In April 2026, the Render Network community completed a governance vote widely regarded in the industry as a "high-stakes gamble." Proposal RNP-023 passed its first round with an overwhelming 98.86% approval, officially integrating Salad Network as the exclusive subnet within the Render ecosystem. This move brought approximately 60,000 daily active GPUs into the fold.
Salad Network isn’t a traditional data center compute provider. It operates the world’s largest consumer-grade GPU network, spanning over 180 countries with more than 450,000 registered nodes and about 60,000 daily active GPUs. Its compute power comes from idle graphics cards owned by gamers and individual users—primarily consumer models like RTX 3070, RTX 3080, RTX 3090, and RTX 4090. This stands in stark contrast to hyperscale cloud providers like AWS and GCP, which rely on enterprise-grade clusters such as A100 and H100.
As of May 19, 2026, Gate market data shows RENDER trading at $1.8254, up 2.90% over 24 hours, with a circulating market cap of roughly $946 million. Market sentiment remains neutral.
Key Facts:
- RNP-023’s first round saw 1.3 million votes in favor and 15,500 against, with a 98.86% approval rate
- Salad Network’s ~60,000 daily active GPUs will be integrated as the exclusive subnet for Render
- Integration is structured around three milestones: Stage 1—Chefs receive RENDER rewards; Stage 2—clients can pay with RENDER; Stage 3—all transactions migrate to on-chain BME model
- Prior to migration, Render Network had about 5,700 active GPU nodes, with over 71.4 million rendered frames processed
- At NVIDIA GTC 2026, Jensen Huang projected that demand for Blackwell and Vera Rubin architecture AI chips would reach at least $1 trillion by the end of 2027, doubling last year’s forecast
From BME to RNP-023: The Render Expansion Timeline
Render Network’s expansion of compute power isn’t an isolated event. Its evolution is nested within two macro trends: the structural hunger for GPU resources driven by AI models, and the transition of decentralized physical infrastructure networks from narrative to real-world deployment.
Timeline:
- 2023: The community passes proposal RNP-002, migrating Render from Ethereum to Solana and introducing the Burn-and-Mint Equilibrium (BME) tokenomics model. Under BME, GPU task payments are burned, and new tokens are minted as needed, dynamically linking token supply to actual network usage
- 2024–2025: The network validates the feasibility of distributed GPU resource scheduling. AI inference and fine-tuning tasks rise steadily, reaching nearly 40% of total network activity by early 2026
- March 2026: Salad submits a formal proposal to join Render Network as the exclusive subnet
- March 2026: NVIDIA GTC 2026 takes place, with Jensen Huang announcing the $1 trillion demand forecast and giving industry-level backing to the GPU shortage narrative
- April 1, 2026: RNP-023’s first round of voting concludes, with 98.86% approval
- April 7, 2026: RNP-023 is officially passed; Salad confirmed as part of Render Network
BME Model & Integration’s Key Transmission Chain: One of RNP-023’s core designs is to channel Salad’s compute revenue into the BME burn mechanism. Salad’s founder publicly stated: "The design of burning more than minting is deliberate—we want Salad’s growth to benefit the entire Render ecosystem, not just ourselves." This suggests (hypothetically) that if Salad’s integration significantly boosts network usage, the BME model will see increased token burning, tightening supply through the logic of "demand growth → accelerated burning → supply contraction." Whether this holds true depends on actual usage, not just the proposal text.
Compute Surge & Cost Realities: A Data Perspective
Step Change in Compute Supply: From 5,700 to Over 65,000
Before integration, Render Network had about 5,700 active GPU nodes. Salad Network’s addition brings roughly 60,000 daily active GPUs, raising the network’s theoretical available GPUs to over 65,000—a discontinuous jump in supply, not just incremental optimization but a shift in scale.
However, "number of GPUs" doesn’t equate to "usable compute power." Consumer-grade GPUs differ significantly from enterprise-grade GPUs across several dimensions:
Consumer vs. Enterprise GPUs: Key Differences
| Dimension | Consumer GPUs (Salad) | Enterprise GPUs (AWS/GCP) |
|---|---|---|
| Typical Models | RTX 3070/3080/3090/4090 | A100 80GB / H100 80GB / H200 |
| VRAM | 8GB–24GB | 40GB–141GB |
| Interconnect Bandwidth | PCIe (no NVLink/NVSwitch) | NVLink + NVSwitch (high-bandwidth interconnect) |
| Use Cases | AI inference, batch processing, small/medium rendering | Large-scale distributed training, full-parameter fine-tuning of 70B+ models |
| Node Reliability | Personal devices, can go offline anytime | Data center-grade, 99.9%+ SLA |
| Unit Cost | Extremely low ($0.02/hour starting) | High (H100 ~$4.50–$5.50/hour) |
Salad’s positioning supports this division of labor. Its official blog notes that open-source AI models increasingly run on consumer hardware, and Agentic AI workloads are surging, with each interaction requiring computational resources several orders of magnitude greater than traditional API calls. Salad’s customer cases show that running workloads on consumer GPUs enables scaling while significantly reducing costs. This means Render’s post-integration network isn’t aiming to replace AWS/GCP in all scenarios, but rather focuses on tasks tolerant of latency, highly cost-sensitive, and divisible/parallelizable.
