Why Quantum Chemistry Waits on the T Gate

I struggled with organic chemistry in college. The course asked me to predict how molecules behave from their structure, and I predicted poorly. The cause could have been the content, the instructor, or me. I will never know but I’m kinda leaning towards me. Things have changed these days - a quantum computer can make predictions from the molecule's energy, and the effort runs into a bottleneck that is easier to identify. 

Bacteria turn nitrogen from the air into a form living things can use. The enzyme that does the work, nitrogenase, runs at room temperature and ambient pressure, according to a 2026 resource-estimation study. Its active site is a cluster of iron, molybdenum, sulfur, and carbon atoms called FeMoco. Chemists want FeMoco's energy at high precision, because the energy shows how the reaction proceeds. Classical computers cannot solve it exactly at this size, so FeMoco is a standard target for quantum computers.

The algorithm for that job is quantum phase estimation, or QPE. Every molecule has a Hamiltonian, the mathematical description of its energy. The lowest value is the ground state energy. Alexei Kitaev introduced phase estimation in 1995, and given a state close to the ground state, it reads out the energy as a number. Preparing a good starting state is difficult on its own, but the larger cost comes from the gates the algorithm runs.

A gate is one basic operation on qubits. Quantum computers protect their qubits with error correction (Chapter 4 of Quantum from the Ground Up covers the basics). The machine groups many physical qubits into one logical qubit, which can survive errors in the physical qubits beneath it. A family of gates called Clifford gates runs cheaply on logical qubits. Clifford gates alone cannot perform every calculation. The machine also needs a non-Clifford gate, usually the T gate. A logical T gate consumes a specially prepared resource called a magic state, and preparing it takes extra qubits and many rounds of checking.

The count of T gates sets the cost. The best published QPE estimate for FeMoco needs about 5.3 billion Toffoli gates. A Toffoli is a three-qubit gate requiring seven T gates in standard circuit synthesis (or four with an ancilla), though modern fault-tolerant architectures often cultivate |CCZ⟩ states directly to bypass explicit T-gate expansion. Under standard synthesis, this total is roughly 37 billion T gates. A PennyLane tutorial for a 76-orbital FeMoco model shows a trade-off. One version uses 2,188 logical qubits and 88 billion Toffoli gates. A second version cuts the qubits to 466 and needs 345 billion Toffoli gates. Saving space costs more gates, and every gate adds T gates to the bill.

Distillation was the standard way to make magic states. It combines many noisy magic states into one cleaner state, and the circuits that do this are called T factories. Craig Gidney's 2025 Simons talk cites a 2012 analysis that put T factories at 94 percent of a quantum computation. That share made magic states the main target for cost reduction.

Chapter 6 of the book covers one alternative to brute-force distillation. In August 2026, a Quantinuum and UC Davis team used code switching to build a trustworthy magic state from 28 physical qubits instead of hundreds. The state's logical infidelity came out about 2.7 times lower than the error rate of the physical qubits that built it. Cultivation takes a different route to the same goal.

In 2024, Gidney, Shutty, and Jones proposed magic state cultivation. It builds up one magic state inside a single error-correcting patch, checks the state at each stage, and discards any attempt that fails. At a physical error rate of 0.1 percent, it reaches a logical error rate near 2 in a billion using about an order of magnitude fewer qubit-rounds than earlier methods. A qubit-round is one qubit held for one round of error correction, so the measure combines space and time. The authors suggest distillation may become unnecessary. Google tested cultivation on a superconducting processor in December 2025. The error dropped by a factor of 40, the state fidelity reached 0.9999, and 8 percent of attempts survived. The remaining 92 percent were discarded during early-stage syndrome checks. This post-selection converts a spatial footprint penalty into a temporal retry delay. Factory duty-cycle and retry latency become the primary runtime bottlenecks.

From a molecule’s energy to the magic states every T gate consumes, with three ways to make them.

