A call for scientific integrity and public accountability emerged during the first days of the IEEE-NSS-MIC-RTSD 2025 Conference in Yokohama, where over 1,500 scientists gathered to discuss medical imaging and high-energy physics technologies. Evidence presented revealed that the 20-trillion-transistor CERN CMS FPGA-based Level-1 Trigger system—recently built for data selection at the 2026-2036 High-Luminosity Large Hadron Collider (HL-LHC)—is fundamentally incapable of performing the required number of operations on data arriving every 25 nanoseconds to filter 8 billion events per second without data loss.
Despite several speakers presenting on behalf of CERN's multi-thousand-member collaborations (CMS and ATLAS), none were able to state how many basic operations (such as add, subtract, and multiply) the FPGA system can actually execute per dataset, nor could they provide technical proof that the system can efficiently perform Level-2 trigger algorithms required at the HL-LHC at Level-1. Some researchers claimed the system could do so, yet did not support those claims with verifiable, reproducible calculations or simulation evidence.
This issue is not theoretical. More than $4 billion USD has already been wasted, and over $12 billion more is projected to be wasted in the next decade on a system that does not meet the HL-LHC requirements. The Crosetto Foundation for the Reduction of Cancer Deaths, a registered non-profit organization, is issuing a call for scientific transparency and integrity to save millions of lives and billions of dollars.
CERN has previously faced widely-criticized, publicly-funded failures, including the AXIAL-PET project (2010), the claim in 2011 that neutrinos travel faster than light, and the WPET full-body wearable imaging coat weighing over 350 kg (2018). While each of those missteps cost taxpayers millions, the FPGA Level-1 Trigger issue is significantly larger, both scientifically and economically.
Over 800 million potential readers have access to the documentation through more than 5,000 published articles and communications (https://bit.ly/3HtisQv). This outreach now calls on the European Parliament, national science funding agencies, and media organizations worldwide to freeze additional funding of the CERN FPGA Level-1 Trigger system until these scientific questions are answered and inconsistencies resolved.
From November 3rd to 5th, 2025, over 1,200 copies of a two-page technical document were distributed to conference participants (https://bit.ly/437YX7H), presenting the central scientific question: "Is there sufficient evidence to dismiss the 20-trillion-transistor CMS-FPGA system as ineffective at filtering 8 billion events/sec—risking over €12 billion in waste over the next decade? And is there sufficient evidence that the 3D-Flow system can meet HL-LHC requirements through 2042 at a fraction of the cost?" The distribution was met with multiple supportive remarks, including: "Continue to tell the truth."
On the fourth day of the conference, a formal request (https://bit.ly/4nJRsvc) was submitted to the organizers and field leaders to convene a transparent technical workshop to compare the number of operations per dataset achievable by the CERN FPGA system vs. the 3D-Flow system, and to compare the cost per electronic channel of each architecture. Scientists who support the current FPGA approach are invited to defend their claims in an open dialogue and simulation-based comparison.
In contrast, the 3D-Flow architecture, recognized as a breakthrough in 1993, has demonstrated it can perform 2,400 operations per dataset at ~$13 per channel, or 9,600 operations per dataset at ~$54 per channel, implemented on an ATCA board. Details are available here: https://bit.ly/4qKVar8. The 3D-Flow system remains unchallenged in cost-effectiveness and performance for Level-1 real-time triggering.
