<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Nimish Sakalle on Medium]]></title>
        <description><![CDATA[Stories by Nimish Sakalle on Medium]]></description>
        <link>https://medium.com/@nimish.sakalle?source=rss-a7dfe155599a------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*1rAf0WjnegMAXvDsZ_1gVw.png</url>
            <title>Stories by Nimish Sakalle on Medium</title>
            <link>https://medium.com/@nimish.sakalle?source=rss-a7dfe155599a------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 17:34:40 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@nimish.sakalle/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[How Atoms and Light Make a Quantum Computer Work]]></title>
            <link>https://medium.com/@nimish.sakalle/how-atoms-and-light-make-a-quantum-computer-work-edcc08eb0d5e?source=rss-a7dfe155599a------2</link>
            <guid isPermaLink="false">https://medium.com/p/edcc08eb0d5e</guid>
            <category><![CDATA[quantum-physics]]></category>
            <category><![CDATA[quantum-computing]]></category>
            <category><![CDATA[emerging-technology]]></category>
            <category><![CDATA[qubit]]></category>
            <category><![CDATA[atom]]></category>
            <dc:creator><![CDATA[Nimish Sakalle]]></dc:creator>
            <pubDate>Tue, 10 Nov 2026 14:18:56 GMT</pubDate>
            <atom:updated>2026-11-10T14:18:56.663Z</atom:updated>
            <content:encoded><![CDATA[<p>A guide to qubits and quantum computation through the physical things happening inside the machine</p><p><em>by Nimish Sakalle</em></p><p>Picture a single atom held in place inside a vacuum chamber. A carefully timed pulse of light changes its state. More pulses follow. At the end, a detector records whether the atom shines brightly or stays dark. That pattern of light becomes part of a computer’s answer.</p><p>How does that become a calculation? This is the question I want to explore: what physically happens between giving a quantum computer a problem and receiving a result? We can follow the process without starting with equations, using a trapped-ion computer as our example. Other kinds of quantum computer use different hardware, but share the underlying quantum rules.</p><h4>Start with an atom and two states</h4><p>An ion is an atom with an electric charge. In our example, it has lost an electron, allowing electric fields to hold it in a trap. Within the atom, electrons occupy allowed quantum states with particular energies. Think of steps on a staircase: the available energy levels are separated, rather than forming a continuous ramp.</p><p>Choose two suitable, long-lived internal states and label them 0 and 1. These labels are our way of representing information. The atom, controlled within those two states, supplies a quantum bit, or qubit. A qubit can also be built from a superconducting circuit or encoded in a photon; it does not have to be an atom.</p><p>Light can connect atomic states because its oscillating electric field interacts with the atom’s electric charges. A photon is one quantum of light: a discrete amount of energy whose size depends on the light’s frequency. An atom can absorb a photon of a suitable frequency and move to a higher energy state. It can also lose energy by emitting a photon. Matching the energy gap matters, although the transition must also be physically allowed.</p><p>Controlling a qubit requires more than making an atom absorb light. A laser provides a field with a well-defined rhythm that can drive a controlled change between states. Some qubits use a direct optical transition; others use two laser frequencies together, or microwaves. The particular arrangement depends on the chosen states. <a href="https://arxiv.org/abs/quant-ph/9710025">[1]</a></p><h4>A pulse can prepare a superposition</h4><p>In the simplest two-state model, a suitable laser pulse can take an atom initially in 0 into 1. A pulse half as long, with the same strength, prepares an equal superposition instead. If we immediately measure 0 or 1, either outcome has a 50% chance.</p><p>The atom has not acquired a new energy level halfway up the staircase. Its quantum state now combines the two allowed states in a way that later pulses can still control.</p><p>A dial is a useful picture for that control. Pulse duration and strength determine how far the state turns; the phase of the light helps determine the direction of the turn. The atom is not physically rotating. This picture of a qubit as an arrow on a sphere is called the Bloch sphere. The arrow’s height sets the chances of 0 and 1; its direction around the sphere carries the relative phase. Under a steady resonant drive, the probabilities cycle between 0 and 1. This is called Rabi oscillation. [1, 2]</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/944/1*_BAJJ6SscnK-q4GHswKVMQ.png" /><figcaption>In this simplified model, the arrow represents the qubit’s state. The labels π/2 and π mean a half-flip pulse and a full-flip pulse. The halfway arrow represents an equal superposition, not a third energy level.</figcaption></figure><p>A quantum gate is a precisely controlled operation that changes a qubit’s state, or the joint state of several qubits. It is one instruction in a quantum calculation. In our trapped-ion example, the hardware carries out these instructions through carefully arranged control pulses and interactions between ions. The gate describes what change we want; the pulses make it happen.</p><p>The Hadamard gate, written H, acts on one qubit. Starting from 0, it prepares an equal superposition of 0 and 1: either result has a 50% chance if measured immediately. Starting from 1, it also gives equal chances, but with a different relative phase. That phase difference matters when later gates make amplitudes interfere. H is useful for preparing possibilities that a calculation can work with. [2, 12]</p><p>For a simple example, start in 0 and apply H twice without measuring in between. The first H prepares the superposition; the second returns the qubit to 0. This predictable return shows why H is more than a random switch: it changes the state in a controlled, reversible way. [2]</p><h4>Interference makes that relationship useful</h4><p>Think of two ripples meeting on water. Where their crests line up, they reinforce. Where a crest meets a trough, they can cancel. Their relative position in the wave cycle is called phase.</p><p>Quantum possibilities have something similar. Each has an amplitude, which can be pictured as a small arrow with a length and a direction. When indistinguishable alternatives lead to the same measurement outcome, their amplitudes combine. Arrows pointing together reinforce; opposing arrows can cancel. This is quantum interference. The probability of an outcome is the square of the resulting arrow’s length. <a href="https://quantum.cloud.ibm.com/learning/en/courses/basics-of-quantum-information/single-systems/quantum-information">[2]</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/944/1*Uf_ZW3SJhiE1g2z_NjsQuQ.png" /><figcaption>For mutually exclusive classical alternatives, probabilities add. For coherent quantum alternatives leading to the same outcome, amplitudes combine first. These arrows illustrate reinforcement and cancellation; they are not paths travelled by an atom.