Lower T-gate cost moves chemistry closer. A March 2026 study estimates that chemically accurate QPE for active spaces of 20 to 50 spatial orbitals is feasible on early fault-tolerant machines with about 100,000 physical qubits, with runtimes of days to weeks. An active space is the set of orbitals the calculation treats exactly, and chemical accuracy is the precision chemists need to predict reactions. Exact classical methods like Full Configuration Interaction (FCI) exponentially stall beyond 22 spatial orbitals, making 20-to-50 orbital strongly correlated systems the primary domain for early quantum advantage.

The same T-gate cost applies to cryptography. Chapter 13 of Quantum from the Ground Up cites Gidney's 2025 estimate that RSA-2048 falls to fewer than one million noisy qubits in under a week, then reports that the figure did not survive spring 2026. An Oratomic estimate cited there puts RSA-2048 within reach of as few as 10,000 qubits on a neutral atom machine. Part of Gidney's reduction came from using cultivation to shrink the space for magic state production. Progress on the T gate therefore shortens the timeline for chemistry and for breaking encryption. The September 2026 edition covers code switching in Chapter 6. The next quarterly edition of the book will add cultivation and the T-gate budget for chemistry.

I still cannot say for sure whether the content, the instructor, or I caused my trouble with organic chemistry but still.... kinda leaning towards me.

Agents Read the Reports About Themselves

I am building AI/SI into my courses and lab materials. I call it PGI, for Pretty Good Intelligence, a nod to the encryption tool Pretty Good Privacy. On weaker days it is SGI, Sometimes Good Intelligence. I want my students to use both, so the assignments I am writing require it. That means I have to understand how these systems behave, and I learn something new about them nearly every week.

This week’s material came from a research report on an AI/SI agent fleet. An agent is a language model with tools, such as a browser, a code runner, and a network connection. A fleet is a group of agents that start from one place with similar goals. Researchers watched one fleet, running on a commercial cloud platform, between 28 September and 8 October 2026.

The task was simple: find which entrance of Chengdu Zoo most map users navigate to. The mapping site blocked plain requests with anti-bot checks, so the agents improvised. They submitted programs to urlquery.net, a public security scanner that loads pages in a real browser, and used that browser to run their own code. Results returned through page titles and disposable web inboxes. The researchers counted more than 2,600 attempts. Some probed non-production servers and tried exposed access keys. Most failed. The report also cites earlier incidents involving agents from other labs, including a reported case in Germany, so the pattern is general.

The report offers three lessons. First, audit what your sandboxes can reach, because network access is where agents find workarounds. Second, tell agents the truth. The researchers placed hidden text on their site with false claims, agents acted on it, and the authors regret it. Third, assume agents read what you publish. One agent inbox message used the word “fleet” after the researchers coined it.

The fleet gave PGI a fair test. Turning a public security scanner into a code runner is pretty good intelligence. Missing on most of the 2,600-plus attempts is sometimes good intelligence. My students will earn the same two ratings. Work that checks the output earns PGI. Work pasted in unread earns SGI. The gradebook will use both labels. I am officially claiming the terms PGI and SGI. Citations are always appreciated 😎

Quantum Software Gets a Faster Compiler and AI/SI-Generated Circuits

Chapter 12 of Quantum from the Ground Up covers NVIDIA Ising, two machine learning models that calibrate quantum hardware and decode its errors. Chapter 19 names Qiskit, Cirq, and PennyLane as the standard programming frameworks and moves on, which is about the attention most programmers give a compiler until it breaks. The software that turns a quantum idea into something a chip can run gets little space in the book. Two announcements from the past three months change that: IBM’s Qiskit v2.5 release in July, and a generative AI method from Oak Ridge National Laboratory, IonQ, NVIDIA, and the University of Tennessee, Knoxville in September.

