Video yükleniyor...

Video Yüklenemedi

Ana Sayfaya Dön

Lumentum CEO Michael Hurlston explains the laser math behind AI compute: Every pluggable light source needs 8 lasers, each rack holds around a thousand light sources, so a single rack needs ~8,000 lasers, which adds up to hundreds of millions across the industry. "For scale-up optics, the idea from...

23,167 görüntüleme • 11 gün önce •via X (Twitter)

0 Yorum

Yorum bulunmuyor

Orijinal gönderinin yorumları burada görünecek

Benzer Videolar

NEW: Lumentum (NASDAQ: $LITE ) CEO Michael Hurlston on the Optics Boom How Lasers Are Transforming AI Data Centers "We're being asked by hyperscalers to deploy in millions, tens of millions of units." "You're not gonna have birds burning up as these lasers shoot data down to the Earth, right?" "Copper can't carry these signals the distances they need to go. You have more & more optics now coming into the data center." We cover: › Why AI is replacing copper with light › Birds vs. space lasers › Space lasers & the future of the internet › Why we're spending $10T moving data from kilometers to millimeters › Lumentum's 3x revenue growth in just 5 quarters › Stock went up 1000% › U.S.-China geopolitics in optical networking › The manufacturing bottlenecks facing optical infrastructure Michael Hurlston became President & CEO of Lumentum (LITE) on February 7, 2025. Fifteen months later the company reported record quarterly revenue of $808.4M, up 90% year over year, joined the Nasdaq-100, & took a $2B investment from Nvidia. 𝐓𝐈𝐌𝐄𝐒𝐓𝐀𝐌𝐏𝐒 (00:00) Michael Hurlston, CEO at Lumentum Holdings (00:58) Backstage with Tony Kim at RAISE Summit (01:53) What optical connectivity actually means (03:31) The $10 trillion irony of the data center build-out (05:37) Scaling up inside a data center (08:38) 3x revenue in 5 quarters (10:13) From Finisar to Synaptics to Lumentum (11:51) Why Lumentum went all-in on data centers (13:10) Why space is the next big market for optical lasers (14:06) Debunking laser myths (space vs birds) (16:40) Why copper is about to disappear from the server rack (19:00) The geopolitics of the optical supply chain (20:16) The advice that shaped Michael's path

Molly O’Shea

152,337 görüntüleme • 10 gün önce

Free Electron Lasers(FEL) look like the future of EUV lithography. This is how most EUV light is made today: A powerful laser hits tiny molten tin droplets thousands of times per second. The droplets turn into plasma and release the extreme ultraviolet light used to print the smallest features on advanced chips. It works, but it comes with clear limits. The process creates debris that contaminates the mirrors, the usable power is restricted, and the overall energy efficiency is low. A Free Electron Laser(FEL) takes a completely different approach. It accelerates electrons to nearly the speed of light and sends them through a series of alternating magnets called an undulator. As the electrons move through the magnetic field they emit clean, coherent EUV light. No tin. No plasma debris. Higher power output. Better efficiency. Narrower spectrum. And in the right design, one source can feed multiple lithography machines at the same time. Those advantages add up. Cleaner light means less downtime and lower maintenance. Higher power means faster throughput and fewer stochastic defects. Better efficiency means lower operating costs at scale. The ability to supply several scanners from a single source changes the economics of a large fab. China is also exploring this route. Research groups there have published designs and studies on energy-recovery Free Electron Lasers specifically for EUV lithography, seeing it as a way to move beyond the limits of current tin-based sources. This is why many see Free Electron Lasers as one of the real long-term paths for next-generation chipmaking. The current method got the industry here. FEL looks like the technology that can take it further.

Joe Hansen

71,021 görüntüleme • 1 gün önce

In our last conversation, Gavin said data centers in space will be the most important thing in 3-4 years. He explains that means "racks in space" and thinks orbital compute will solve the watts shortage: "When people hear data centers in space, they picture a Pentagon-sized building in space. That's not what it is. A Blackwell rack weighs 3,000 pounds. It's eight feet high. Four feet deep. Three feet wide. It's racks in space. It has these solar wings that are probably 500 feet long on each side. You keep it in a Sun-synchronous orbit, so those solar panels are always in the sun. And then because it's in an exactly Sun-synchronous orbit, the radiator, which extends behind it for hundreds of feet is in the shade. You link these racks using lasers traveling through vacuum which are already on every Starlink. SpaceX operates the world's largest satellite fleet, which is 98 or 99% of all satellites in orbit. Every Starlink, they're cooling it today. I think Starlink V3 is going to operate at 20 kilowatts. A Blackwell rack is only 100 kilowatts. And people talk a lot about density. Well, if you're connecting the racks with lasers through vacuum, you can make the rack bigger physically. In space, there's all sorts of things that SpaceX can do. They also now operate the largest data center on Earth. I've spent a lot of time at Starbase over the years, and I've talked to a lot of SpaceX engineers. It is the most talented group of engineers on planet Earth, and they're very confident they have solved this."

