Quick Answer: Is RTX 5090 discontinued? NVIDIA says no: SVP Jeff Fisher stated on October 16 that production continues and the card “has definitely not been discontinued.” But street prices near $9,000, “Not Carried” listings at Micro Center, and vanishing stock mean the flagship is effectively unbuyable for gamers. The smart move is an RTX 5080, RX 9070 XT, or RTX 5070 Ti at your price tier instead.
About our earlier coverage Our previous article covered the original October 10 rumor, when leakers claimed NVIDIA had halted RTX 5090 production to reserve GB202 chips for workstation cards. NVIDIA had said nothing at that point. Everything below is new: the official denial, fresh October 19 pricing data, and a tier-by-tier buying guide for what to purchase instead.
 Introduction
On October 10, hardware leakers claimed the RTX 5090 was finished. NVIDIA stayed silent, and the internet filled the silence with panic. Then on October 16, NVIDIA finally spoke: Jeff Fisher, Senior Vice President of the GeForce business, told Taiwanese outlet 4Gamers that the RTX 5090 “has definitely not been discontinued” and that shipping will continue.
That should have ended the story. Instead, the market called NVIDIA’s bluff. In the five weeks since mid-September, flagship 5090 models have climbed from around $6,900 to nearly $9,000 at major marketplaces. Micro Center has reclassified most of its RTX 5090 listings as “Not Carried.” NVIDIA’s own marketplace no longer lists the card. And two pallets of Zotac RTX 5090s, worth about $470,000, reportedly vanished inside a Sao Paulo airport.
“Not discontinued” and “available” are two very different things. A product can stay on the roadmap while being impossible to buy. This article explains what the denial actually covers, why prices say otherwise, and most importantly, what gamers should buy at every price tier right now, with CPU pairings that avoid bottlenecks. A plain-language note: the news keywords around this story (like the Fisher statement and the October price spikes) may have no search volume yet, since the events are only days old.
What NVIDIA Actually SaidÂ
The statement came at the RTX Spark launch event in Taipei on October 16, where Fisher spoke with Taiwanese media including 4Gamers and Taiwan’s Central News Agency. Notebookcheck, Dexerto, Digital Trends, and HotHardware all carried the translated remarks. Here is the key quote, translated from Chinese:
“The supply chain (capacity) is indeed somewhat constrained, but NVIDIA is doing everything it can to get the product on the market. Although there are other variants of the RTX 5090 (chip), production has not been discontinued, and NVIDIA will continue to ship, which is very important.”
Three things in that statement deserve a closer look.
First, the denial is real and on the record. This is the first time NVIDIA has publicly addressed the October 10 rumors, and it directly contradicts claims that new GB202 chip allocations for GeForce cards have ended. If you own a 5090, driver support and warranty coverage were never actually in question, and this statement removes the last doubt about the card being quietly killed.
Second, Fisher admitted the supply chain is constrained. He offered no shipment numbers, no restocking timeline, and no explanation for why the card vanished from NVIDIA’s own marketplace. Some outlets also read his remarks as pushing back on the rumored 24GB RTX 5080, though NVIDIA has not officially confirmed or denied that card. Either way, “we will keep shipping” without volumes or dates is a promise with no deadline.
Third, the mysterious “other variants of the RTX 5090 chip.” Fisher did not explain what those variants are. The most likely reading is that he means the different RTX 5090 products built on the GB202 die, such as the standard card and the China-market RTX 5090 D, but that is speculation. Nobody outside NVIDIA knows for sure.
The honest summary: the RTX 5090 is alive on paper. On shelves, it is a different story.
The Price Charts Tell a Different Story
PC Guide tracked four popular RTX 5090 models across five weeks of the shortage. The numbers below are their reported marketplace prices. The trend is the point: every card either climbed hard or disappeared.
| Model | September 14 | September 29 | October 19 |
|---|---|---|---|
| ASUS ROG Astral RTX 5090 White OC | $6,899.99 | $7,999.00 | $8,979.00 |
| Gigabyte AORUS RTX 5090 Infinity | $6,549.99 | $7,177.68 | $8,599.99 |
| Gigabyte AORUS Xtreme WATERFORCE | $6,899.99 | $7,055.90 | Out of stock |
| ASUS TUF Gaming RTX 5090 | $6,699.99 | $7,399.97 | Out of stock |
That is roughly a 30 percent jump in five weeks on cards already priced at more than triple the $1,999 launch price. Digital Trends found Amazon marketplace listings up to $8,500 and Newegg listings stretching toward $10,000. Tom’s Hardware reported third-party sellers asking as much as $9,500. One limited anniversary bundle, the ASUS ROG Astral Edition 20, was spotted at Micro Center with a $12,000 price tag, six times the launch MSRP.