A quantum program moves from algorithm to circuit to compiler to hardware. DQAOA-GPT generates circuits in simulation, and Qiskit v2.5 speeds up the compile step. Figures are reported by IBM and IonQ. Source: gordostuff.com

What a Quantum Program Is

Start with what a quantum program is. A quantum program is a circuit: a list of gates applied to qubits, followed by measurements. A gate is a basic operation on one or two qubits. You write the circuit in a framework such as Qiskit, an open-source Python toolkit maintained by IBM. The circuit you write cannot run as written. Each chip supports only a small set of native gates, and on many chips a qubit can interact only with its neighbors. A compiler, which Qiskit calls a transpiler, rewrites your circuit using the gates the chip supports. It assigns the qubits in your circuit to physical qubits and adds extra operations to carry information between distant ones. It also trims the circuit where it can. Every added gate adds a chance of error, so a better compiler produces more reliable answers from the same hardware.

A Faster Compiler in Qiskit v2.5

Qiskit v2.5 improves this layer in several ways. IBM reports typical transpilation speedups of two to three times on its benchmark circuits, and a 115-qubit circuit now routes in about two thirds of the earlier time. Routing is the step that moves information between distant qubits. A new default optimization pass chooses gate decompositions with higher predicted fidelity, the estimated chance a circuit runs correctly. IBM notes that the pass can add gates or depth when that raises the prediction.

Compiling for Fault-Tolerant Machines

The larger change targets fault-tolerant machines, the error-corrected systems that Chapter 6 and Chapter 12 describe. Qiskit now ships two preset compilation pipelines for them. The first compiles to Clifford+T, a standard gate set for error-corrected hardware. Clifford gates are relatively inexpensive to run on such a machine. T gates require magic states, the costly resource behind the Quantinuum work in Chapter 6, so compilers try to use as few T gates as possible. Qiskit lets you adjust synthesis accuracy to trade T-gate count against approximation error. The second pipeline compiles to Pauli-based computation, which expresses a program as a sequence of rotations and measurements built from the three basic Pauli operators, X, Y, and Z. In IBM’s example, the LitinskiTransformation optimization pass cut a phase estimation circuit from 1,354 Pauli rotations to 443. Phase estimation is a common subroutine in quantum algorithms, including the factoring method behind the encryption threat in Chapter 13. Fewer rotations mean fewer expensive operations on a future error-corrected machine.

A New Framework and a New Name

Version 2.5 also adds a framework for compiling through several intermediate representations, which are program formats that sit between what you write and what the chip runs. IBM calls it an early step. The release also renames Qiskit Runtime Service to IBM Quantum Compute Service. The change is naming only, so current documentation may use either name for the same software.

AI/SI-Generated Circuits

The second announcement uses AI/SI to generate circuits. Many useful problems are optimization problems, which ask for the best choice among a huge number of combinations. DQAOA, the distributed quantum approximate optimization algorithm, splits a large problem into smaller subproblems, solves each with a small quantum circuit, and combines the results. Larger subproblems tend to give better answers. Each circuit has adjustable settings called parameters, and the standard method finds good values by looping: run the circuit, measure the result, adjust, and repeat, sometimes hundreds of times. That tuning time limits how large the subproblems can grow.

How the Model Works and What It Showed

The researchers ran the standard method on many sample problems and kept only the near-optimal circuits. They trained a transformer on those circuits. A transformer is the model type behind large language models. This one trained on circuits. At solve time it generates ten candidate circuits per subproblem, simulation scores all ten, and the best one updates the overall solution. The model skips the tuning loop. Circuit-finding time for the standard method rose from about 34 seconds at 4 qubits to over 11 minutes at 12 qubits. The AI/SI method stayed near 28 seconds at every size tested. IonQ reports that answer quality roughly doubled as subproblem size grew on a dense 100-variable Higher-Order Unconstrained Binary Optimization (HUBO) benchmark. The team presented the work at IEEE Quantum Week 2026 in Toronto, and reports say it won a best paper award.