</figcaption></figure><p>This explains why two preparations can give identical 50–50 results when measured immediately, yet behave differently after another pulse. Their relative phases differ. A later operation can turn that phase difference into different measurement outcomes.</p><p>An algorithm uses controlled changes of amplitude and phase to make useful results more likely. Designing that sequence is the difficult part. Quantum mechanics does not automatically recognise a correct answer, and interference does not guarantee that every problem becomes faster to solve.</p><h4>Entanglement joins qubits into a shared state</h4><p>The controlled-NOT gate, called CNOT or CX, acts on two qubits. One is the control; the other is the target. For inputs of 0 or 1, the rule is: leave the target alone when the control is 0, and flip the target when the control is 1. “Flip” means changing 0 to 1 or 1 to 0. The gate applies this rule without measuring the control, so it also works coherently when the control is in superposition. This lets CNOT create entanglement from suitable inputs. [3, 12]</p><p>Here is how H and CNOT work together. We write the two qubits as a pair of digits, A first and B second: “00” means both are 0, and “10” means A is 1 while B is 0. Each qubit still holds only one digit.</p><ol><li><strong>Start. </strong>Both qubits are 0, so the pair is 00.</li></ol><p>2. <strong>Apply H to A. </strong>A is now in an equal superposition of 0 and 1, while B stays 0. The pair has equal amplitudes for 00 and 10.</p><p>3. <strong>Apply CNOT, with A as the control and B as the target. </strong>Where A is 0 (the 00 part), nothing happens. Where A is 1 (the 10 part), B flips, so 10 becomes 11. The gate acts on both parts at once, so the superposition is preserved: the pair is now an equal superposition of 00 and 11.</p><p>The result is a Bell state, specifically the phi-plus Bell state: an equal superposition of 00 and 11 with the same phase. Measuring both qubits as 0 or 1 gives 00 or 11, each half the time. Each individual result is unpredictable. Hadamard prepares the superposition; CNOT turns it into entanglement. [3, 12]</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/944/1*e5-qMjmhBBzBUX73FV2b3A.png" /><figcaption>Preparing 00, applying H to A, then CNOT from A to B produces the Bell state shown here. Matching cards can imitate its 0-or-1 results. Tests with different measurement settings reveal the quantum difference across many fresh pairs. [3, 12]</figcaption></figure><p>Matching results alone are not proof of entanglement. Two envelopes containing matching cards could do that. The difference emerges when we change how the qubits are measured. Across carefully chosen settings, entangled states can produce correlations that no locally predetermined set of answers can reproduce. The correlations depend on those settings; the outcomes do not always match.</p><p>The pair needs a joint quantum description. Trying to give each qubit a separate, complete description misses something essential. That shared state lets operations on several qubits act on relationships between their possible outcomes. <a href="https://quantum.cloud.ibm.com/learning/en/courses/basics-of-quantum-information/multiple-systems/quantum-information">[3]</a></p><p>In a trapped-ion processor, lasers can couple the qubits through their shared motion. A sequence of these interactions and single-qubit rotations can implement a CNOT gate. A gate name describes the intended operation; it need not correspond to just one laser pulse. [1, 4] Separating entangled qubits does not create instant messaging: neither side can choose its random result to send a signal.</p><h4>How a calculation happens inside the machine</h4><p>The program specifies which gates to apply and in what order. Control electronics turn those instructions into timed signals for the hardware. Single-qubit gates such as H and two-qubit gates such as CNOT are building blocks for longer sequences. The calculation happens through the physical changes they produce in the joint quantum state.</p><p>Grover’s algorithm gives us a small example we can follow. Consider four candidates, labelled 00, 01, 10 and 11, with exactly one satisfying a test. Knowing how to test a candidate is different from knowing which candidate passes: checking whether a key fits a lock is easier than knowing which key to choose. The example below marks 11.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/944/1*9NjYBEVJLqEzftn3o4w8xQ.png" /><figcaption>Grover’s algorithm on two qubits. Hadamard gates spread the amplitudes, the oracle marks 11 by reversing its phase, and diffusion makes the unwanted amplitudes cancel. Measurement produces readable bits, shown here using bright = 1. [1, 5, 13]</figcaption></figure><p>First, prepare two qubits in 00 and apply a Hadamard gate to each. This creates equal amplitudes for all four candidates. Measuring now would simply give a random candidate. These two independent Hadamard gates create superposition; they do not yet entangle the qubits.</p><p>Next, apply a quantum version of the test, called an oracle. It reverses the phase of the candidate that passes, without measuring which candidate that is. In the picture, the arrow for 11 turns down while the other three keep pointing up. We must build the test into the program; the computer is not guessing what we mean by a good answer.</p><p>Then comes the diffusion step, another sequence of gates that turns the phase marking into interference. In this example, it reflects each amplitude about their average, like moving each arrow’s tip to the other side of a mirror placed at that average. The three unwanted amplitudes become zero, while the marked amplitude grows to one. A final measurement therefore returns 11 with certainty in this ideal four-candidate example. Real devices introduce errors, so laboratory results fall short of the ideal. [5, 13]</p><p>What does “measurement” look like? In a common trapped-ion method, a readout laser makes one qubit state scatter many photons while the other remains comparatively dark. A detector distinguishes bright from dark and the electronics record a 0 or 1. The laser can access additional atomic levels for this readout; the stored qubit still uses the two states we chose. <a href="https://arxiv.org/abs/quant-ph/9710025">[1]</a></p><p>A run produces ordinary bits, not a list of every possibility in the superposition. Experiments usually repeat the preparation, gates and measurement to estimate outcome probabilities. For larger unstructured searches, Grover’s method reduces the number of tests from roughly proportional to the number of candidates to proportional to its square root. Building the test and controlling the hardware still take work. <a href="https://arxiv.org/abs/quant-ph/9605043">[6]</a></p><h4>Light can also connect separate processors</h4><p>Adding more qubits to one device makes precise control harder. One alternative is to connect smaller processors so that they can cooperate. Here, light takes on a different job: individual photons establish connections between qubits.