What I Read This Week…

Google faces a new antitrust trial this week, AI startups are shifting away from developing large language models, and why innovative drugs for life-saving treatments are so expensive

Chamath Palihapitiya

Sep 08, 2024

Watch All-In E195

Read our Deep Dive on Drug Development

Caught My Eye…

Google faces a new antitrust trial this week, brought by the Department of Justice and a coalition of states. Unlike the previous case focused on Google's dominance in internet search, this trial examines the company's advertising technology. What's at stake in this trial? The trial's outcome could determine the future of Google's digital advertising business, which generates over $200 billion in annual revenue. The Department of Justice argues that Google formed its monopoly through strategic acquisitions, including its acquisition of DoubleClick in 2007. These acquisitions have contributed to Google’s 91% market share in ad servers, which are the platforms that enable website owners to manage, display, and track ads on their websites. The Department of Justice argues that this gives Google undue influence over the process of bidding, serving, and placement of online ads, enabling it to extract disproportionate profits from both advertisers and publishers.
AI startups are pivoting away from developing large language models towards other areas of the AI ecosystem. Why is this happening? Training a competitive large language model currently costs around $100 million, with projections suggesting this could increase to $1 billion in the near future. Intense competition from well-funded tech companies and open-source projects has significantly reduced the potential for startups to generate the outsized returns that would justify the high cost of training a leading-edge large language model. Meanwhile, major tech companies continue to invest heavily in AI. Alphabet's capital expenditures nearly doubled to $25.2 billion in the first half of 2024, driven by investments in data centers and servers to support its AI products. Similarly, Meta plans to invest up to $40 billion in AI this year alone. Despite the potentially poor returns on developing large language models, the tech giants view their investments in AI as key to maintaining their competitive edge in a transformative technology.
Opdivo is a novel immunotherapy drug that treats certain forms of cancer. It works by blocking a protein on specific immune cells, enabling the immune system to attack cancer cells it would normally ignore. This allows the body's natural defenses to recognize and destroy cancer more effectively. However, a year's supply of Opdivo at its list price costs more than $150,000. Why is the price so high? Developing innovative cancer drugs like Opdivo requires substantial investment in research, extensive clinical trials, and complex regulatory approval processes, often costing billions of dollars. Given the high costs and low success rates associated with drug development, companies and investors set high prices for successful drugs like Opdivo to recover their investments and earn a return that justifies the risk they took. The price also reflects Opdivo's potential to be a life-saving treatment, which creates high demand despite the cost. And as a targeted therapy, Opdivo may only be effective for a small subset of patients, leading to even higher prices to generate enough return from a smaller customer base. Thus, in many cases where there is drug innovation and a potentially life-saving treatment, it often comes with a high price tag for patients and our healthcare system.