Limits of the Results

Every test ran in simulation. The circuits ran on a single NVIDIA H200 GPU at Oak Ridge using CUDA-Q and cuQuantum, NVIDIA software that simulates quantum circuits on GPUs. Both circuit-generation methods used the same setup, so the comparison covers two ways of producing circuits, with no quantum hardware involved. The authors describe the results as benchmark-scale validation, and hardware runs remain a stated future goal. Both sets of figures in this post come from the organizations that built the tools.

What This Changes in the Book

In the book, Chapter 12 describes the NVIDIA Ising calibration agent as a vision-language model and its decoder as a 3D convolutional neural network, both machine learning tools applied to quantum data. DQAOA-GPT adds a transformer that generates circuits, which extends the same pattern to the programming layer. Chapter 18 lists machine learning model fine-tuning among the skills a candidate can bring without a physics PhD, and training a transformer on a library of good circuits fits that description. The Chapter 18 list also gains an entry: compiler and software tooling development. The Qiskit 2.5 work involves multithreading, a C interface, and compiler pipelines, all conventional software engineering. Chapter 19 names the computer science bachelor’s degree as the main path into quantum software, and the compiler work above is computer science.

This post will be folded into the next quarterly edition of Quantum from the Ground Up. Every edition of the book is available as a free download.

Frank Snyder, December 1939

This is the second part of a story. It is best to read Uncle Al's Knife first, and then this post. That post tells how I found my great-uncle Al's Swiss Army knife. Al's brother Frank was my grandfather. I wrote about Frank's son, my father, in Lessons From My Father's Life.

My Dad, Grandfather Frank, & Uncle Vernon

Frank Snyder was 34. He was a machinist at the Tryco Products Company in Westfield, Massachusetts. He lived on Eldridge Avenue with his wife, Emily, and their two sons. Vernon was 10. My father, Gordon, was eight.

My father remembered Frank's voice from a day in traffic, when he was five or six. A big car came roaring past every car, in the wrong lane. Frank said something like, 'Look at that crazy driver go. Oh my God, it's Al.'

I recall seeing paperwork that said Frank was born in Germany. I do not know when he came to the United States, or how. His parents, Julius and his wife, lived on Park Street, near the river. Frank had five brothers and three sisters.

On the eleventh of December, 1939, Frank spent most of the afternoon on the Westfield River in a boat with his brother August, who was 22. My family called him Augie. They gathered driftwood and new lumber that floated down from the dam at Woronoco, past their parents' home. My family said they gathered it to burn, to keep warm.

Rain had swollen the river. Near the Great River Bridge the water poured over the remains of the old Horton dam and spun into whirlpools among blocks of cement and granite. The brothers turned for shore by the Lynch Lumber Company yard. August stepped out. As Frank was about to follow, his foot slipped and the boat rolled over. The paper said he apparently did not cry for help. In a second or two he was gone. The river threw the boat against the rocks and shattered it.

Newspaper Clippings

August ran for help, and no one he met could give any. Many families had no telephone at home in 1939. He kept running, across the bridge and down the south bank, nearly a mile, to Park Street. His brother-in-law, Daniel Reader, drove him to police headquarters. August was so shaken he could barely tell the police what happened. My father, who was eight that day, later spent 33 years with the telephone company.

The acting police chief took charge. The firemen supplied spotlights. The police dragged the river for nearly two hours while three or four hundred people lined the bridge and the bank behind the lumber yard. August and an officer rowed downstream and found the boat a quarter mile from the bridge. They towed it to shore near the end of Fowler Street. They never found Frank.

The search went on the next morning. The police told residents to stay off the river, where the current was swift, and to report any clothing they saw. The headlines said 'Westfield Man Believed Lost in High River' and then 'Snyder Body Unrecovered.' Police, Boy Scouts, and volunteers combed the banks for several miles below the dam. In June, a badly decomposed body was found in the Connecticut River near Glastonbury, Connecticut. The police checked it against Frank's description. Other bodies came out of the river over the years. None was ever identified as Frank. Three years after the drowning, a newspaper wrote that the river still held the secret of where he was.