</p><p>In one method, an ion in each processor emits a photon entangled with its internal state. The photons travel through fibre to a beam splitter, an optical component that mixes their paths. When the photons are otherwise indistinguishable, a suitable pattern of detector clicks can announce that the distant ions have become entangled.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/944/1*sx37rUtn_WxIhdyELMHZEw.png" /><figcaption>A possible optical link between processors. The purple dashed line represents shared entanglement, not a physical wire. Successful detection creates a resource that local operations and ordinary communication can use.</figcaption></figure><p>Local operations, measurements and ordinary messages can then use this shared entanglement to perform a gate across the processors. This is called a teleported gate: the operation acts across the modules while the atoms stay in their own machines. Classical communication remains necessary. [7]</p><p>In work published in 2025, Dougal Main and colleagues at Oxford demonstrated this with trapped-ion modules about two metres apart. Their remote gate had approximately 86% fidelity, a measure of agreement with the intended operation. They also ran the four-candidate Grover search across the modules, obtaining the correct answer about 71% of the time. These are different measures: gate quality and the success rate of the complete search. <a href="https://doi.org/10.1038/s41586-024-08404-x">[7]</a></p><h4>What still makes it difficult</h4><p>The atom must keep the phase relationships that later operations need. Uncontrolled interactions with the environment can damage them, a process called decoherence. Noise in the control signals can also make a pulse turn the state by the wrong amount. A longer calculation creates more opportunities for errors.</p><p>Networks add further problems. An atom can emit into many directions, so collecting a photon into one fibre is difficult. Cavities, structures that confine light, can favour emission into a wave pattern that couples efficiently into the fibre. Some photons also need their frequency converted to a band that travels through fibre with less loss. <a href="https://doi.org/10.1038/nature07127">[8]</a></p><p>Even good telecom fibre transmits only about 1% of photons over 100 km if its loss is 0.2 decibels per kilometre, before other losses. Ordinary amplification cannot restore arbitrary unknown quantum states perfectly. The no-cloning theorem rules out making perfect copies of an unknown state. Quantum repeater schemes instead establish shorter entangled links and connect them, often using memories to hold successful links while others are prepared. <a href="https://doi.org/10.1126/science.aam9288">[9]</a> <a href="https://doi.org/10.1038/299802a0">[10]</a></p><p>A 2024 experiment by Can Knaut and colleagues connected quantum memories through a 35-km fibre loop in the Boston area. The loop returned to the laboratory; it did not join computers 35 km apart. It nevertheless tested the interfaces under deployed-network conditions. <a href="https://doi.org/10.1038/s41586-024-07252-z">[11]</a></p><p>For me, this makes the practical questions clearer. How reliably can we control a qubit? How long can it preserve the relationships a calculation needs? How often do processors establish entanglement compared with the speed of their local gates, and how accurate are the remote operations? Following the atom, the pulses and the detector turns these questions into something concrete. It connects the promise of quantum computing to the physical work still required to make it useful.</p><h3>Sources and further reading</h3><p><em>[1] Wineland, D. J. et al. (1998). Experimental issues in coherent quantum-state manipulation of trapped atomic ions. Journal of Research of NIST, 103, 259. Atomic control, Rabi oscillations and fluorescence readout. </em><a href="https://arxiv.org/abs/quant-ph/9710025"><em>Read source</em></a></p><p><em>[2] Watrous, J. Basics of quantum information: Single systems — Quantum information. IBM Quantum Learning. States, amplitudes and measurement. </em><a href="https://quantum.cloud.ibm.com/learning/en/courses/basics-of-quantum-information/single-systems/quantum-information"><em>Read source</em></a></p><p><em>[3] Watrous, J. Basics of quantum information: Multiple systems — Quantum information. IBM Quantum Learning. Joint states and entanglement. </em><a href="https://quantum.cloud.ibm.com/learning/en/courses/basics-of-quantum-information/multiple-systems/quantum-information"><em>Read source</em></a></p><p><em>[4] NIST (2009). NIST Demonstrates ‘Universal’ Programmable Quantum Processor for Quantum Computers. Physical implementation of gates using trapped ions. </em><a href="https://www.nist.gov/news-events/news/2009/11/nist-demonstrates-universal-programmable-quantum-processor-quantum"><em>Read source</em></a></p><p><em>[5] IBM Quantum. Grover’s algorithm. A practical explanation of the oracle and amplitude amplification. </em><a href="https://quantum.cloud.ibm.com/docs/en/tutorials/grovers-algorithm"><em>Read source</em></a></p><p><em>[6] Grover, L. K. (1996). A fast quantum mechanical algorithm for database search. Proceedings of STOC 1996, 212–219. </em><a href="https://arxiv.org/abs/quant-ph/9605043"><em>Read source</em></a></p><p><em>[7] Main, D. et al. (2025). Distributed quantum computing across an optical network link. Nature, 638, 383–388. </em><a href="https://doi.org/10.1038/s41586-024-08404-x"><em>Read source</em></a></p><p><em>[8] Kimble, H. J. (2008). The quantum internet. Nature, 453, 1023–1030. </em><a href="https://doi.org/10.1038/nature07127"><em>Read source</em></a></p><p><em>[9] Wehner, S., Elkouss, D. and Hanson, R. (2018). Quantum internet: A vision for the road ahead. Science, 362, eaam9288. </em><a href="https://doi.org/10.1126/science.aam9288"><em>Read source</em></a></p><p><em>[10] Wootters, W. K. and Zurek, W. H. (1982). A single quantum cannot be cloned. Nature, 299, 802–803. </em><a href="https://doi.org/10.1038/299802a0"><em>Read source</em></a></p><p><em>[11] Knaut, C. M. et al. (2024). Entanglement of nanophotonic quantum memory nodes in a telecom network. Nature, 629, 573–578. </em><a href="https://doi.org/10.1038/s41586-024-07252-z"><em>Read source</em></a></p><p><em>[12] IBM Quantum Learning. Build and run your first quantum program. Hadamard and CNOT gates preparing a Bell state. </em><a href="https://quantum.cloud.ibm.com/learning/en/courses/use-a-qc-today/build-and-run-your-first-quantum-program"><em>Read source</em></a></p><p><em>[13] IBM Quantum Documentation. Grover operator. Phase marking and diffusion as reflection about the mean. </em><a href="https://quantum.cloud.ibm.com/docs/en/api/qiskit/qiskit.circuit.library.grover_operator"><em>Read source</em></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=edcc08eb0d5e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How Policymakers and Entrepreneurs Can Develop the US Quantum Industry]]></title>
