Anthropic Is Playing Hard to Get with…

1. Key Themes

AI Company Valuations Are Decoupling From Normal Fundraising Logic

Anthropic's revenue trajectory is so steep that the company can afford to ignore investors offering 2.3x its last-round valuation. The company reportedly surged from $9 billion to $30 billion in annualized revenue in just months, yet still rebuffed offers valuing it at $800B+.

"Anthropic has rebuffed investor offers that would value it at $800 billion or more even as its revenue has reportedly surged to $30 billion from $9 billion just months earlier. The company last raised $30 billion in February at a $350 billion pre-money valuation."

Infrastructure-First Fintech Is the Defensible Moat Play

Airwallex spent a decade building proprietary payment rails and accumulating regulatory licenses rather than chasing growth. That foundation is now becoming a strategic weapon as it expands into physical-world payments and challenges Stripe, Square, and Adyen simultaneously.

"The startup currently boasts close to 90 regulatory licenses across roughly 50 markets, direct connections to local payment networks in over 120 countries, and the ability to settle transactions in more than 90 currencies. It is the very infrastructure, Zhang says, that Stripe and Square lack in meaningful ways — particularly the local banking licenses that allow funds to be held, converted, and deployed within a given market rather than immediately repatriated."

AI as Corporate Survival Tactic (Not Just Growth Driver)

Snap's mass layoff illustrates how established tech companies are restructuring around AI efficiency to cut costs at scale — not just to grow faster, but to survive financially.

"Snap is laying off about 1,000 employees or 16% of its workforce, claiming that AI efficiency gains coupled with a smaller workforce could lead to more than $500 million in savings by the second half of 2026."

AI Infrastructure Capital Is Concentrating Fast

From Accel's $5B raise (with $4B dedicated to late-stage AI), to nEye's $80M for AI data center switching chips, to Parasail's $32M for distributed AI inference routing, capital is flowing heavily into the plumbing beneath AI applications.

"Accel has raised $5 billion in new capital, including $4 billion for its late-stage Leaders Fund, which is expected to invest an average of $200 million in about 20 AI-focused companies."

Nuclear Power as AI Infrastructure Play

X-energy's IPO roadshow is directly tied to AI data center power demand — signaling that nuclear energy is becoming a legitimate investment category within the AI infrastructure stack.

"X-energy, a nuclear startup backed by Amazon that is developing high-temperature gas-cooled reactors, has begun its IPO roadshow targeting $16 to $19 per share and up to about $814 million in proceeds amid rising power demand from AI data centers."

2. Contrarian Perspectives

Anthropic Is Rationally Leaving Hundreds of Billions on the Table — and That's the Right Call

The consensus VC playbook says: raise as much as you can at the highest valuation possible. Anthropic is doing the opposite — declining offers at $800B+ despite its revenue tripling in months. This is a signal that Anthropic's leadership believes the current investor demand reflects froth, not fundamental value, and that dilution at a potentially peak valuation is the greater risk.

"Anthropic has rebuffed investor offers that would value it at $800 billion or more even as its revenue has reportedly surged to $30 billion from $9 billion just months earlier."

Slow Infrastructure Building Beats Fast Growth Hacking in Payments

Conventional fintech wisdom rewards fast scaling and distribution. Airwallex rejected a $1.2B acquisition offer from Stripe when it had just $2 million in revenue and instead spent years building its own rails. That contrarian patience has produced an $8B company processing $100B annually.

"In 2019, Stripe offered to acquire Airwallex for $1.2 billion, when Airwallex had just $2 million in revenue. But Zhang decided to keep building... 'What really got me to change my mind is when I actually flew back to Melbourne and went deep on what motivated me to build Airwallex.'"

AI Is Not the Reason Hiring Is Down — Consensus Is Wrong

The dominant narrative blames AI for the labor market slowdown. LinkedIn's own data pushes back on that, pointing to interest rates as the primary driver. This matters for investors and operators making workforce bets predicated on AI displacement.

"LinkedIn says hiring is down about 20% since 2022 but finds no evidence AI is responsible, instead pointing to higher interest rates as the primary driver."

3. Companies Identified

Anthropic AI safety and frontier model company Turning down $800B+ investor offers despite revenue tripling to $30B ARR in months; case study in valuation discipline.

