Good morning. It's Thursday, May 21, and we're covering NVIDIA’s high-stakes earnings report, OpenAI’s growing influence over U.S. AI policy, and the IPO plans intensifying the battle between the biggest AI labs.
Know someone who'd love this? Forward it their way. First time reading? Sign up here to keep up with the future of tech.
OpenAI is preparing to confidentially file its IPO prospectus as soon as Friday — two days after a federal court rejected Elon Musk's challenge to its for-profit conversion — with Goldman Sachs and Morgan Stanley advising on what could be one of the largest debuts in history. The ChatGPT maker is targeting a September listing at an $850B+ private valuation, on the back of $30B in annualized revenue, 900M weekly active users, and 50M+ paying consumer subscribers. The filing sets up a three-way AI IPO race with SpaceX (publicly filing the same Wednesday) and Anthropic (eyeing a 2026 listing at ~$950B).
Alibaba says its new Zhenwu M890 AI chip delivers triple the performance of its prior model, even though analysts say it still trails leading Western processors. The company unveiled the chip alongside plans for its upcoming Qwen3.7-Max model, reinforcing its push into China’s domestic AI hardware market as NVIDIA faces export restrictions. Alibaba says it has shipped 560,000 Zhenwu units to 400 customers, with the M890 featuring 144 GB GPU memory and 800 GB-per-second bandwidth.
NVIDIA is opening its first Singapore research hub as the city-state doubles down on embodied AI, despite its relatively small domestic market. The lab will focus on robotics and AI infrastructure efficiency, supporting Singapore’s broader push to become a regional center for real-world AI deployment. Singapore also plans to launch a robotics testbed later this year with companies including Grab, DHL, and Certis, while new trials will explore AI-powered delivery, cleaning, and security patrol systems.
Rather than cede CPUs to AWS, Intel, and AMD as analysts had worried, Jensen Huang told Wednesday's earnings call that NVIDIA's new Vera CPU — purpose-built for agentic AI — opens "a brand new $200 billion TAM" the company has never addressed before. Huang's pitch: while GPUs handle the reasoning, agents themselves run mostly on CPUs to execute tasks, and Vera is optimized for token throughput rather than the multi-core parallelism of classic cloud CPUs. NVIDIA already booked $20B in standalone Vera sales this year on top of another record quarter ($81.6B revenue, $91B forecast next).
POWERED BY ELEVENLABS
Build Real-Time Conversational AI Agents
ElevenAgents by ElevenLabs enables you to easily build intelligent, real-time conversational AI agents. By using your own knowledge base, SOPs and policies, ElevenAgents creates agents that sound like your brand in over 70 different languages at low latency and in minutes.
Ship voice agents for sales, support, and operations. Configure, deploy, and monitor natural, human-sounding agents that scale your team without scaling headcount.
INTERVIEW
Inside Google’s AI Strategy Shift
Matt sits down with Google’s Sundar Pichai to unpack AI agents, open-source models, cyber warfare, compute limits, and what’s next for the internet.
MARKETS
NVIDIA Beats With Record $81.6B Quarter
Rather than show signs of plateau, Nvidia posted $81.6B in Q1 FY27 revenue (up 85% YoY) on non-GAAP EPS of $1.87, beating Wall Street estimates of ~$79B and $1.76 and extending its sequential-growth streak to 14 straight quarters. Data Center alone delivered $75B (up 92% YoY) as hyperscalers absorbed Blackwell 300 capacity at scale and sovereign customers joined the buyer list.
Guidance for Q2 came in at $91B (±2%) excluding any China Data Center compute, alongside expected non-GAAP gross margins of ~75%. On the call, Huang positioned Vera Rubin as the next leg up, called the new Groq LPX chips a narrow niche, and pointed to inference as Nvidia's fastest-growing share gain — consistent with his GTC claim that Nvidia will sell $1T in Blackwell and Vera Rubin chips across 2026–2027. → Read the full article here.
READER POLL
If You Were a Top AI Researcher…
Andrej Karpathy just joined Anthropic — if you were a top AI researcher today, which lab would you choose? |
Sharing what everyone said tomorrow.
POLICY
OpenAI Pushes State-by-State Strategy to Shape U.S. AI Rules
OpenAI is shifting its AI policy strategy from Washington to state capitals as federal lawmakers remain deadlocked on national AI legislation. In an interview published May 20, 2026, OpenAI chief global affairs officer Chris Lehane described the company’s approach as “reverse federalism” — encouraging major states like California, New York, and potentially Illinois to adopt similar AI safety laws that could effectively become a national standard
The company-backed frameworks focus on transparency, reporting requirements, and third-party audits while avoiding tougher liability rules favored by some AI safety advocates. The effort reflects a broader battle over who will shape AI regulation in the U.S., as tech companies, safety groups, and political leaders increasingly treat state governments as the frontline for AI policy. → Read the full article here.
POWERED BY
The AI Report
The AI Report is a daily AI newsletter read by 400,000+ founders, directors, and senior operators at companies like Google, Microsoft, and OpenAI.
Five minutes a day on what actually matters in AI — practical, data-led, no filler. If you're getting value from Forward Future, this one belongs on your list too.
NEWS
What Else is Happening
Figma Launches Native Design Agent: Figma added an AI agent directly to the canvas, letting designers edit, iterate, and explore ideas without leaving files.
Google DeepMind Expands Singapore AI Push: New partnerships target healthcare, education, and science, with AI projects projected to add $3.3B.
Anthropic Nears First Profitable Quarter: The AI lab projects $559M Q2 operating profit on $10.9B+ revenue.
Intuit Cuts 3,000 Jobs: The TurboTax maker will lay off 17% of staff as CEO Sasan Goodarzi pushes to simplify operations and improve efficiency.
Stability AI Unveils Stable Audio 3.0: The open-weight model adds six-minute audio generation, on-device song creation, and commercial rights.
Google AI Studio Adds Android App Builder: Users can now generate Android apps from prompts in minutes, expanding AI Studio into app creation.
