NVIDIA Predicts $1TRN in Revenue: Everything You Need to Know From GTC & Anduril Lands $20B Contract

TL;DR

NVIDIA's GTC keynote forecasted $1 trillion in cumulative revenue over two years while Anduril secured a $20B Army contract, signaling massive sustained AI capex.

Key Points

• 1.NVIDIA's $1 trillion revenue figure is less dramatic than it sounds. Adding analyst forecasts of ~$500B for 2026 and ~$400B for 2027, then applying a 'salesman's round up,' yields roughly $1 trillion — meaning the stock moved less than 1% because analysts had already priced it in.
• 2.NVIDIA grew from $20B to $215B in revenue in ~5 years, and is forecasting ~60% growth again this year. The real headline is that unprecedented capex levels — potentially $1.2 trillion industry-wide if NVIDIA hits $600B — are expected to continue for four to five more years.
• 3.Jensen Huang launched NeMo Cloak (NVIDIA's open-source LLM) and announced data centers in space. The open-source push is strategic: more open models burn more tokens, driving inference demand and GPU revenue.
• 4.The panelists see NVIDIA as firing on all cylinders while rivals like OpenAI show signs of strain. OpenAI is reportedly concentrating on enterprise and cutting side projects — described as a classic startup 'code red' moment.
• 5.Atlassian announced 1,600 layoffs and Meta is reportedly cutting 20% of its 79,000-person workforce (~16,000 people). The hosts categorize these into five types: overhiring correction, growth slowdown, AI efficiency gains, capex reallocation (paying Jensen instead of people), and talent reshuffling for AI-fluent staff.
• 6.Meta's layoffs are specifically driven by capex reallocation — the company must fund massive GPU spending, and with free cash flow nearly zero after honest capex accounting, it literally cannot afford both Nvidia and headcount.
• 7.The new must-hire profile is the 'agentic deployment expert' — someone who has brought at least one commercial AI tool into their organization in the past 30 days. Jason Calacanis argues 70%+ of current management teams at even the best startups fail this standard.
• 8.You do not need to be technical to win with AI agents in 2026. Anyone who has successfully deployed enterprise software (Salesforce, HubSpot, Outreach) has the skills to deploy agentic tools — the only non-intuitive step is training the agent.
• 9.Anduril's $20 billion, 10-year Army contract (5-year base + 5-year option) consolidates 120+ separate procurement actions into one enterprise deal. The primary product is Lattice, a software connectivity layer that integrates disparate physical hardware systems in real time — critical for drone defense where humans have only seconds to respond.
• 10.The Anduril deal signals it has been elevated to a prime supplier for the Army's AI-enabled systems layer. The panelists note Anduril has only 4–5 customers, so $20B per customer is necessary to build a massive company, and the contract essentially locks in Lattice as the default connectivity standard.
• 11.Seed fund economics are breaking down as power-law thinking inflates pre-seed valuations to $60M+ post-money. To get 100x with dilution from a $60M post valuation requires a ~$14–20B exit — fewer than 50 public tech companies globally have that market cap — making $50–100M seed funds potentially the worst-performing of this vintage.