The retail signals are just as stark. Micro Center switched most RTX 5090 listings to “Not Carried,” which usually means the retailer does not expect to stock the item again, not merely that it is temporarily out. The card also disappeared from NVIDIA’s own marketplace model list. Neither move has been explained.
Who is buying at these prices? Increasingly, not gamers. Multiple outlets report that AI firms are buying 32GB RTX 5090s for server use, since the card offers far more memory bandwidth per dollar than workstation cards. The GeForce RTX 5090 launched as a gaming flagship in January 2025. In October 2026, it behaves like a rationed industrial component. Our GPU shortage coverage explains how memory prices and AI demand created this mess across every tier, not just the flagship.
Practical takeaway: Ignore the word “discontinued” and watch the shelves instead. A card you cannot buy at any sane price is dead to your build, whatever NVIDIA calls it.
A Pallet Heist Shows How Wild This Market Is
On October 19, Tom’s Hardware reported one of the strangest GPU stories of the year. Brazilian distributor Oderço says two pallets of Zotac RTX 5090 graphics cards went missing after arriving at Guarulhos International Airport in Sao Paulo. The shipment was valued at about $470,000, or 2.5 million reais, which was the distributor’s own acquisition cost. Based on local retail prices near 34,000 reais per card, the pallets likely held at least 63 graphics cards.
Oderço, a company operating in Brazil for 38 years, said it had never lost a shipment like this and posted a public appeal after getting no explanation from airport or freight authorities. No official statement has come from the airport operator, police, or Zotac.
Do not read this as an investment tip. Read it as a market signal. When pallets of graphics cards become worth stealing at an airport, you are not looking at a normal retail shortage. You are looking at a product whose street value has detached completely from its launch price. For gamers, that is one more reason to stop chasing the flagship and shop the tiers below it.
What to Buy Instead Around $700
This is the sweet spot for 1440p high-refresh gaming and perfectly good 4K with upscaling. Two cards own this tier.
RX 9070 XT (16GB GDDR6, $599 MSRP). This is the value pick of the current generation. It launched in March 2025 and delivers strong raster performance that handles 1440p at high refresh rates and 4K with FSR upscaling. Its 16GB of memory is the comfortable standard for modern games, and it draws 304W, which keeps power supply requirements reasonable. Our full RX 9070 XT vs RTX 5080 comparison shows exactly where the cheaper card wins and where it gives ground.
RTX 5070 ($549 MSRP) or RTX 5070 Ti ($749 MSRP). The 5070 Ti stretches this budget slightly, but it buys you DLSS 4 with multi frame generation and stronger ray tracing. NVIDIA’s upscaling remains the best in the business, and at 4K that software advantage is worth real frames. The 5070 Ti’s 16GB of GDDR7 and 300W power draw make it an easy fit for most builds.
CPU pairings that avoid bottlenecks: At this tier, a Ryzen 5 9600X or Ryzen 5 7600X feeds either card cleanly at 1440p and 4K. On Intel, a Core i5-14400F or Core i5-14600K does the same job. These chips are strong enough that the GPU stays the limiting factor in demanding games, which is exactly what you want. Competitive esports players chasing 240Hz plus at 1080p may prefer an X3D chip, but for most gamers these mid-range CPUs are the balanced choice. Check compatibility details in our CPU and GPU compatibility guide.
Practical takeaway: If your budget tops out near $700, the 9070 XT is the smart money and the 5070 Ti is the feature pick. Neither needs an expensive CPU to shine.
What to Buy Instead Around $1,000
Welcome to the new high end, at least for cards you can actually buy. The RTX 5080 is now effectively the flagship of the reachable market.