            <link>https://medium.com/@nimish.sakalle/how-policymakers-and-entrepreneurs-can-develop-the-us-quantum-industry-a1420392d9c7?source=rss-a7dfe155599a------2</link>
            <guid isPermaLink="false">https://medium.com/p/a1420392d9c7</guid>
            <category><![CDATA[public-policy]]></category>
            <category><![CDATA[quantum-initiative]]></category>
            <category><![CDATA[quantum-computing]]></category>
            <category><![CDATA[enterpreneurship]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[Nimish Sakalle]]></dc:creator>
            <pubDate>Sun, 04 Oct 2026 19:55:38 GMT</pubDate>
            <atom:updated>2026-10-04T19:55:38.699Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*MIKaD9H61asLamvu1EwREA.png" /></figure><h4>The science is proven. The money is moving. The next few years decide who leads</h4><p>Quantum theory started in the 1920s. Niels Bohr, Werner Heisenberg, Erwin Schrödinger, and a small group of physicists built it to explain experiments that made no sense any other way. For most of the century since, it stayed mostly theoretical — interesting physics, not much else.</p><p>That’s changed. The same physics now powers un-hackable communication, faster drug discovery, and sharper imaging. It’s becoming a real industry, and the US is racing to lead it, fund it, and staff it.</p><p>This piece looks at that race: where the US stands, what just changed in the last few weeks, and what policymakers and entrepreneurs need to get right from here.</p><h4>Four pillars, one hard engineering problem</h4><p>Quantum technology breaks into four areas: computing, communication, sensing, and simulation. All four run on the same strange physics.</p><p><strong>Superposition</strong> means a quantum system can hold multiple states at once. A classical bit is either 0 or 1. A qubit can be 0, 1, or a mix of both. <strong>Entanglement</strong> means two quantum objects can stay linked, so measuring one instantly tells you something about the other, no matter how far apart they are. Together, these let a quantum computer represent far more information per qubit than a classical computer can per bit.</p><p>That power comes at a cost. Quantum information is hard to prepare, store, and read without disturbing it. The systems run at the atomic scale, where the tolerance for noise is close to zero. This is the core engineering challenge, and it’s why progress here looks less like one big breakthrough and more like a decade-long grind — across hardware, error correction, and materials science. As you’ll see below, that grind just had a very eventful few weeks.</p><h4>Where the US has stood: bet on the science first</h4><p>The federal push started with the National Quantum Initiative Act in 2018. It set up a whole-of-government strategy for quantum leadership. Between 2018 and 2022, the Subcommittee on Quantum Information Science committed $1.2 billion, with more flowing through the Department of Energy, the National Science Foundation, and the Department of Defense. Other countries did the same — heavy public investment paired with private capital, in China and India, across Europe, and in the US.</p><p>The core idea behind that early strategy was “science-first”: quantum information science is still young, and most applications are still prototypes, so the priority should be steady investment in basic research, not a rush to sell products. It’s an easy idea to state and a hard one to fund consistently, because it asks policymakers to keep paying for research whose payoff isn’t obvious yet.</p><p>That bet just paid off, publicly. The 2025 Nobel Prize in Physics went to John Clarke, Michel Devoret, and John Martinis for experiments they ran in 1984 and 1985, discovering quantum tunnelling in an electrical circuit. Forty years later, that same physics is the foundation for the superconducting quantum chips Google, IBM, and others are building today. Science-first isn’t just a slogan. It’s the actual pipeline this industry runs on.</p><p>The numbers back this up too. Quantum patent filings grew five-fold from 2014 to 2024. Quantum computing patents specifically jumped more than 300% between 2016 and 2021. Corporations and universities hold 91% of them between them. China leads with 60% of quantum patents as of 2024, ahead of the US and Japan — a reminder that funding the science isn’t enough on its own if the resulting ideas end up patented somewhere else.</p><h4>What just changed: June and July 2026</h4><p>Here’s where this stops being a history lesson. In the last few weeks, US quantum policy moved from long-term strategy to funded, dated commitments — on several fronts at once.</p><p><strong>Congress is extending the National Quantum Initiative.</strong> Both the House and Senate have moved their own versions of a reauthorization bill. The House Science Committee approved its version (H.R. 8462) in an April markup. The Senate Commerce Committee passed its version (S.3597), which would extend the initiative five more years, to December 2034. The two bills still need to be merged and signed into law, but the direction is bipartisan, and the funding window just got much longer.</p><p><strong>The White House signed a dedicated quantum executive order.</strong> On June 22, 2026, the administration signed “Ushering in the Next Frontier of Quantum Innovation,” alongside a companion order on post-quantum cryptography. The order updates the National Quantum Strategy, starts a national effort to build a quantum computer powerful enough to drive real scientific discovery, and calls for quantum sensors and networks to be deployed within five years. It also creates National Quantum Workforce Development Institutes, expands apprenticeships, restarts the National Quantum Initiative Advisory Committee, and adds a new counterintelligence team focused on protecting quantum supply chains.</p><p><strong>Commerce put $2 billion behind nine companies, by name.</strong> In May 2026, the Department of Commerce signed letters of intent for $2.013 billion in CHIPS Act incentives across nine quantum companies. The details matter more than the total. Two companies got money specifically to build domestic manufacturing capacity: GlobalFoundries ($375 million), for a foundry covering multiple qubit types, and <strong>IBM ($1 billion)</strong>, to build a new quantum foundry on top of its existing US lead in superconducting quantum wafer manufacturing. Seven more companies — Atom Computing, Diraq, D-Wave, Infleqtion, PsiQuantum, Quantinuum, and Rigetti — got smaller, targeted grants (mostly around $100 million each) aimed at specific unsolved engineering problems, spread across five different qubit approaches. In exchange, Commerce takes a small equity stake in each company. That’s not a research grant. That’s industrial policy, spread deliberately across competing approaches instead of picking one winner.</p><p><strong>Agencies set a hard deadline.