[sdecoret/Adobe Stock]
The fact that the biopharmaceutical industry has a large carbon footprint is well established. A 2022 study from My Green Lab confirmed that biotech and pharma are still among the globe’s top polluters. The research highlights that a mere 4% of the largest publicly-traded biotech and pharmaceutical firms have climate commitments in line with the UN’s Intergovernmental Panel on Climate Change (IPCC) to cap warming at 1.5°C by 2030.
Yet climate change remains the most pressing threat to human health, potentially causing 250,000 additional deaths per year, as the World Economic Forum has noted.
AI as a doubled-edged sword in sustainability
For biopharma, AI-driven drug discovery could serve as a double-edged sword when it comes to sustainability. On the one hand, these technologies promise to enable new discoveries and optimized processes that could slash emissions.
“We’ve developed our AI to determine factors impacting our clients’ operations, including their environmental footprint and energy usage patterns,” said Chris Noble, CEO of Cirrus Nexus, an IT services and consulting company.
Chris Noble
Noble is referring to TrueCarbon, a carbon monitoring platform that provides real-time data on their cloud emissions and AI-backed recommendations to optimize renewable energy use.
The company’s AI system has learned to maximize the use of renewable energy by predicting when solar, wind and other green power will be available at a given location. By forecasting renewable energy supply and shaping computing workloads around these predictions, AI enables biopharma firms to tap into green power that might otherwise be wasted, Noble noted.
GPUs and the energy intensity of AI in drug discovery
One example of an energy-intensive AI-driven drug discovery process involves the use of graphic processing units (GPUs), which can accelerate computationally intensive tasks such as molecular docking and molecular dynamics simulations. By tapping GPUs, drug companies can speed up identifying promising candidates. But GPUs, whether used locally or in a cloud data center, are power hungry. A single data center grade GPU, such as the NVIDIA V100, can consume between 250 and 300 watts. That’s more than most modern large-screen LED TVs, which typically consume between 60 and 200 watts, depending on the model and settings.
Added to that are the significant energy required to cool data centers, which house the powerful computing infrastructure enabling these technologies.
Roughly 30% to 55% of a data center’s energy consumption powers its cooling and ventilation systems, with the average in the ballpark of 40%.
The massive data crunching required for AI may substantially boost energy usage and, by extension, carbon emissions from the sector if not properly managed, according to Noble.
“Consider email,” Noble said. “We did a study with one biopharmaceutical company and found that sending an email with a one-megabyte attachment generates about 36 grams of carbon from end to end.” Powerful AI systems in drug discovery, whether crunching data related to drug hits or clinical trial participants, often demand massive energy to monitor, learn and update themselves.
AI for data-powered operations
Noble emphasizes the potential of AI to improve the environmental sustainability of the biopharmaceutical industry. He points out that AI can identify inefficiencies in data center operations, stating, “AI excels at finding patterns in huge datasets that humans often miss. By applying AI to analyze the energy consumption data from data centers, it can determine optimal efficiency settings and find waste that can be eliminated.”
For instance, Renewable-powered data centers and precision load balancing can help achieve sustainability, especially with health data continuously fueling biopharma AI systems. Noble argues that the strategic use of AI and predictive analytics may be key to overcoming the technology’s own environmental impact in sectors, like biopharma, that rely on massive computing power and data use.
Balancing AI-driven drug discovery with sustainability
Noble underscores that companies have options when choosing cloud service providers (CSPs) based on their processing and timing needs, financial constraints, and the environmental impact of their energy sources. Noble recommends timing workloads based on renewable energy availability to further reduce costs and carbon impact.
Biopharma may lag in sustainability, but Noble emphasizes the sector’s growing environmental awareness and the urgency to tackle AI and big data analytics’ massive energy demands. He warns, “By 2030 or 2040, [technology] could account for 15% of global carbon emissions. If tech companies claim carbon neutrality or negativity, why are emissions rising? We must reduce carbon production, not shift responsibility.”
Noble calls for prioritizing sustainable data infrastructure and renewable energy, stressing, “AI’s hidden cost is the immense energy consumed while ingesting and processing data.” To mitigate AI’s carbon footprint while harnessing its power to optimize energy efficiency, he advocates for a solution involving renewable-powered data centers, optimized cooling and load balancing. Noble’s message is clear: If AI is a double-edged sword, biopharma companies must wield it wisely — tapping it to fuel scientific breakthroughs while also minimizing its environmental impact and promoting sustainable practices in the industry.