i am looking to extract some quotes from the recent invest like the best podcast with the cfo of anthropic. Can you help with that?

Anthropic's strategy is defined by extreme capital intensity and exponential growth, positioning it as a key player in a market where access to compute is a primary prerequisite for competition . The company has raised vast sums to fund its operations, with one source citing over **$7.5 billion** and another claiming a much larger figure of $75 billion since the current CFO joined [1, 3]. This capital fuels a business experiencing rapid expansion, evidenced by a 500%+ net dollar retention rate, revenue that doubled in the few months prior to one report, and a trajectory from zero to over $100 million in revenue in one year with projections to exceed $1 billion in the following year [1, 14, 28]. This growth has been accompanied by rapidly escalating valuations, with various sources citing investments at $60 billion, $180 billion, and a more recent private round valuing the company in the high $300s of billions, underscoring the market's belief in the high returns of frontier intelligence [2, 12, 18, 24]. The scale of these private funding rounds for Anthropic and its peers now exceeds that of any public tech IPO in the last five years .

To manage its largest capital expenditure, Anthropic treats compute as a strategic asset rather than a commodity . The company is reportedly "really compute constrained," making its hardware strategy critical . For years, Anthropic has invested in a proprietary orchestration layer and compilers that allow it to **fungibly use NVIDIA GPUs, Google TPUs, and AWS Tranium chips** [4, 17]. This multi-platform approach provides crucial flexibility in a supply-constrained environment, de-risking dependency on any single hardware provider and enabling cost and performance optimization for different workloads . This technical infrastructure is a core differentiator, supporting a global AI industry where annual CapEx spending is estimated at $300 billion and doubling roughly every year [13, 15].

Anthropic accelerates its own operational velocity and product development through extensive internal use of its models, creating a powerful flywheel for enterprise adoption . The finance team, for example, utilizes a library of over **70 custom Claude skills** to automate statutory financial statements and has reduced the time for weekly reporting on key metrics from hours to minutes [9, 30]. This "dogfooding" serves as a direct feedback loop for the product team and acts as a compelling proof point for enterprise customers, demonstrating tangible productivity gains . The demand from enterprise users is strong, with partners observing that employees consistently use their full token allocations for models like Claude and request more each month .

Looking forward, company leadership has made bold predictions about the pace of AI advancement, with CEO Dario Amodei forecasting that models will be smarter than all PhDs in any given subject by 2027 and could achieve a 90% score on a key coding benchmark by the end of 2025 [7, 19]. This rapid progress toward "transformative AI" is discussed alongside its profound societal implications, including the potential for massive job displacement and a co-founder's concrete forecast of a **0-10% probability** of an existential risk from AI [8, 23]. In response, the company publicly details its proactive safety testing for dangerous capabilities, such as bioweapon creation, and maintains transparency with policymakers, reflecting a dual focus on pushing the frontier of AI while attempting to mitigate its gravest risks .