</strong> At a White House Quantum Summit, federal agencies pledged over $2.2 billion toward building a fault-tolerant quantum computer by 2028. The Defense Innovation Unit added up to $200 million for quantum sensing on top of that. Paired with the Genesis Mission — a separate executive order from November 2025 using AI to speed up scientific discovery, quantum included — the pattern is consistent: fewer strategy documents, more actual deadlines.</p><p>Taken together, this is the biggest run of US quantum policy activity since the original 2018 Act. And it matches the theory this industry has run on for years: public money reduces the risk on early science, private capital scales what works.</p><h4>What this means if you’re building in this space</h4><p>Three things follow from the last few weeks.</p><p>First, look at where the CHIPS money went. It tells you what the government thinks is still broken: getting chips to behave consistently, cutting error rates, keeping systems cold enough to work, building the control and readout hardware that talks to the qubits, and cutting signal loss in light-based (photonic) systems. If your work touches any of those problems, there’s now a real federal on-ramp — nine signed letters of intent, not just a strategy paper.</p><p>Second, watch the supporting supply chain. The June executive order specifically calls out support for the unglamorous stuff every quantum computer depends on — cryogenic coolers, control electronics, specialized materials, packaging. That layer has historically gotten less attention than the headline-grabbing “how many qubits” race. It’s a real opening for smaller, focused companies that don’t need to build a whole quantum computer to matter.</p><p>Third, the workforce pipeline is being built in public now, not left to individual companies. The new workforce institutes and apprenticeship programs matter because the field needs people at every level — deep specialists in physics and engineering, but also a much larger group who just need enough training to work in supporting roles: manufacturing, systems engineering, sales, product. Get involved early and you get a hiring edge. Wait, and you’re competing with everyone else for the same small pool of PhDs.</p><h4>What this means for policymakers</h4><p>Don’t drop the science-first approach just because things are moving faster now. The 2025 Nobel Prize is proof that patient, curiosity-driven research pays off decades later. The bigger risk now is the opposite mistake: funding the science and the factories, but failing to coordinate the parts that tie it all together.</p><p>Three things matter most from here. Merge and pass the reauthorization bill quickly — the House and Senate versions are close enough that delay, not disagreement, is the real risk, and a funding window through 2034 gives companies and universities the kind of long-term certainty they need to plan multi-year programs. Fund the new workforce institutes for real, not just on paper — training programs take years to build, and the industry’s own hiring projections (250,000 new jobs by 2030, 840,000 by 2035, per The Quantum Insider) will outrun the pipeline if the institutes are announced but never properly staffed or funded. And keep spreading hardware funding across different qubit approaches — superconducting, trapped-ion, neutral-atom, photonic, silicon-spin — rather than picking one now. No single approach has a clear, lasting lead yet, so this is exactly the wrong moment to bet on just one.</p><h4>The stakes</h4><p>The Quantum Insider projects quantum computing could add $1 trillion in economic value by 2035. The uses aren’t hypothetical anymore: simulating molecules and proteins for drug discovery, catching fraud and optimizing portfolios in finance, quantum navigation for defense where GPS doesn’t reach, and modeling materials that classical computers simply can’t handle. Meanwhile, China holds 60% of global quantum patents. The question isn’t whether quantum technology will matter — the physics and the market both already answered that. The real question is whether the country that funded the underlying science for a century actually captures the economic payoff.</p><p>The last two months say the US just decided the answer has to be yes, and started backing that decision with real money. Whether it holds up through 2035 comes down to unglamorous execution: merging two bills in Congress, actually staffing workforce institutes that don’t exist yet, and making sure nine companies with signed letters of intent deliver on genuinely hard engineering problems. Science-first got the country this far. Execution is what happens next.</p><p><em>I’m learning and working in quantum computing at IBM, one of the companies named above. I’ve tried to keep the analysis independent of that — every source below is public, so you can check the numbers yourself.</em></p><h4>Sources &amp; further reading</h4><ul><li><a href="https://www.congress.gov/bill/119th-congress/senate-bill/3597/text">National Quantum Initiative Reauthorization Act — S.3597 text (Congress.gov)</a></li><li><a href="https://www.congress.gov/bill/119th-congress/house-bill/8462">National Quantum Initiative Reauthorization Act — H.R.8462 (Congress.gov)</a></li><li><a href="https://www.whitehouse.gov/fact-sheets/2026/06/fact-sheet-president-donald-j-trump-ushers-in-the-next-frontier-of-quantum-innovation/">The White House: Fact Sheet — Ushering in the Next Frontier of Quantum Innovation (June 2026)</a></li><li><a href="https://www.nist.gov/news-events/news/2026/05/department-commerce-announces-letters-intent-9-companies-2-billion">NIST/Department of Commerce: Letters of Intent with 9 Companies for $2 Billion in Quantum Incentives (May 2026)</a></li><li><a href="https://thequbitreport.com/security-policy/2026/07/08/white-house-quantum-summit-details-over-2-2-billion-in-support-for-2028-fault-tolerant-computing/">White House Quantum Summit: $2.2B toward fault-tolerant computing by 2028</a></li><li><a href="https://www.nobelprize.org/prizes/physics/2025/press-release/">Nobel Prize in Physics 2025 — press release</a></li><li><a href="https://qir.mit.edu/patents/">MIT Quantum Index Report — patent landscape</a></li><li><a href="https://thequantuminsider.com/2024/09/13/the-quantum-insider-projects-1-trillion-in-economic-impact-from-quantum-computing-by-2035/">The Quantum Insider: $1 trillion economic impact and job projections by 2035</a></li><li><a href="https://www.quantum.gov/wp-content/uploads/2022/02/QIST-Natl-Workforce-Plan.pdf">Quantum.gov: QIST National Workforce Plan</a></li><li><a href="https://www.lockheedmartin.com/en-us/news/features/2025/unlocking-the-power-of-quantum-navigation-lockheed-martin-and-partners-awarded-contract.html">Lockheed Martin: Quantum navigation system (QuINS)</a></li><li><a href="https://www.ibm.com/quantum/blog/large-scale-ftqc">IBM: path to fault-tolerant quantum computing</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a1420392d9c7" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Building a data pipeline with IBM Agentic Data Integration]]></title>
            <link>https://medium.com/@nimish.sakalle/building-a-data-pipeline-with-ibm-agentic-data-integration-1eb330f230be?source=rss-a7dfe155599a------2</link>