"Anthropic has rebuffed investor offers that would value it at $800 billion or more even as its revenue has reportedly surged to $30 billion from $9 billion just months earlier."

Airwallex Australian global payments fintech, valued at $8B Entering in-person POS payments as the culmination of a decade-long infrastructure build; direct challenger to Stripe, Square, and Adyen.

"Airwallex claims it generates annualized revenue of about $1.3 billion, and that the number is growing by roughly 85% every year. The startup says it now serves more than 46,000 U.S. businesses and processes $100 billion in annual volume."

Accel Global venture firm Raised $5B with $4B earmarked for late-stage AI companies; signals where institutional LP conviction is concentrated.

"Accel has raised $5 billion in new capital, including $4 billion for its late-stage Leaders Fund, which is expected to invest an average of $200 million in about 20 AI-focused companies."

Artemis New York AI cybersecurity startup, 6 months old Raised $70M Series A to replace rule-based security tools with an AI "brain"; notable for its stage and size of round relative to its age.

"Artemis, a six-month-old New York startup that aims to use an AI 'brain' to replace rule-based security tools, raised a $70 million Series A round led by Felicis."

nEye Santa Clara optical switching chip startup Raised $80M Series C for chips that connect servers in AI data centers; plays directly into AI compute infrastructure demand.

"nEye develops optical switching chips that connect servers in data centers for AI workloads."

X-energy Nuclear reactor startup, Amazon-backed IPO roadshow underway targeting $814M; positions nuclear power as a direct infrastructure layer for AI data centers.

"X-energy has begun its IPO roadshow targeting $16 to $19 per share and up to about $814 million in proceeds amid rising power demand from AI data centers."

Snap Social media company Cutting 16% of workforce citing AI-driven efficiency; early data point on how AI is being used to justify large-scale tech layoffs.

"Snap is laying off about 1,000 employees or 16% of its workforce, claiming that AI efficiency gains coupled with a smaller workforce could lead to more than $500 million in savings by the second half of 2026."

Wayve UK autonomous driving startup Raised $60M Series D extension from AMD, Arm, and Qualcomm; notable for its sensor-data-only neural net approach (no maps or fixed hardware), now $2.8B raised total.

"Wayve uses end-to-end neural networks trained on sensor data rather than maps or fixed hardware to power autonomous driving systems."

Flock Safety Atlanta surveillance technology company, valued at $7.5B In talks for new round; license plate–reading AI for law enforcement is a large and growing segment of public safety tech.

"Flock Safety provides license plate–reading surveillance technology used by law enforcement."

Mercor AI hiring startup Run by 23-year-old billionaire founders; facing fraud and suspected North Korean infiltration as it scales contractor workforce for AI model training. Significant operational risk signal for the remote AI labor market.

"Mercor is grappling with employee fraud and even suspected North Korean infiltration as it scales a vast global contractor workforce to train A.I. models."

Allbirds / NewBird AI Shoe brand turned GPU-as-a-service company Sold shoe assets for $39M, secured $50M investment, and rebranded as an AI compute provider; emblematic of the "AI pivot" phenomenon reaching absurdist levels.

"Allbirds, formerly the favored shoe brand of the Silicon Valley vest-wearing set, is rebranding as NewBird AI and pivoting to a GPU-as-a-service business after selling its shoe brand and assets for $39 million and securing a $50 million investment."

Thinking Machines Lab AI startup valued at $12B Has lost at least five founding engineers to Meta in an intensifying talent war; serious signal about retention challenges even at well-capitalized AI startups.

"Meta has hired Joshua Gross, a founding engineer at Thinking Machines Lab, marking its fifth such hire from the $12 billion AI startup as competition for top talent intensifies."

Live Nation / Ticketmaster Live entertainment and ticketing monopoly Found by a jury to have illegally maintained monopoly power; potentially opens path for competitive disruption in ticketing.

"A jury found Live Nation and its subsidiary Ticketmaster illegally maintained monopoly power in the ticketing market, potentially opening the door to further damages and remedies."

Parasail San Francisco AI inference routing startup Raised $32M Series A to route AI workloads across distributed data centers to reduce compute costs; solves a real cost problem as AI inference spend scales.

