Alex Rives, head of science at Biohub, calls one of his institute's newest instruments "the most powerful standing wave laser in the world." Its job is to make individual molecules visible. In a conversation with Nathan Labenz, host of The Cognitive Revolution, Rives said the device reflects one of Biohub's core principles: "so many historic paradigm shifts in science have really come from new technologies for measurement." The episode was published on October 20, 2026.
Biohub wants to build what Rives calls a world model of life. This would be an AI system that predicts how a cell or tissue changes from one state to the next, backed by an expanded $1.8 billion Virtual Biology Initiative. Rives argues that a model like this can only be as good as the observations it learns from, and that biology still cannot observe much of what it needs to.
Richer data or more of it
Labenz framed the question with an analogy from self-driving cars, where people have long debated expensive LiDAR sensors against cheap cameras. The question there is how much richer data matters compared with simply collecting more of it. In biology, he asked, should researchers do "super expensive stuff, not as much of it," or rely on cheaper techniques? And how should they build a portfolio of different kinds of data?
Rives didn't choose between the two. Biology has parallels with other fields where AI is advancing, he said, but its biggest difference is the limit on our ability to "observe, see, and measure what matters." Much of biology's progress has come from new tools for seeing more, and more powerful tools allow richer models. He described Biohub's measurement work as pushing in four directions at once. The first is measuring more parameters in every cell. The second is widening the field of view so more cells can be watched at the same time. The third is adding time, so cells can be followed as they change. The fourth is the molecular dimension: seeing what is happening inside cells "down at the most fundamental levels."
Phase contrast for electrons
The laser phase plate serves that fourth direction. Rives summed it up as "phase contrast for electron microscopy." When a wave passes through a sample, it can be shifted in phase (its timing changes) without getting any dimmer, and a detector can't see that shift directly. Phase contrast converts those hard-to-detect phase differences into visible differences in brightness, as UC Berkeley's explanation of the device puts it.
The laser version applies that idea to a beam of electrons. A mirrored optical cavity repeatedly amplifies and focuses an intense laser, and the light shifts the phase of the electrons passing through it. Rives described it as "a beam of light, which interferes with the electron beam to shift its phase."
The main application, he said, is cryo-electron tomography. In cryo-electron microscopy, samples are frozen to limit damage from the electron beam. Tomography combines views taken from many angles into a three-dimensional reconstruction, so researchers can study molecules where they sit in preserved cells. Rives described the method as sectioning the cell and using electron microscopy to see proteins and other large molecules in "their living context." The difficulty is that the images are, in his words, "incredibly, incredibly noisy." Better contrast would let researchers pick out individual molecules, identify them and see how they interact. He expects the result to be "a totally new understanding" of the "dynamic structural biology of the living cell."
Rives said Holger Müller at Berkeley imagined the device "many decades ago." Berkeley traces it to a 2010 proposal by Müller and Robert Glaeser, which makes the idea about 16 years old. Biohub's announcement says the institute began supporting Müller's work in 2021, and both institutions announced results on June 11, 2026. Berkeley's design uses a single laser cavity. Biohub's version crosses two perpendicular laser beams, each running at roughly half the power the single-cavity system needs. That reduces heating of the components and optical distortions.
So far the demonstrations have used isolated proteins, not molecules inside cells. Biohub reports a larger improvement for small hemoglobin particles than for larger aldolase particles, which conventional cryo-electron microscopy already images fairly well. Cellular cryo-electron tomography, the in-cell imaging Rives described, is the planned next step, along with further instrument development and data collection.
From one mutation to a whole fish
Labenz then asked about scale. Cells, tissues, organ systems and whole organisms are separated by many orders of magnitude, he noted. Because everything in the body is connected, perhaps one random cell carries enough information to predict a person's overall health. Or perhaps it does not. He asked whether researchers already have the techniques to bridge those scales or will need new ones, such as microbiopsies.
Rives called it "a multi-scale problem." Complexity at the molecular level can shape events at the scale of the whole body, he said, and a single point mutation in a gene can cause disease. He called it "very much a technology problem" and said that measuring an entire system down to the molecular level will not be possible in the near future. For now, researchers can look at tissue and at cells within tissue, an in-between level he called the mesoscale.
That affects what Rives means by a model of the cell. Biologists often study cells in isolation, grown in lab cultures as stand-ins for cells in the body. A complete model, he said, would have to capture the cell in its living context, because living systems are created by interactions between cells. Cultured cells, including transformed ones, can be very different from cells in tissue. But they still share "that core machinery and those core programs," so researchers can learn a great deal from them, even through a "poor proxy" like a culture system. They can learn more by seeing cells in tissues and systems.
Biohub's imaging institute is building tools for each level. The laser phase plate opens an atomic-resolution view, but only for part of a cell; "it's not going to give you the picture for an entire tissue," Rives said. A middle tier of technologies captures fields of cells. Researchers can perturb those cells and then measure their responses statistically across many cells in culture. At the top is whole-organism imaging of zebrafish, which are transparent, so every cell can be seen.
Rives said Biohub's microscopes can follow a developing zebrafish embryo and track every single cell as it grows from one cell into a fully developed organism. Researchers can also label particular cell types, such as all the immune cells, and follow them in a living fish as they move and interact with each other and with the surrounding tissue. Biohub's open Zebrahub atlas, from its Royer Group, shows what this kind of data looks like. It combines single-cell RNA sequencing, which reads the genes each cell is using, with live light-sheet imaging and reconstructions of which cells descended from which. Its sequencing time course covers ten developmental stages, from ten hours after fertilization to ten-day-old larvae. The resource also shares instrument designs, software and lab protocols.
Rives said it is possible to gather information at each of these scales. What remains, in his account, is to build a model that can integrate across "the hierarchies of biological complexity."