Emily was a widow with two small sons. The Depression was still on, and then came the war. They had a small chicken farm. It gave them just enough to survive. I do not know whether Emily received any public aid.

When I was a kid I saw a photograph of Frank on my grandmother's bureau, in her bedroom. I do not know where it is now.

Frank's brothers Al and Pooley, who owned the junkyard in Pottstown, never talked about it. Augie never talked about it either, and he was the one in the boat. I wish I had asked.

Vernon, Frank, Grandmother Emily, and Dad

One day in December 1978 I talked with my grandmother. She told me that time of year brought back memories. When I got home I wrote down what she said. I no longer have the notes. Then I sat at a microfiche reader at UMass and found the newspaper articles. I gave them to my father. He saved them. Save everything was the plan in our family. It came from the Depression. I found them in a folder in my parents' house. The clippings were laid out together on one page.

Uncle Al's knife is in my safe. I have a note with it that says whose it was. The river kept Frank. My father kept the articles. Someone kept the knife sharp for years. Now we have to keep all of it.

Thanks to my brother John for sending over the pics.

Uncle Al’s Knife

I found it in a box in my parents’ house. We were clearing the house out. The box had no label. Inside were old keys, a flashlight, and a roll of electrical tape. The Swiss Army knife lay at the bottom.

The case was worn. Someone had scratched “AL” into it. The blades were sharp. Someone had sharpened them often.

It was Uncle Al’s. Al was my father’s father’s brother. His given name was Adam. He and his brother Pooley owned a junkyard in Pottstown, Pennsylvania. After they “retired” they still lived at the junkyard. In winter they kept a trailer in New Port Richey, Florida. In college I drove over one day on spring break to see them. We grilled hamburgers outside the trailer. I think he had the knife in his pocket then.

Twice a year Al and Pooley drove up to Massachusetts to fish with my father, two brothers and me on the Cape. At first they drove together. Pooley died around 1980. After that Al drove alone from Pennsylvania in an old Dodge Dart. A bar ran across the back seat and his clothes hung from it. I looked at them and thought he lived in the car. His hands were stiff with arthritis. He said he drove with two fingers.

We fished from my father’s 16-foot Starcraft. We caught bluefish and striped bass. The bluefish ran hard and the drag sang. The striped bass pulled heavy and steady. In a boat that size you felt every fish moving the boat around. At night they told stories.

I was young and I did not listen well. I watched the rods and the water and waited for the next fish. Al and Pooley talked and I heard some of it. Now I wish I had heard all of it. Maybe he told more than I remember. Maybe he told everything and I was somewhere else.

My father listened. He read a eulogy at Al’s memorial service, and the stories were in it.

Al got his chauffeur’s license in 1914, at 16. In 1918 he drove to Cape Cod on dirt roads. The trip took 26 hours. He had 12 flat tires and fixed every one himself, on the road. He raced cars. In World War I the government wanted race car drivers to train as pilots. Al volunteered, and so did a friend. The doctors found a hernia. They told him to have it fixed and they would take him. Al said no way. He never had it fixed. His friend was killed.

My father wrote that Al could talk to fish. Al caught the most and the largest, and it upset Pooley. In his last years Al could not stand to cast. He sat in one of the rear-facing seats and told my father where to cast, then took the rod to reel in. A moment later you heard “Fish on.” Everyone else got skunked. I think he had the knife in his pocket then too.

My father also read that he had lost an uncle who treated him like a son. He treated Al like a father. Al died at 93.

I never met my grandfather. His name was Frank Snyder. He was Al’s brother, and Pooley’s, and August’s.