            <guid isPermaLink="false">https://medium.com/p/1eb330f230be</guid>
            <category><![CDATA[agentic-etl]]></category>
            <category><![CDATA[watsonx]]></category>
            <category><![CDATA[agentic-data-pipeline]]></category>
            <category><![CDATA[data-integration]]></category>
            <category><![CDATA[data-pipeline]]></category>
            <dc:creator><![CDATA[Nimish Sakalle]]></dc:creator>
            <pubDate>Sun, 04 Oct 2026 16:25:29 GMT</pubDate>
            <atom:updated>2026-10-04T16:25:29.900Z</atom:updated>
            <content:encoded><![CDATA[<p><em>I described what I wanted. Twenty-eight seconds later there was a scheduled job.</em></p><p>Most finance teams I work with have the same problem, and it is not the modelling or the reporting. It is the part before that. Pulling data out of four systems, lining it up, cleaning it, and getting to one table everyone agrees on.</p><p>The work itself is not hard. It is joins, filters and deduplication. But it sits inside a tool that takes real skill to drive. So the person who understands the numbers is usually not the person who can build the pipeline. They raise a ticket and wait.</p><p>I spent a few weeks building a demo to test a different idea. Can the person who needs the data just describe it and get a working pipeline back? Not a script, not a notebook. A scheduled, supported job. Here is what I found.</p><h4><strong>The setup</strong></h4><p>Four sources. I picked them to look like something a tax or finance team would actually recognize, not a toy dataset.</p><p>• A <strong>trial balance</strong> — 425 journal lines across three legal entities, three period ends and three currencies.</p><p>• A <strong>chart of accounts</strong> — 66 distinct accounts with their classification, normal balance and rollup group.</p><p>• An <strong>entity master</strong> — ownership on one subsidiary moves from 60% to 75% partway through the year.</p><p>• A <strong>currency rate table</strong> — closing and average rates by period, including one rate that was restated.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/465/1*Simyhvrl-6EVP9IuDVCaQg.png" /><figcaption><em>The project before anything is built. Connections and data assets, nothing else.</em></figcaption></figure><p>The ask was ordinary. Combine them into one table, drop the duplicates, filter to the current period, and have it run again next period without anyone rebuilding it.</p><p>I planted one trap. The chart of accounts arrived with duplicate rows — <strong>77 rows holding 66 accounts</strong>. Leave them in and the totals go up. That is the kind of thing nobody notices in a spreadsheet and everybody notices in a filing.</p><h4><strong>What I actually did</strong></h4><p>I opened a chat and typed this:</p><p><em>“I would like you to create a data pipeline with chart_of_accounts_duplicates, trial_balance and entity_master tables. Remove duplicates from chart_of_accounts_duplicates, join with trial_balance on account id. Filter data for 2026 and then join with entity_master on entity code. Save output file as csv.”</em></p><p>What came back was not a result. It was a plan. Five numbered steps, naming the tables it would read, the keys it would join on, the filter it would apply and the file it would write. Nothing had run.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/465/1*izgP_SKMAyBDpxPSsjFnpw.png" /><figcaption><em>The request, and the plan that came back. Nothing has executed at this point.</em></figcaption></figure><p>That matters more than it sounds. For a finance audience an AI that quietly produces a number is useless, because you cannot defend the number. One that shows you its method and waits is something you can work with. I read the plan, approved it, and then it ran.</p><h4><strong>The numbers, which are the point</strong></h4><p>The row counts told the story as it went. Chart of accounts in at 77, out of the deduplication step at 66. Join to the trial balance, 425 rows. Period filter, 280. Then the entity join, and a CSV written out.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/465/1*NF5wMYeAXDnh3SO_vH77pg.png" /><figcaption><em>The generated flow with live row counts on every link.</em></figcaption></figure><p>Twenty-eight seconds end to end.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/465/1*b037b8Q6SKiJ5rBy-acUuQ.png" /><figcaption><em>Run successful, with the usual metrics and logs kept.</em></figcaption></figure><p>If the deduplication step had been missed, the totals would have been wrong and nothing on screen would have looked off. That is the failure worth designing against, and it is why the review step earns its place.</p><h4><strong>What it generated, and why that matters more than the speed</strong></h4><p>The thing I was most skeptical about going in was what sits underneath. A natural language layer that produces something proprietary is a problem. You cannot hand it to an engineering team, you cannot put it under change control, and nobody is going to support it at three in the morning.</p><p>What it produced was an ordinary ETL flow. The same artifact an engineer would have built by hand on the canvas. It is versioned, governed, monitored and scheduled like anything else in the estate, and an engineer can open it and change it without needing to know it was generated.</p><p>So this is not a parallel system sitting alongside the real one. It is a second front door onto the same engine. A platform team will accept that. They will not accept the alternative.</p><h4><strong>Pushing it further: the agent as a tool, not a destination</strong></h4><p>The chat interface is useful, but you still have to go to the product and open it. The more interesting setup is exposing the same capabilities over the <strong>Model Context Protocol</strong>, so an assistant the business already uses can call them.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/465/1*-6_YSulIhfpjJ3CuAk0Lrw.png" /><figcaption><em>The same capabilities reached from a general-purpose assistant over MCP.</em></figcaption></figure><p>Set up that way, the pipeline tools become ordinary capabilities for a general purpose agent. List the project assets, build a flow, run it, monitor it, schedule it. The user never opens the integration product at all. They ask the assistant they already have, and a governed job turns up where it belongs.</p><h4><strong>What about Unstructured Data</strong></h4><p>The agent generates StreamSets flows as well as DataStage ones, which brings streaming and semi-structured data — JSON, XML, log files — into scope, along with the schema drift handling StreamSets is good at. What it does not do on its own is document extraction. Getting tables out of PDFs is Unstructured Data Integration, which has its own canvas, its own prebuilt modules for text extraction and redaction, and its own prerequisites. In my demo that step was built in advance. So the accurate line is not “structured only”. It is that documents need a separate pipeline in front of the one you ask for.