Axios Pro Rata: 🤖 Sourcing AI | Lucinda Shen Summary

1. Key Themes

Theme 1: The Open-Source AI Race Is Now an American Story, Not Just a Chinese One

Silicon Valley is pushing back hard against proposed bans on Chinese AI models, driven by fears of vendor lock-in with OpenAI and Anthropic. The community is actively seeking a competitive U.S.-built open alternative.

"The open frontier is no longer just a Chinese story. It is an American one too." — Benchmark's Bill Gurley, Washington Post op-ed

"Nearly everyone in the AI economy has a reason to prefer an open foundation — everyone, that is, except the big incumbents Anthropic and OpenAI, whose fortunes depend on keeping it closed." — Bill Gurley

Theme 2: AI Model Benchmarks Are Ephemeral — Volatility Is the New Normal

The competitive landscape shifts so rapidly that today's leading model may be irrelevant in months, making long-term bets on specific model providers extremely risky.

"The conversation today has centered on Kimi K3, but the leading model in three to six months could be a whole different story."

"The most important things to understand about the new AI race right now is that every metric is changing really fast right now." — Artificial Analysis CEO Micah Hill-Smith

Theme 3: AI Infrastructure Is Attracting Institutional Capital at Massive Scale

Private equity and institutional investors are committing tens of billions to the physical layer of AI — data centers and compute.

The Artificial Intelligence Infrastructure Partnership, a consortium including BlackRock's Global Infrastructure Partners and MGX, "completed their previously announced acquisition of Aligned Data Centers from Macquarie Asset Management and investors for $40b. Investors are considering putting another $5b into the deal."

Theme 4: ESG as an Investment Theme Is in Structural Decline

The retreat from ESG is no longer just rhetorical — it is now materialized in deal failures and distressed asset transfers, signaling the end of the ESG consulting boom.

"The talks land amid a broader reckoning for environmental, social and governance consulting, which rode a wave of corporate net-zero pledges through the early 2020s but has since run into mounting political backlash." — Silas Brown, Bloomberg

"The term ESG is even falling off in corporate earnings calls."

2. Contrarian Perspectives

Banning Chinese AI Models Would Entrench American Incumbents, Not Protect American Interests

The consensus framing of Chinese AI bans is national security; the contrarian view from within Silicon Valley is that banning Chinese models is primarily anti-competitive, benefiting OpenAI and Anthropic at the expense of the broader AI economy.

"Efforts to ban Chinese AI models in Washington, D.C., are meeting strong resistance in Silicon Valley from entrepreneurs and investors who fear it hands more power to Anthropic and OpenAI."

"Nearly everyone in the AI economy has a reason to prefer an open foundation — everyone, that is, except the big incumbents Anthropic and OpenAI, whose fortunes depend on keeping it closed." — Bill Gurley

Closed Models' Revenue Advantage May Be a Structural Trap

While closed models appear financially superior now, their monetization moat may be narrowing as open-source alternatives improve, costs fall, and companies deliberately diversify away from single vendors.

"Closed models can make more money by charging for each query, but the pathway to monetization for open models is rockier."

"AI costs are ballooning, and companies don't want to be reliant on just one or two vendors for such critical tech."

Chinese AI Firms Are Still Creating Equity Value Despite Model Competition

Despite Kimi K3's release rattling global markets and triggering a selloff in rival Z.AI, Z.AI's market cap remains robust — suggesting competitive model releases haven't destroyed investor confidence in the sector overall.

"Shares of Z.AI are still above IPO levels, valued at HK$543.4 billion, or $69.3 billion, as of Tuesday morning."

3. Companies Identified

Thinking Machines

Description: AI model lab led by former OpenAI CTO Mira Murati
Why mentioned: Released its first open-weights model last week, an early signal of the emerging U.S. open-source AI ecosystem
Quote: "Thinking Machines, led by former OpenAI CTO Mira Murati, dropped its first model (an open-weights one), last week."

Reflection AI

Description: AI model company valued at $25 billion
Why mentioned: Closely watched by the industry despite having yet to release a model; represents speculative institutional confidence in U.S. open AI alternatives
Quote: "The industry is also closely watching $25 billion Reflection AI, which has yet to release a model."

Moonshot AI (Kimi K3 maker)

Description: Chinese AI company behind the Kimi K3 model
Why mentioned: Its newest model release rattled global markets; the company is reportedly preparing for a Hong Kong IPO within six months
Quote: "Kimi K3 maker Moonshot is reportedly preparing for a Hong Kong IPO in as soon as six months, per Bloomberg. That follows the release of its newest model last week rattling global markets, leading to a selloff in rivals including Z.AI."

Fluidstack

Description: NYC-based AI cloud platform
Why mentioned: Raised $830M Series A at a $7.5B valuation, one of the largest AI infrastructure rounds noted in this issue
Quote: "Fluidstack, a NYC-based AI cloud platform, raised $830m in Series A funding at a $7.5b valuation in January, led by Situational Awareness."

Anthesis Group

Description: London-based ESG consultancy, majority owned by Carlyle since 2023
Why mentioned: Central case study in the ESG industry's collapse; Carlyle is in talks to hand control to lender Bridgepoint in what amounts to a distressed exit
Quote: "Carlyle is in talks to hand control of ESG consultancy Anthesis Group to its lender Bridgepoint Group."