RTX 5080 (16GB GDDR7, $999 MSRP). With 10,752 CUDA cores and 16GB of GDDR7 on a 256-bit bus, it delivers roughly 70 percent of the RTX 5090’s 4K gaming performance for a fraction of the street price chaos. It draws 360W, so a quality 850W power supply handles it comfortably. DLSS 4 and best-in-class ray tracing come standard. Street prices have inflated above MSRP along with everything else, but the card is actually in stock at most retailers, which the 5090 cannot claim.
The used RTX 4090 question. Leftover and used RTX 4090s float around this tier with 24GB of GDDR6X memory. For pure gaming, a used 4090 often matches or beats the 5080 while offering more VRAM. Our RTX 5080 vs RTX 4090 buying guide breaks that decision down completely. The catch is the same as any used flagship purchase: verify the seller, confirm the return policy, and stress-test on arrival. A 4090 also draws 450W, so budget for a stronger power supply and better case airflow.
CPU pairings that avoid bottlenecks: The natural partners are the Ryzen 7 9800X3D or 7800X3D, the gaming chips that leave nothing on the table, or a Core i7-14700K on the Intel side. At 4K, even a Ryzen 5 9600X keeps up fine because the GPU does the heavy lifting, but at 1440p high-refresh rates the X3D chips earn their price. Do not pair a $1,000-class GPU with a five-year-old mid-range CPU and expect clean frame times at 1440p.
Practical takeaway: The 5080 is the card the 5090’s absence promotes to flagship. Buy it for 4K with DLSS 4, or hunt a verified used 4090 if 24GB of VRAM matters to your work.
What to Buy Instead Around $1,500
This tier is for buyers who wanted 5090-class memory capacity without 5090-class insanity.
RTX 4090 (24GB GDDR6X, $1,599 MSRP). New old stock still surfaces at some retailers, and the used market has steady supply from owners upgrading or cashing out. Twenty-four gigabytes of VRAM covers heavy 4K texture packs, 3D rendering, and local AI models up to roughly 30 billion parameters at 4-bit quantization. For gaming alone it is overkill next to the 5080, but for mixed gaming and creative work it remains the rational big-memory choice.
What about a used RTX 5090 at $1,500 over MSRP? Skip it. Paying around $3,500 for a used 5090 buys you scalper pricing plus the warranty and connector risks covered in the next section, while a new 5080 with full warranty costs far less. The math only works if you genuinely need 32GB of VRAM for professional workloads, and even then, buy from a verifiable seller or not at all.
CPU pairings that avoid bottlenecks: Same guidance as the $1,000 tier: Ryzen 7 9800X3D or 7800X3D, Core i7-14700K or better. One extra check at this tier is the power supply. A 4090’s 450W draw plus a high-end CPU wants a quality 1000W unit with a native 12V-2×6 cable, not an adapter chain from an old PSU. The connector needs to be fully seated, with no bend stress near the plug.
Practical takeaway: At $1,500 you are buying VRAM and longevity, not just frames. Make sure your power supply and case are ready for a 450W card before you commit.
The Used Market When It Makes Sense and When It Does Not
Scarcity plus hype creates the perfect conditions for bad decisions, so here is the blunt version.
Used RTX 5090 prices remain deep in scalper territory, and the scam risk is documented. Buyers have reported empty boxes, bait-and-switch listings, and in one widely covered case, a card that arrived with no PCB at all, just a cooler and backplate. The 5090’s 12V-2×6 connector has a known history of melting incidents when not perfectly seated, and a used card usually carries limited or no warranty on a 575W component with that history.
A used 5090 only makes sense when three things are all true: the price is within striking distance of the $1,999 launch MSRP, the seller is verifiable with a real return policy, and you can stress-test the card the day it arrives. In October 2026, the first condition almost never holds.
The smarter used buys are one tier down. A used RTX 4090 or RTX 4080 Super from a verifiable seller sidesteps the worst of the 5090 madness while delivering proven performance. Whatever you buy used, run a demanding benchmark loop for at least an hour, check temperatures under load, and inspect the power connector closely. If anything feels off, the return policy is your safety net. Use it.
VRAM Reality Check How Much Memory Do You Actually Need
VRAM anxiety sells graphics cards, so let us ground this in facts.
12GB is the new minimum for comfortable 1440p. Cards like the RTX 5070 handle modern games fine at 1440p, but heavy texture packs and ray tracing can push past 12GB at 4K.