Patrick OShaughnessy

268,098 görüntüleme • 2 ay önce

Universities teach many erroneous models. One example of a misleading teaching is the Newtonian model of the prism. A man named Johann Goethe corrected and complimented Newton's model about 100 years after Newton. It is not the same light passing through the prism and distributing. It is a brand-new light with each beam of color. The primary light hitting the prism is extinguished by the electrons making up the glass of the prism itself. Those same electrons re-emit a brand new light from each angle. That process happens at the rate of c. Serving the illusion that the original white light came it... separated and exited to continue on. Most people rely on outdated text books or the Pink Floyd album cover "Dark Side of the Moon" to get their information about light and color and prisms. The depictions of light "splitting" and "bending" through a prism based on their color is false. (See first comments for attached images and substantiation) I filmed these videos playing around with prisms, lasers, magnifying glasses, mirrors, slits, and colored filters. I've gained my own perspectives on light and color over the years from real world, hands-on experimentation and observation. Most people have never held a real glass prism their own hands and done the tests to see for themselves. 2 prisms for $10 Harmonics of Light and Sound: This guy said his units were confiscated by the FBI and he was threatened not to pursue manufacturing them. The unit is simply a light with a magnifying glass and colored filters to isolate the light. Like burning an ant with a focal point... but with a specific color. The lens can spin and create a stroboscopic effect in an isolated band. The inventor claimed it healed stuff. Dinshah Spectro-Chrome Machine - Tour and Explanation Magenta monochromatic ray of shadow - Pehr Sahlstrom: Color and Physics, Newton versus Goethe - Pehr Sahlstrom: Thomas Joseph Brown Color Theory: Clay Taylor Color Theory: Testing the "Rightness" of relativity. The impossibility of proving something through experiments. Laser and Mirror observation: It's never the same light. Re-emission Example: I put this interactive page together going back and forth with ChatGPT to code using Javascript. Here is a video playing with the slide controls to show 3 different sinewaves (3 different lights/emissions). The packets are ALWAYS in phase from medium to medium. The frequency shifts BECAUSE the velocity of light shifts. NOT a Doppler shift of "the same light" distorting within the same frame of reference. It's a brand new light at each point. The incoming light is a different frame of reference. All the light you ever see is the re-emitted light from the electrons making up yourself. Propagation and Re-Emission of Light: Dr. Edward Dowdye Slides - LOTS of math and context. E.H. Dowdye, Can Stars BEND LIGHT? General Relativity and Gravity with Dr. Edward Dowdye Challenges to Gravitational Lensing and More: 71 Part Video Series (Each vid is 1 min - 1 and a hafl min) The Rebirth of Classical Physics - Time, Light & Gravity Classical Physics vs Relativity - History, Examples and Alternatives: Illusions of Relativity: Space-Time vs Real-Time The Rebirth of Classical Physics: Time, Light & Gravity - Article Packed with Info & Links Thought Experiment for Light and Absolute Time. (This scenario got me banned from talking on a Podcast.) "Too off the rails."

TheRealVerbz (Jason Verbelli)

44,361 görüntüleme • 1 yıl önce

The speed of light is too slow. That is the buried physics problem inside Elon Musk's plan for 1 million AI satellites, and his solution is a constellation that refuses to think together. The world pictures the same machine, a million Starlink satellites wired into one giant brain in the sky. Physics vetoes it on contact. Inside a single ground rack, AI chips exchange 260 terabytes a second across 2 meters of copper and synchronize in millionths of a second. Between satellites hundreds of kilometers apart, light itself needs a full millisecond to cross the gap, a thousand times too slow. The fastest thing that exists cannot keep up with the conversation inside one server rack. A computer cannot be stretched across the sky. Think about it. Maybe Elon will agree. So SpaceX collapsed the computer into the satellite. The spacecraft is the rack. Each AI1 carries a complete data center rack's worth of compute, 120 kilowatts on average, on wings spanning 70 meters, wider than a Boeing 747 jet. The spec Elon posted today pushes further, a full NVIDIA Rubin rack in every Starlink satellite, 72 GPUs, 250 kilowatts peak. Nothing important ever leaves the machine. The weights, the memory, the model itself, all stay home. The laser links between Starlink satellites carry questions and finished answers, never the thinking in between. For 20 odd years computing moved in one direction, toward the cloud, thousands of machines pretending to be one. Orbit just reversed it. A million computers pretending to be nothing but themselves. The moat is not beating the speed of light. If you think about it it is all about building the machine so the most valuable data never enters the race. Two prototypes fly in early 2027. The rocket that has to carry them flies tonight. The piece works out whether the sky becomes one mind or a million. My money is on Elon!

Shanaka Anslem Perera ⚡

56,918 görüntüleme • 23 gün önce