The Frontier

Alex Kendall states Wave's self-driving technology has moved from scientific risk to engineering execution and deployment risk, with a clear scaling path involving increased data, compute, and algorithmic innovation.

Kendall says Wave partners with major automakers including Nissan, Mercedes, and Stellantis. Nissan announced it will deploy Wave's AI driver in 90% of its vehicles, which amounts to 2.7 million cars annually - double Tesla's yearly production.

Wave has raised over $2 billion in capital, positioning it to achieve free cash flow positive status without needing further funding to reach 'escape velocity'.

Wave's world models function as powerful representation learning tools and simulators, trained on petabytes of diverse sensor data aggregated from over a dozen partner companies.

Kendall expects autonomy to become a subscription service for consumers, similar to Tesla's $100/month model, due to ongoing updates and potential manufacturer-borne insurance costs.

Raquel Urtasun argues that scalable, safe self-driving requires a convergence of ready OEM hardware platforms, evolving regulatory frameworks, clear consumer demand, and powerful AI, with the market now primed for deployment.

Urtasun states Wabi's core technology is a 'verifiable end-to-end' system combined with a world model simulator, designed from inception to be generalizable across multiple robotics verticals like trucks and robotaxis.

Wabi's partnership with Uber for robotaxis is for a minimum of 25,000 vehicles, not merely 'up to' that number. The company is a technology provider and will partner with an OEM to supply the hardware.

Wabi operates on a 'driver-as-a-service' business model charging per mile, avoiding ownership of trucks or robotaxis. Its commercial trucking operations began in 2023 with a fleet in the double-digits.

Urtasun believes the industry mistakenly assumed L2 would sequentially evolve to L4, but building native L4 technology is a distinct safety problem and the only viable path to true driverless systems.

Both Kendall and Urtasun argue that licensing their AI driver to automakers and fleets is the most scalable business model, avoiding the capital intensity of building custom vehicles or operating city-specific fleets.

Wave is conducting supervised robotaxi trials in London, Tokyo, and 10 other cities via Uber this year, with consumer vehicle deployments starting next year.

The UN established a legal pathway for L3 and L4 autonomous driving two months ago, covering most countries except the US and China.

Jenny Fielding relays that a founder who raised a $15M Series A plans to return the capital, fearing AI will displace their legal-tech product.

Jenny estimates only 50% of SaaS-era portfolio companies will successfully pivot to AI, citing painful steps like firing non-AI-native executives and moving to usage-based pricing.

Dave McClure argues SPVs with excessive load fees create misalignment, drawing 'Wolf of Wall Street' characters, while authorized single-layer SPVs remain viable tools.

Jason Calacanis notes that only 6% of Americans are accredited investors, which Dave McClure disputes, saying it's more about concentrated wealth in areas like Palm Beach.

Dave McClure criticizes opaque private company finances, arguing transparency should lower capital costs, contrary to the current system rewarding secrecy.

Jenny Fielding says late-stage investors often pressure founders to force seed investors to waive pro rata rights in competitive rounds.

Jason Calacanis predicts a 'vibe shift' where retail investors fueled SpaceX's rise, but institutional scrutiny of cash flow will weigh Anthropic and OpenAI valuations.

Jason cites Intercom's pivot to an AI agent named Finn as a successful late-stage transformation, requiring bold leadership and painful changes.

Jason Calacanis argues founder-led companies like Tesla and SpaceX operate on 'founder authority,' enabling drastic pivots like killing products, unlike manager-led firms like Google.

Dave McClure states sovereign wealth funds in Qatar, UAE, and Saudi Arabia control over $4 trillion and aren't shutting off venture investment spigots.

Jenny Fielding's advice to solo founders pitching VCs is to focus on business momentum over traction metrics and avoid apologizing for lacking a co-founder.

Jason Calacanis notes OpenAI's recent tender offer let 600 employees sell shares worth $6.6B, highlighting secondary market liquidity for concentrated wealth.

AI21's Maestro platform uses a proprietary 'meta-model' to orchestrate multiple AI models, predicting cost, latency, and accuracy to route enterprise queries for maximum efficiency.

AI21's open-weight Jamba model family includes a 400-billion-parameter version and a 13-billion-parameter mixture-of-experts model, combining transformer and Mamba architectures for efficient long-context processing.

AI21 claims its Maestro orchestration system can reduce enterprise AI token costs by up to 50% by dynamically routing queries across a portfolio of frontier and open-weight models.

Magrathea Metals is building the first US primary magnesium smelter in a generation, targeting a production cost of $3,000 per ton to compete with Chinese imports priced around $7,000.

China currently controls 95% of global magnesium supply, primarily using the coal-intensive 'Pidgeon' process, creating a strategic vulnerability for US defense and aerospace industries.

Magrathea's initial commercial smelter in Arkansas will target the 10,000-ton annual defense industrial base demand, using high-grade brine from 10,000 feet underground as feedstock.

Cerebras raised its IPO price range to $150-$160 per share, valuing the company between $34.4B and $48.8B, representing a 50x to 71x multiple on its $686M annualized revenue run rate.

OpenAI has spun out a separate 'OpenAI Deployment Company' funded by TPG, Warburg Pincus, and Bain to push its AI into private equity portfolio companies, a structure Calacanis calls convoluted financial engineering.

Google Trends data shows searches for 'OpenClaw' peaked in mid-March 2025 and have declined since, which Wilhelm attributes to competition from tools like Perplexity Computer, Co-Work, and Grok.

TikTok is testing an ad-free subscription tier priced at £4 per month in the UK, a model Calacanis sees as a potential regulatory hedge against privacy complaints by offering a non-tracking option.

Calacanis argues that intractable federal issues like education certification, healthcare, and housing policy should be delegated to states for experimentation, citing cannabis and online poker as successful precedents.