Artificial Analysis

Description: AI benchmarking and analysis firm
Why mentioned: Cited as a source of authoritative perspective on AI model cost and performance trends
Quote: "Costs are going down inside of closed-source AI labs, while services companies are emerging aiming to also bring down bills."

Vertiv Holdings (NYSE: VRT)

Description: Public infrastructure company
Why mentioned: Acquired Strategic Thermal Labs, a liquid cooling tech startup — a direct signal of data center thermal management becoming a critical investment area as AI compute scales
Quote: "Vertiv Holdings (NYSE: VRT) acquired Strategic Thermal Labs, a Georgetown, Texas-based liquid cooling technology company."

Tempus AI (Nasdaq: TEM)

Description: AI-driven precision medicine company
Why mentioned: Agreed to acquire Personalis for $1.7B, demonstrating large-scale M&A consolidation in AI-enabled oncology
Quote: "Tempus AI (Nasdaq: TEM) agreed to buy Personalis (Nasdaq: PSNL), a Fremont, Calif.-based cancer genomics company, for $1.7b."

Scribe Therapeutics

Description: Alameda, Calif.-based molecular engineering startup co-founded by Jennifer Doudna
Why mentioned: Filing for an IPO to raise up to $108M; notable for its scientific pedigree and blue-chip investor base
Quote: "Scribe Therapeutics...plans to raise up to $108m in an IPO...It's raised $121m from a16z, Avoro Ventures, Eli Lilly, OrbiMed, T. Rowe Price, Perceptive Advisors, Wellington Management, RA Capital Management, and Menlo Ventures."

Neo (Boston)

Description: Boston-based agentic cybersecurity company
Why mentioned: Raised $100M in seed and Series A — a very large early-stage raise signaling strong investor conviction in AI-native security
Quote: "Neo, a Boston-based agentic security company, raised $100m in seed and Series A funding. Andreessen Horowitz, Bessemer Venture Partners, Craft Ventures and Merlin Ventures invested."

Bluecore Energy

Description: Port of Long Beach, Calif.-based maker of small modular reactors on barges
Why mentioned: Pre-seed raise for an innovative modular nuclear energy approach; notable for the novel deployment format (barge-mounted SMRs)
Quote: "Bluecore Energy, a Port of Long Beach, Calif.-based maker of small modular reactors on barges, raised $10m in pre-seed funding."

4. People Identified

Bill Gurley

Description: General Partner at Benchmark, veteran venture capitalist
Why mentioned: Authored a Washington Post op-ed making the affirmative investment and policy case for U.S. open-source AI models
Quote: "The open frontier is no longer just a Chinese story. It is an American one too."

Mira Murati

Description: Former CTO of OpenAI, now CEO of Thinking Machines
Why mentioned: Her new company released its first open-weights AI model, positioning her as a key figure in the U.S. open-source AI movement
Quote: "Thinking Machines, led by former OpenAI CTO Mira Murati, dropped its first model (an open-weights one), last week."

Micah Hill-Smith

Description: CEO of Artificial Analysis
Why mentioned: Provided the clearest articulation of the AI competitive dynamic — metrics shifting rapidly, costs falling across both closed labs and emerging services companies
Quote: "The most important things to understand about the new AI race right now is that every metric is changing really fast right now."

Jennifer Doudna

Description: Nobel Prize-winning biochemist, co-inventor of CRISPR
Why mentioned: Co-founded Scribe Therapeutics, which is now filing for an IPO
Quote: "Scribe Therapeutics, an Alameda, Calif.-based molecular engineering startup co-founded by Jennifer Doudna, plans to raise up to $108m in an IPO."

5. Operating Insights

Diversify AI Vendor Exposure Now — Single-Vendor Dependency Is a Business Risk

The article makes clear that enterprise reliance on one or two closed AI vendors is increasingly viewed as a strategic liability, not just a cost problem. Operators should evaluate open-weight model alternatives, even if they require more integration work, to preserve negotiating leverage and business continuity.

"AI costs are ballooning, and companies don't want to be reliant on just one or two vendors for such critical tech."

Treat AI Benchmarks as Lagging Indicators, Not Procurement Criteria

Given the 3-6 month model leadership cycle, operators who build product strategy around the current "best" model risk costly pivots. Architecture decisions should favor model-agnostic or multi-model approaches.

"The conversation today has centered on Kimi K3, but the leading model in three to six months could be a whole different story."

6. Overlooked Insights

Hugging Face Turned to Chinese Open-Source AI for Cybersecurity Defense

Buried in the article's links is a Fortune reference noting that Hugging Face — a central node in the global AI ecosystem — turned to Chinese open-source models after American AI guardrails blocked its defensive AI capabilities. This is a concrete, operational example of why the Chinese model ban debate has real consequences beyond ideology.

Companies don't want to "be reliant on just one or two vendors for such critical tech" — and Hugging Face's pivot illustrates what happens when U.S. regulatory constraints force that hand.

Anthropic Is Simultaneously Suing a Competitor and Settling a $1.5B Author Class Action

Anthropic's dual legal exposure — offensive trademark litigation against Abnormal AI while simultaneously closing a $1.5B copyright settlement with authors — is a signal of the growing legal complexity facing frontier AI labs, with implications for compliance costs and IP strategy across the industry.

"Anthropic has sued Abnormal AI alleging trademark infringement. At the same time, a judge granted Anthropic final approval for its $1.5b settlement of a class action lawsuit brought by a group of authors."