16GB is the comfortable standard. This is where the RTX 5080, RTX 5070 Ti, and RX 9070 XT sit. Here is the telling detail: the official PC requirements for Call of Duty: Modern Warfare 4, which launches October 23, list the RTX 5070 Ti or RX 9070 XT with 16GB of video memory for its top ray tracing tier at 4K upscaled. When the biggest shooter launch of the year targets 16GB for its maximum settings, 16GB is not a compromise. Our Modern Warfare 4 PC requirements breakdown covers every tier.
24GB is for creators and future-proofers. Heavy 3D work, 8K video timelines, and large local AI models genuinely use it. For pure gaming, it is insurance rather than necessity today.
32GB is a workstation feature wearing a gaming badge. Only local AI work with very large models, extreme 3D rendering, or 8K production timelines need what the 5090’s frame buffer offers. If that is not you, you are not missing playable frame rates by skipping it. You are missing bragging rights.
Practical takeaway: Match VRAM to your actual workload. For gaming through the next few years, 16GB is the sweet spot, 24GB is comfortable insurance, and 32GB is a professional tool.
Should You Wait or Buy Now
This is the hardest question, so here is a decision framework instead of a guess.
Reasons to buy now. Your current GPU is holding you back in games you play today. Call of Duty: Modern Warfare 4 launches October 23, and if your card cannot deliver the experience you want, waiting means missing the launch window with friends. Street prices across the whole market are inflated by the memory shortage, and nothing on the calendar promises relief this year.
Reasons to wait. Your current card still runs your games acceptably. The rumored 24GB RTX 5080, tipped by BenchLife for the first quarter of 2027, would be a genuinely better long-term card if it materializes, with 50 percent more VRAM on the same proven architecture. But treat it as what it is: an unconfirmed rumor. Some reports suggest Fisher’s remarks pushed back on it, and neither NVIDIA nor any board partner has confirmed specifications, pricing, or a date. Waiting is a bet stacked on three uncertainties: that the card is real, that it arrives on time, and that it is priced sanely. Our RTX 5080 24GB rumor tracker will be updated the moment anything is confirmed.
The 2028 question. Reports suggest the RTX 60 series may not arrive until 2028, which means Blackwell cards bought now will not be replaced quickly. That cuts both ways: your purchase stays current longer, but NVIDIA has little competitive pressure to fix pricing.
Practical takeaway: Buy now if your current GPU is the problem. Wait only if your current GPU is fine and you specifically want more VRAM for a card you will keep four or more years.
CPU Pairings That Avoid BottlenecksÂ
A quick note on expectations: no gaming PC has literally zero bottleneck. One component always limits performance in any given scene, and chasing a perfect zero is a misunderstanding of how hardware works. What matters is keeping the bottleneck small (under about 10 percent in the calculator) and on the GPU side at your target resolution. Our bottleneck percentage guide explains the numbers in plain language.
| Your GPU tier | Safe CPU picks (AMD) | Safe CPU picks (Intel) | Notes |
|---|---|---|---|
| RX 9070 XT / RTX 5070 (1440p) | Ryzen 5 7600X, Ryzen 5 9600X | Core i5-14400F, Core i5-14600K | Balanced; GPU stays the limit |
| RTX 5080 / RTX 5070 Ti (1440p/4K) | Ryzen 7 7800X3D, Ryzen 7 9800X3D | Core i7-14700K | X3D chips shine at high refresh |
| RTX 4090 (4K / creative) | Ryzen 7 9800X3D | Core i7-14700K, Core i9-14900K | Pair with a 1000W PSU |
At 4K, the graphics card does the heavy lifting, so CPU bottlenecks shrink dramatically. At 1440p with a 240Hz monitor, the CPU matters much more, which is where the X3D chips justify their price. After building, verify your real-world balance with our guide to finding a PC bottleneck rather than guessing from specs alone.
Frequently Asked Questions
Is the RTX 5090 discontinued?
No. NVIDIA Senior Vice President Jeff Fisher told Taiwanese media on October 16, 2026 that the RTX 5090 “has definitely not been discontinued” and that production and shipments continue. He did acknowledge that supply chain capacity is constrained. The card is alive as a product but extremely hard to buy at sane prices.
Why is the RTX 5090 so expensive in October 2026?
