oct 2026
    AI inference spending will surpass databases by 2027 to become software's largest market, fundamentally reshaping unit economics: usage consumption will dwarf platform fees, gross margins will compress below SaaS norms, and infrastructure costs will dominate COGS.
  • Do We Grow Software or Do We Design It?↗ 2 Oct 2026
    Software development splits into two paradigms: organic 'vibe-coding' with iterative exploration, and deterministic 'design' via state machines and formal verification—each suited to different problems in the AI era.
  • The Fastest Growing Market in Tech isn't AI↗ 30 Sept 2026
    Tokenized real-world assets (RWA)—stocks and commodities trading on blockchains—are growing at 53% monthly, outpacing AI startups, with major financial institutions and blockchain infrastructure like Allium powering institutional adoption.
  • Segmentation Drives Market Share Wins in AI↗ 29 Sept 2026
    Anthropic and OpenAI are competing on business model innovation—segmentation and price discrimination—as much as technical capability. Anthropic's enterprise metering doubled revenue in a quarter; OpenAI's 80% price cut on its cheapest tier keeps them competitive, with both approaching $100B revenue run rates by year-end.
  • How GPU Prices Can Double While AI Gets Cheaper↗ 28 Sept 2026
    GPU costs have doubled while AI inference prices plummet—a seeming paradox resolved by rapidly improving model efficiency that's outpacing infrastructure inflation, though the balance remains precarious as capital becomes more expensive and demand explodes.
  • Thinking in Systems, Shipping in Loops↗ 24 Sept 2026
    AI has eliminated hand-coding as economically viable; software engineering now means designing systems and feedback loops that enable AI agents to write and validate code at scale. Donella Meadows's framework of resilience, self-organization, and hierarchy defines how these loops should be built.

All writers · all sources · by topic