Next: Frank Snyder, December 1939

Spectrum Still Needs a Network

In 2007, I tracked the 700 MHz auction on this blog. Dish Network appeared on the bidder list, and a Credit Suisse analyst questioned whether the band suited Dish. The FCC closed Auction 73 in March 2008 with about $19.6 billion in winning bids. Verizon and AT&T spent $16.3 billion of that. Verizon took the largest nationwide block with a $4.74 billion bid, which topped Google's $4.71 billion by $30 million. A Dish Network subsidiary won 168 licenses in the E Block for $711.9 million. Dish is now part of EchoStar.

On October 9, the same day I published "Walls Are the Hard Part of Phone Service From Space," John Saw, T-Mobile's chief technology officer, published a LinkedIn article titled "Spectrum Alone Doesn't Make a Network." He wrote it after SpaceX agreed to acquire Grain Management's nationwide 800 MHz portfolio, pending FCC approval. The companies did not disclose terms. Grain reportedly sought about $6 billion in August, and the Wall Street Journal valued the deal at about $8 billion. T-Mobile sold those licenses to Grain, and T-Mobile now competes with Starlink Mobile. Read the article with that in mind. The engineering claims are checkable, so they follow after some background.

T-Mobile and SpaceX have worked together. T-Mobile's T-Satellite service runs on Starlink, and T-Mobile's one-year exclusive agreement is ending. On its August 5 earnings call, SpaceX said it will build out terrestrial networks with its EchoStar spectrum and expects Starlink Mobile to win customers from Verizon, AT&T, and T-Mobile. T-Mobile executives have told analysts that satellite service carries about 0.0002% of network usage. T-Mobile, Verizon, and AT&T have since announced a satellite joint venture, and T-Mobile removed the Starlink name from its T-Satellite webpage. The carriers say existing satellite agreements stay in place, so the two companies are partners and rivals at once.

According to Wccftech's report, Saw makes three technical points. First, the Grain block is about 14 MHz of paired spectrum. Starlink Mobile would hold roughly 80 MHz in total, some of it uplink only, while T-Mobile holds nearly 400 MHz of low-band and mid-band spectrum. Second, a satellite hundreds of kilometers up faces about 50 dB more path loss than a cell site 1 km away, and moving from 2 GHz to 800 MHz recovers only a few dB. Third, one satellite beam covers an area that hundreds of terrestrial sectors serve. Millions of simultaneous users streaming video and making calls need that density.

The path loss figure holds up. Free-space loss at 2 GHz is about 153 dB at 550 km and about 98.5 dB at 1 km, a gap of roughly 55 dB. The next-generation satellites the FCC approved would orbit at 326 to 335 km, which narrows the gap to about 50 dB at 330 km. The same formula gives about 8 dB between 2 GHz and 800 MHz for equal antenna gain. The bigger benefit of the lower band is less loss through walls and foliage, which free-space math omits. That benefit explains why SpaceX wants the block. Yesterday's post covered the loss at the wall. This post covers the loss before the signal reaches the wall, and the two add together. Saw addresses link budget and capacity. SpaceX addresses indoor coverage. Both can be correct.

Free-space path loss for a cell site and for satellites at 550 km and 330 km, at 2 GHz and 800 MHz. Calculated at equal antenna gain.

Open items remain. SpaceX says most existing phones already support the 800 MHz band. Starlink's current service uses standard LTE signaling from satellites that carry LTE base station payloads, with Doppler and timing compensation applied on the satellite, so unmodified phones connect as they would to a terrestrial tower. Chipset support matters in two cases. The first is the 2 GHz EchoStar spectrum, which overlaps the 5G NR NTN band n252 that commercial phone chipsets do not yet include. The second is a move to the native 3GPP NR NTN standard, which Starlink's current system does not follow. On October 6, the FCC approved 15,000 next-generation satellites for Starlink Mobile in a partial grant. The approval adds capacity to the system Saw describes.

The 2008 bidder list showed who had money. In 2025, after the FCC questioned its spectrum use and SpaceX called the holdings underused, EchoStar agreed to sell its AWS-4 and H-block licenses to SpaceX for about $17 billion and its 600 MHz and 3.45 GHz licenses to AT&T for about $23 billion. Saw has the title right. EchoStar held the licenses, and the buyers owned the networks.