</p><p>It builds pipelines. It does not answer questions about the data. The tools cover the flow and job lifecycle — create, run, monitor, schedule. Querying the resulting table is a different connection entirely. If you are picturing one assistant that both builds the pipeline and analyses the output, that is two integrations, not one</p><h4><strong>Further reading</strong></h4><p>• <a href="https://www.ibm.com/products/watsonx-data-integration/agentic-data-integration">Agentic Data Integration</a> — the product page for the capability this piece is about</p><p>• <a href="https://www.ibm.com/docs/en/watsonx/wdi/saas?topic=integration-integrating-data-agent">Integrating data with an agent</a> — the documentation, including the full list of what the agent can do and the MCP server</p><p>• <a href="https://www.ibm.com/products/watsonx-data-integration">IBM watsonx.data integration</a> — the platform underneath, covering batch, streaming and replication</p><p>• <a href="https://www.ibm.com/products/streamsets">IBM StreamSets</a> — the streaming engine the agent can generate flows for</p><p>• <a href="https://www.ibm.com/docs/SSNFH6_5.2.x/svc-welcome/undata.html">Unstructured Data Integration</a> — the separate capability for getting text and tables out of documents</p><p>• <a href="https://github.com/IBM/ibm-watsonx-data-integration-skills">ibm-watsonx-data-integration-skills</a> — IBM’s repository of skills for wiring an external agent to the MCP server</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1eb330f230be" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Foundation of Quantum Computing — Explained Without the Jargon]]></title>
            <link>https://medium.com/@nimish.sakalle/foundation-of-quantum-computing-explained-without-the-jargon-00b3969c758b?source=rss-a7dfe155599a------2</link>
            <guid isPermaLink="false">https://medium.com/p/00b3969c758b</guid>
            <category><![CDATA[beginners-guide]]></category>
            <category><![CDATA[quantum-physics]]></category>
            <category><![CDATA[quantum-superposition]]></category>
            <category><![CDATA[quantum-computing]]></category>
            <category><![CDATA[quantum-entanglement]]></category>
            <dc:creator><![CDATA[Nimish Sakalle]]></dc:creator>
            <pubDate>Sun, 12 Jul 2026 01:00:10 GMT</pubDate>
            <atom:updated>2026-07-12T01:00:10.441Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/250/1*IEhTjyayIKWw3n0GWpTOJg.jpeg" /><figcaption>IQM Quantum Computer</figcaption></figure><p>A lock with four possible codes, a spinning coin, and a pair of noise-cancelling headphones — good examples for learning how quantum computers actually work</p><p>Every few months, quantum computing has another “breakthrough” moment in the news. A chip with a name like a bird (Willow, Nighthawk, Condor). A headline about being “13,000 times faster” than the world’s best supercomputer. A Nobel Prize for something that happened in a lab in 1984. And somewhere in the piece, a paragraph about superposition and entanglement that reads like it was translated twice, badly, from another language.</p><p>Here’s the thing: we don’t need a degree in math or physics to understand the basic concepts. All we need is one running example and three ideas — so let’s set up the example first.</p><h4>The four-digit lock</h4><p>Imagine a lock with a two-bit code — the possibilities are `00`, `01`, `10`, or `11`. Only one of them opens it. You don’t know which.</p><p>A classical computer solves this the way you’d expect: it tries `00`, checks, tries `01`, checks, and so on. Worst case, four attempts. On average, 2.5.</p><p>A quantum computer solves it differently — not by trying codes faster, but by trying a fundamentally different strategy: loading all four codes into the machine *at once*, then using a trick called interference to make the correct one the most likely thing you see when you finally check. For this particular lock, that trick needs exactly one step.</p><p>That’s the whole pitch of quantum computing in one sentence: it doesn’t just compute faster, it computes differently. Here’s how.</p><h4>Classical bit vs. qubit: what actually changes</h4><p>A classical bit is a light switch. It’s up or it’s down — 0 or 1 — at every instant, full stop.</p><p>A qubit (quantum bit) is more like a dimmer switch that’s spinning. While it’s spinning, it isn’t cleanly “on” or “off” — it’s holding some blend of both. The moment you look at it (measure it), it snaps to one definite answer. That in-between, blended state is called **superposition**, and it’s the first of three ideas that make quantum computers work. The other two are **entanglement** (qubits can be linked so their outcomes are correlated) and **interference** (algorithms can be designed so wrong answers cancel out and right answers get amplified). Everything else in quantum computing is built from these three.</p><h4>Interference: how the answer actually gets found</h4><p>Superposition alone just gives you a mess of possibilities — all four lock codes, sitting there with equal weight, is not an answer. Something has to make the *right* one stand out. That something is interference, and the best analogy is noise-cancelling headphones.</p><p>Noise-cancelling headphones don’t identify background noise and delete it. They generate an *opposing* sound wave that cancels it out through destructive interference. Quantum algorithms are engineered the same way, deliberately, using constructive and destructive interference to cancel out wrong answers and reinforce the right one.</p><p>Here’s the lock, solved, step by step:</p><ol><li>**Mark the answer.** The machine checks all four codes against the lock simultaneously and flips the sign of the correct one’s “weight” from positive to negative. Say the real code is `10`. Three codes are now positive, one is negative — but if you measured right now, you’d still get any of the four at random. Flipping a sign doesn’t change the odds by itself.</li><li>**Amplify around the average.** A second step reflects every weight around the group’s average. Because the marked code is the odd one out — negative while the rest are positive — this reflection pushes its weight sharply up while pulling the other three down.</li><li>**Read the answer.** Check the register. For a four-code lock, this whole mark-then-amplify sequence needs just **one round**, and the correct code comes out essentially every time.</li></ol><p>Compare that to the classical approach — up to four sequential tries, 2.5 on average — against **one step, high confidence**. This exact mark-and-amplify pattern is what’s known as <a href="https://en.wikipedia.org/wiki/Grover%27s_algorithm">Grover’s search algorithm</a>, and it works identically at much larger scale: millions of entries instead of four, just with more rounds of mark-and-amplify.</p><p>Interference isn’t just a party trick for search problems, either. In October 2025, Google’s Quantum Echoes algorithm used the same principle — sending a signal through a quantum system, reversing it, and listening for an “echo” amplified by constructive interference — to extract molecular structure information about 13,000 times faster than the best classical supercomputer could even verify. That’s not a toy example; it’s the same three-step logic above, aimed at real chemistry.