A mix of constrained supply, surging AI demand for its 32GB of GDDR7 memory, and retailer scarcity. PC Guide’s tracking shows flagship models climbing from about $6,900 in mid-September to nearly $9,000 by October 19, with some models going out of stock entirely. Third-party Amazon and Newegg listings reach $8,500 to $10,000. These are marketplace asking prices, not NVIDIA’s $1,999 MSRP.
Should I sell my RTX 5090 while prices are this high?
This is a real question owners are asking on hardware forums right now, and it is worth considering seriously. If you only game, selling a 5090 at current prices and replacing it with an RTX 5080 or RX 9070 XT could pocket you thousands while barely changing your actual gaming experience. Keep the 5090 if you need its 32GB of VRAM for AI or creative work. Note that warranty replacements on a discontinued-then-not product could mean a refund at MSRP rather than a replacement card.
Should I buy an RTX 5090 for Microsoft Flight Simulator?
For flight sims at 4K with heavy add-ons, the 5090’s muscle is genuinely useful, but the price makes it a terrible value. An RTX 5080 or even RTX 5070 Ti handles Microsoft Flight Simulator well at 1440p and respectably at 4K, especially with DLSS. Paying triple MSRP for the last few percentage points of smoothness is not rational unless the budget is truly unlimited.
Why would I need an RTX 5090 when an RTX 5080 handles 4K fine?
You probably would not. That is the correct instinct. The 5080 delivers roughly 70 percent of the 5090’s 4K performance for a fraction of the street price, with full warranty and actual availability. The 5090’s real advantages are its 32GB frame buffer for professional workloads and its halo status. For gaming, the 5080 plus DLSS 4 is the rational flagship.
Should I wait for the RTX 5090 to come back in stock or buy something else now?
NVIDIA confirms shipments continue but gives no restocking timeline, so waiting has no guaranteed payoff. If you need a GPU now, buy the best card in your tier today: RX 9070 XT near $700, RTX 5080 near $1,000. If your current card still plays your games well, waiting costs you nothing and lets you see whether the rumored 24GB RTX 5080 materializes in early 2027.
Are GPU prices going to drop soon?
Nothing on the calendar points to near-term relief. The price pressure comes from a structural memory shortage and AI demand absorbing high-end supply, not a temporary retail hiccup. Prices could stabilize if memory supply improves, but no credible report promises drops in 2026. Buy for your current needs rather than timing a market nobody can predict.
Is a used RTX 5090 worth the risk?
Almost never at current prices. You would pay scalper pricing while accepting scam exposure, limited warranty on a 575W card, and the documented 12V-2×6 connector concerns. A used 5090 only makes sense near the $1,999 launch MSRP from a verifiable seller with a real return policy, a combination that barely exists right now.
Key Takeaways
- NVIDIA officially denies the RTX 5090 is discontinued, but $9,000 street prices, “Not Carried” retail listings, and vanishing stock make it effectively unbuyable for gamers.
- Around $700: RX 9070 XT for value, RTX 5070 Ti for DLSS 4 and ray tracing. Pair with a Ryzen 5 9600X or Core i5-14600K.
- Around $1,000: RTX 5080 is the new reachable flagship. A verified used RTX 4090 is the alternative if 24GB VRAM matters.
- Around $1,500: RTX 4090 for big VRAM and creative work. Skip used 5090s at scalper prices.
- 16GB of VRAM is the gaming sweet spot, confirmed by Modern Warfare 4’s official 4K ray tracing requirements.
- Buy now if your current GPU holds you back; wait only if it does not and you want the unconfirmed 24GB 5080.
November 10 Update: NVIDIA Repeats the Denial as the Shortage Reaches Prebuilt PCs
On November 10, 2026, NVIDIA said it again. In a formal statement reported by Guru3D and ComputerBase, the company confirmed the GeForce RTX 5090 has not been discontinued and that GB202 allocations remain active for board partners. It is the second public denial in less than a month, and it arrived as fresh evidence showed the shortage spreading from standalone cards into prebuilt gaming PCs.
Senior Vice President Jeff Fisher put it plainly: “The supply chain (capacity) is indeed somewhat constrained, but NVIDIA is fully committed to bringing products to market. Although there are other variants of the RTX 5090 chip, it has not been discontinued, and NVIDIA will continue shipping, which is very important.” The statement is a repeat, not a reversal. Fisher gave the same denial on October 16, and the October section above covers it in full. What is new is the date, the formal framing, and the fact NVIDIA said it twice in 25 days.