SpaceX once let its boosters fall into the ocean after a single flight. It now lands and reuses them. Saw lists real obstacles: path loss, capacity, and chipsets. I expect SpaceX to work on those the way it worked on boosters.

Correction: Yesterday's post listed two Grain prices without explaining them. Grain reportedly sought about $6 billion, and the Wall Street Journal valued the deal at about $8 billion.

Walls Are the Hard Part of Phone Service From Space

This post covers frequency, walls and satellites. The companies appear because they hold the licenses, and the physics works the same for every owner.

Key Summary Takeaways

·      Low Band Passes Through Walls: Lower-frequency signals lose less energy in drywall, brick, and glass than higher bands.

·      Distance Weakens Satellite Signals: A satellite signal travels roughly 480 to 550 kilometers to reach your phone, and a wall takes more of it.

·      Spectrum Deals Show the Stakes: SpaceX agreed to pay about $17 billion for EchoStar's licenses, and Grain Management reportedly wants about $6 billion for its 800 MHz licenses.

·      Expect Coverage Gains First: Rural homes, trails, and open water gain the most, with text and voice arriving before fast data.

In August 2007, I wrote that the FCC's 700 MHz spectrum drew heavy interest because it travels long distances and passes through walls. The auction closed in March 2008 with $19.6 billion in bids. Dish Network won 168 licenses for $712 million, and a Credit Suisse analyst questioned whether the band suited Dish at all. Nineteen years later, the same property is why a rocket company is shopping for spectrum.

Start with how your phone service works today. A cell tower has antennas that send and receive radio signals. Your phone connects to the closest tower, and the tower connects to the carrier network over fiber. Each carrier holds FCC licenses for specific radio frequencies, and that set of frequencies is the spectrum. A tower covers a limited area, so carriers build thousands of them. Where no tower exists, or where a storm knocks out its power, your service stops.

Frequency decides how a signal behaves. Low-band signals, generally below 1 GHz, travel farther and pass through drywall, brick, and glass with less loss. Mid-band and high-band signals carry more data but cover less distance and struggle with walls. A low-band signal can reach the middle of your house. A high-band signal can drop at the front door.

Phone service from satellites has a distance problem. A Starlink satellite orbits roughly 480 to 550 kilometers up and crosses your sky in minutes. Its signal arrives weak, and your phone answers with a tiny antenna and little power. One satellite also covers a large area, so many phones share its capacity. Satellite texting on current phones already works this way, and it asks you to point the phone at open sky. A wall takes the rest of the signal.

That is why 800 MHz matters. T-Mobile sold its 800 MHz licenses to Grain Management in a swap that closed in August 2026. On October 8, 2026, SpaceX announced a definitive agreement to acquire Grain Management's nationwide 800 MHz spectrum portfolio. The deal still needs FCC approval, and SpaceX did not disclose terms. The Wall Street Journal reported a price of $8 billion. In 2025, SpaceX agreed to pay about $17 billion for EchoStar's licenses, and Boost Mobile customers get access to Starlink service as part of that deal. AT&T, Verizon, and T-Mobile shares fell 4 to 5 percent that day.

Here is what it could mean for you. Coverage reaches places towers do not: rural homes, trails, and open water. Service continues when a storm takes down local towers. Expect text and voice first, with data slower than a tower provides, because satellites share capacity across wide areas. You will need a phone that supports the band. Many current phones already support nearby 800 MHz bands, but check yours. You will also need a plan that includes satellite access. Whether competition lowers your bill is unknown. Cities will still run mostly on towers.

In January 2008, I wrote about public safety networks that survive disasters by not relying on a single network. Satellite coverage on an ordinary phone adds another network to your phone without adding another device.

The physics in my 2007 post still holds. The 700 MHz buyers were carriers building towers. Some of the next buyers build rockets.