</p><h4>Entanglement: the linked coins</h4><p>Now for the third idea. Entanglement is what happens when two or more qubits become linked so that measuring one instantly tells you something about the other — no matter how far apart they are.</p><p>Analogy: imagine two coins magically paired so that whenever one lands heads, the other <em>always</em> lands heads too, even if one is on your desk and the other is on the far side of the planet. Neither coin “decided” its outcome on its own — they’re correlated as a pair.</p><p>Made concrete: entangle two qubits (physicists call this a “Bell pair”), then physically separate them — one with Alice, one with Bob.</p><p>| Alice measures | Bob measures (same instant, no signal sent) |<br>| — -| — -|<br>| 0 | 0 |<br>| 1 | 1 |</p><p>Before either of them measures, neither qubit has a fixed value — it’s genuine superposition. But the moment Alice measures hers and gets, say, 0, Bob’s is *guaranteed* to read 0 too, instantly, even though nothing was sent to it and no one touched it. Ordinary bits can’t do this — you’d have to physically copy a value ahead of time, and a copy can’t be checked for tampering.</p><p>That “checkable for tampering” part is the actual computational payoff. If anyone intercepts or measures an entangled qubit in transit, it disturbs the correlation, and the disturbance is detectable. That’s the basis of quantum key distribution: communication that’s provably eavesdrop-proof, not merely hard to crack.</p><h4>Why a handful of qubits can do so much</h4><p>One bit holds one of two values. One qubit in superposition holds a blend of two. Two bits hold one of four combinations at a time; two qubits hold all four *simultaneously*. In general, n qubits represent 2ⁿ states at once.</p><p>That’s the same exponential curve as the old “grains of rice on a chessboard” story — double the grains on every square, and by square 64 you’re at 18 quintillion grains. Three hundred fully entangled qubits could, in principle, represent more states than there are atoms in the observable universe.</p><p>Worth saying plainly, though: this doesn’t make quantum computers faster at *everything*. They’re a different tool, suited to a specific shape of problem — search, optimization, simulation, factoring — not a general upgrade over classical machines. For most day-to-day computing, your laptop is still the right tool.</p><h4>Where it actually pays off</h4><p>The problem shapes quantum computing is good at happen to sit under some very consequential industries:</p><p>Healthcare and drug discovery — molecular simulation, protein folding, faster candidate screening, because simulating a molecule is itself a quantum problem. Finance — fraud detection, portfolio optimization. National security and navigation — quantum navigation systems, secure communications, underground mapping. Materials and energy — simulating chemistry that’s currently intractable for classical computers, from batteries to catalysts.</p><p>The Quantum Insider projects quantum computing could add $1 trillion in economic value by 2035, and generate roughly 840,000 new jobs by then (250,000 by 2030). Those aren’t small numbers, and they’re a big part of why governments and companies are moving as fast as they are.</p><h4>Where things stand, right now</h4><p>This part is easy to get wrong in either direction — either “it’s all hype” or “it’s already here.” The honest answer: it’s real, it’s moving fast, and it’s still early.</p><p>The physics has been validated the hard way. The 2025 Nobel Prize in Physics went to John Clarke, Michel Devoret, and John Martinis for discovering macroscopic quantum tunneling in an electrical circuit back in 1984 — the same physics that underlies today’s superconducting quantum chips at companies like Google and IBM. Hardware is compounding: qubit counts, gate fidelity, and circuit depth have all been climbing, and multiple companies (Google’s Willow chip, IBM’s Nighthawk processor) hit milestones described as “verifiable quantum advantage” within months of each other in 2025 and 2026. And the market is scaling with it — valued at roughly $3.5 billion in 2025, projected to reach $20 billion by 2030.</p><p>None of that means quantum computers are sitting on your desk next year. It means the gap between “interesting physics” and “useful computer” is closing faster than most people outside the field realize — and the ideas behind it (a spinning coin, a pair of linked coins, and noise-cancelling headphones) are simple enough that anyone can follow along as it happens.</p><p><strong><em>Note* I’m learning and working in AI and quantum computing, and wrote this to explain these ideas without the jargon. Next step: try it yourself — running an actual quantum circuit takes about fifteen minutes with free tools like IBM’s Qiskit. That’s what I’ll cover next, along with some of these concepts in more depth.</em></strong></p><h4>Sources &amp; further reading</h4><ul><li><em>Nobel Prize in Physics 2025 — press release](</em><a href="https://www.nobelprize.org/prizes/physics/2025/press-release/"><em>https://www.nobelprize.org/prizes/physics/2025/press-release/</em></a><em>)</em></li><li><em>[Google: Quantum Echoes algorithm and verifiable quantum advantage](</em><a href="https://blog.google/innovation-and-ai/technology/research/quantum-echoes-willow-verifiable-quantum-advantage/"><em>https://blog.google/innovation-and-ai/technology/research/quantum-echoes-willow-verifiable-quantum-advantage/</em></a><em>)</em></li><li><em>[MarketsandMarkets: Quantum Computing Market forecast](</em><a href="https://www.marketsandmarkets.com/PressReleases/quantum-computing.asp"><em>https://www.marketsandmarkets.com/PressReleases/quantum-computing.asp</em></a><em>)</em></li><li><em>[IBM: path to fault-tolerant quantum computing](</em><a href="https://www.ibm.com/quantum/blog/large-scale-ftqc"><em>https://www.ibm.com/quantum/blog/large-scale-ftqc</em></a><em>)</em></li><li><em>[IBM Quantum: Grover’s algorithm tutorial](</em><a href="https://quantum.cloud.ibm.com/docs/en/tutorials/grovers-algorithm"><em>https://quantum.cloud.ibm.com/docs/en/tutorials/grovers-algorithm</em></a><em>)</em></li><li><em>[Microsoft Quantum: superposition explainer](</em><a href="https://quantum.microsoft.com/en-us/insights/education/concepts/superposition"><em>https://quantum.microsoft.com/en-us/insights/education/concepts/superposition</em></a><em>)</em></li><li><em>[The Quantum Insider: $1 trillion economic impact by 2035](</em><a href="https://thequantuminsider.com/2024/09/13/the-quantum-insider-projects-1-trillion-in-economic-impact-from-quantum-computing-by-2035/"><em>https://thequantuminsider.com/2024/09/13/the-quantum-insider-projects-1-trillion-in-economic-impact-from-quantum-computing-by-2035/</em></a><em>)</em></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=00b3969c758b" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>