Hilbert San Francisco AI growth analytics, 1 year old Raised $28M Series A from a16z to connect cross-team data and surface AI-driven growth recommendations.

Keebler Health Durham, NC, health AI startup Uses LLMs to process unstructured clinical data for risk adjustment; $16M Series A — early signal of LLMs finding real workflow traction in healthcare billing.

Ralio London payments infrastructure, 1 year old Building payment rails specifically for AI agents with audit trails and controls — early-mover in the emerging "agentic payments" category.

4. People Identified

Jack Zhang CEO and co-founder, Airwallex Turned down a $1.2B Stripe acquisition offer in 2019 when Airwallex had $2M in revenue; credited the infrastructure-first strategy and founder conviction for building to $8B valuation and $1.3B ARR.

"I even said yes to the deal. But what really got me to change my mind is when I actually flew back to Melbourne and went deep on what motivated me to build Airwallex."

Marc Andreessen & Ben Horowitz Co-founders, Andreessen Horowitz Contributed $25M to the pro-AI super PAC Leading the Future, boosting its total to $51M+; signals that Silicon Valley's most prominent VC firm is moving aggressively into political influence on AI regulation.

"Marc Andreessen and Ben Horowitz have contributed $25 million to the pro-AI super PAC Leading the Future, boosting its war chest to more than $51 million as Silicon Valley ramps up spending to influence regulation ahead of the November midterms."

Joshua Gross Founding engineer, Thinking Machines Lab (now hired by Meta) Fifth founding engineer Meta has poached from Thinking Machines; emblematic of the talent poaching war between large-cap tech and well-funded AI startups.

"Meta has hired Joshua Gross, a founding engineer at Thinking Machines Lab, marking its fifth such hire from the $12 billion AI startup as competition for top talent intensifies."

5. Operating Insights

Building Proprietary Infrastructure Is the Highest-Leverage Defensibility Strategy in Fintech (and Likely Elsewhere)

Airwallex's decade of building its own payment rails — rather than relying on third-party acquirers — is now the competitive moat enabling it to expand into new geographies and product categories faster than rivals. The lesson: resist the temptation to use off-the-shelf infrastructure if your business model is fundamentally about moving value across borders or systems.

"When a business expands into a new market, they typically have to onboard a new local acquirer, navigate fragmented compliance, and manage yet another set of vendor relationships."

AI Is Now a Legitimate Tool for Restructuring Cost Bases, Not Just Growing Revenue

Snap's announcement frames AI not as a product feature but as an operational efficiency lever that justifies eliminating 16% of headcount. Operators should be modeling AI's impact on their own cost structures — not just their product roadmaps.

"Snap is laying off about 1,000 employees or 16% of its workforce, claiming that AI efficiency gains coupled with a smaller workforce could lead to more than $500 million in savings by the second half of 2026."

Fundraise Preparation Is Evaluated Beyond the Pitch Deck

Investors are scrutinizing operational hygiene — cap table clarity, financial consistency, and responsiveness under follow-up — as part of diligence. Founders who cannot answer follow-on questions without scrambling signal execution risk before a term sheet is issued.

"Investors don't just listen to what you say — they look at how your company operates. Is ownership clear? Do your numbers match your story? Can you answer follow-up questions without digging through spreadsheets?"

6. Overlooked Insights

Agentic Payment Infrastructure Is Becoming Its Own Category

Ralio's $2.5M seed round — building payment rails specifically designed for AI agents to execute transactions with built-in controls and audit trails — is a quiet but significant bet. As AI agents take on autonomous financial tasks, the payment infrastructure they operate on will need entirely different compliance and control architectures than human-facing payments. This is an early but structurally important space.

"Ralio provides a payments infrastructure that enables AI agents to execute transactions with built-in controls and audit trails."

Meta Is Systematically Dismantling Thinking Machines Lab's Founding Team

Five founding engineers from a $12 billion AI startup hired away by a single buyer is not a coincidence — it is a targeted acquisition of human capital. For investors in Thinking Machines, this is a material risk signal. For competitors and recruiters, it illustrates that Meta is using its balance sheet aggressively to consolidate frontier AI talent.

"Meta has hired Joshua Gross, a founding engineer at Thinking Machines Lab, marking its fifth such hire from the $12 billion AI startup as competition for top talent intensifies."