The production side of the picture is unchanged. TSMC 4N lines for GB202 remain operational, and board partners ASUS, MSI, Gigabyte, and Palit report that existing allocations will keep shipping to fulfill backorders. That is good news for anyone already in a queue. The catch is the pace: availability is expected to stay tight over the coming quarter, so backorders will clear slowly and walk-in stock will stay rare.
The new development is where the shortage shows up next. Notebookcheck reported on November 10, 2026 that the RTX 5090 is now disappearing from prebuilt gaming PCs too, a channel many buyers had treated as a workaround. The HP Omen Max 45L with an RTX 5090, listed at $7,999.99, is out of stock. Lenovo shows no RTX 5090 desktop configurations at all. On Newegg, the MSI Suprim lists at $5,899.99, about 195 percent above the $1,999 launch price, and the Asus TUF Gaming lists at $5,099.99, about 155 percent above it. Skytech and CLX prebuilts carrying the card run about $9,000 to $9,850.
| Seller or model | Price | Against the $1,999 launch price |
|---|---|---|
| MSI Suprim (Newegg) | $5,899.99 | About 195 percent above launch |
| Asus TUF Gaming (Newegg) | $5,099.99 | About 155 percent above launch |
| HP Omen Max 45L prebuilt | $7,999.99 | Out of stock |
| Skytech and CLX prebuilts | About $9,000 to $9,850 | Listed, at extreme premiums |
| Lenovo RTX 5090 desktops | No configurations listed | Not available |
Why is this happening? Reports point to AI demand. According to those reports, retail cards are being converted into blower assemblies for AI servers, which pulls supply away from gamers. Treat that as reports, not confirmed fact. NVIDIA has not confirmed that retail stock is being diverted, and no official statement ties the shortage to a specific buyer. What is confirmed is the outcome on the shelves.
So what does the November 10 statement actually change? On the facts that matter to a buyer, nothing yet. The denial is real and on the record, and production continues, but no restocking timeline was given, no shipment numbers were shared, and prices moved the wrong way the same day. A second denial without a date is reassurance, not relief. The price picture is unchanged: the card stays effectively unbuyable near its launch price, and the prebuilt route is now showing the same strain.
What to do now. If you can wait, wait. A card in a queue with an allocation behind it beats a panic buy at triple the launch price. If you need a card now, buy an alternative and keep your money. The RTX 5080 is the closest stand-in for 4K gaming, and a Ryzen 7 9800X3D or Core i7-14700K feeds it without waste. The RX 9070 XT is the value pick for 1440p high-refresh play, and a Ryzen 5 9600X or Ryzen 7 7700X is plenty of CPU for it. The RTX 5070 Ti sits between them on features, with DLSS 4 in its favor, and pairs cleanly with a Ryzen 7 7800X3D or Core i5-14600K. Buying used is the third option, and it only works with the rules the earlier section sets out: a verifiable seller, a price near launch MSRP, and proof the card is genuine. At current street prices, a used 5090 fails all three tests for most buyers.
The bottom line of this update is simple. NVIDIA has now said twice that the RTX 5090 is not dead, and the second statement carries the same message as the first. Believe the denial, but buy on the prices. Until the shelves tell a different story, the smart money stays on the 5080, the 9070 XT, or the 5070 Ti, matched to a CPU that can actually feed them.
Conclusion
The RTX 5090 is not discontinued. It is something stranger: a flagship that exists on NVIDIA’s roadmap but barely exists on shelves, priced like a luxury good and hunted by AI firms instead of gamers. Jeff Fisher’s denial answers the rumor, but the market answered louder. When pallets of graphics cards vanish at airports and anniversary bundles cost $12,000, the rational gamer stops chasing the halo.
The good news is that the tiers below the flagship have never been more capable. An RX 9070 XT or RTX 5070 Ti handles 1440p beautifully and 4K with upscaling. An RTX 5080 delivers genuine high-end 4K gaming with a warranty and an honest price. And Call of Duty: Modern Warfare 4’s own requirements confirm that 16GB cards are the target for maximum settings, not a compromise.
Before you buy, run your planned combination through our CPU bottleneck calculator to confirm the pairing is balanced at your resolution. The best upgrade is the one that removes your actual bottleneck, not the one with the biggest box.
Leave a Reply