The Price Gap with AWS: Up to 90% Savings
This is the most crucial data dimension for understanding Render’s competitive relationship with AWS/GCP. The following compares publicly available price data from the first half of 2026:
H100 GPU Price Comparison
| Provider | GPU Type | On-Demand Price ($/hour) | Notes |
|---|---|---|---|
| AWS (single card) | 1×H100 80GB | ~$4.50–$5.50 | Securities.io industry estimate |
| Decentralized Networks (Akash/Render) | 1×H100 80GB | ~$1.20–$1.80 | Securities.io data |
| Salad (consumer-grade) | Starting price | 0.02 | salad.com homepage data |
Sources: H100 single card estimates and decentralized prices from Securities.io; Salad starting price from salad.com. Prices vary by region, supply fluctuations, and priority settings; for reference only.
Decentralized networks offer H100-grade GPUs at 25%–35% of AWS’s on-demand price, saving 65%–75%. Consumer GPUs (RTX series) start as low as $0.02/hour, with price differences exceeding 90% compared to hyperscale cloud providers.
But a key logic must be clarified: Low price does not mean replacement is possible. For large-scale synchronous training tasks requiring InfiniBand high-speed interconnects, centralized clusters remain the only viable architecture. AWS and GCP have hardware interconnect advantages that decentralized solutions can’t match. Render Network’s value proposition is to fill the gap for "tasks that don’t need high-end interconnects but require massive parallel compute"—AI inference, batch processing, small/medium model fine-tuning, 3D rendering, etc.
Over 1.22 Million Burned: Network Usage & Token Fundamentals
According to Q1 2026 data, Render Network processed over 71.4 million rendered frames, with AI workloads accounting for nearly 40%. More than 1.22 million RENDER tokens have been burned.
Official Q1 2026 Render Network metrics:
| Metric | 2026 Q1 |
|---|---|
| Active GPU Nodes | Over 5,700 |
| Total Frames Processed | 71,269,082 |
| AI Workload Share | Nearly 40% |
| Total RENDER Burned | 1,228,380 |
| Circulating Supply | 552,011,095 / 644,168,762 max supply |
After Salad integration, theoretical GPU node count jumps to 65,000+, but actual concurrent online numbers depend on scheduler efficiency and Chefs’ participation, requiring ongoing operational data.
Tokenomics Perspective (Facts & Analysis): Render’s BME model creates a structural link between network usage and token supply/demand. Salad’s integration will channel part of its revenue into the BME burn process. The actual impact should be tracked via ongoing burn and usage data, not overinterpreted.
Market Divergence: Three Factions Interpret RNP-023
Expansion Optimists: Scale as a Moat
Supporters argue that Render, through Salad integration, gains a compute supply source that traditional cloud providers can’t replicate—millions of gamers’ idle GPUs worldwide. This supply has unique features: ultra-low marginal cost (hardware already purchased, compute is a "byproduct"); highly dispersed geography (180+ countries); scale with network effects (more Chefs means more compute, attracting more clients).
Salad founder Bob Miles stated after the proposal passed: "Open-source AI models are increasingly running on consumer hardware. Agentic AI workloads are surging—each interaction requires computational resources several orders of magnitude greater than traditional API calls. The machines our Chefs run are precisely the infrastructure the industry needs."
Render’s disclosed institutional partners reinforce this narrative—NVIDIA, Stability AI, WME, and others have partnered with Render. NVIDIA’s involvement is especially noteworthy: why would a GPU manufacturing giant focus on decentralized compute networks? (Hypothetically) The logic may be that any ecosystem expanding GPU use cases benefits NVIDIA’s core chip business.
Cautious Observers: Scale Doesn’t Equal Revenue
More reserved viewpoints focus on hard data. Salad’s integration brings significant expansion in compute supply, but how much actual revenue does it contribute? Salad’s founder hasn’t publicly projected specific revenue. Crypto protocol valuation models don’t match traditional P/E frameworks; network effects, narrative premiums, and growth expectations weigh more in token pricing.
Analysts also note that RNP-023 is a governance event; its real impact depends on execution, not the vote itself. In crypto markets, "buy the rumor, sell the news" event-driven logic is common.
Competitive Structure: DePIN’s Internal Rivalry
Salad’s proposal explicitly states it "chose not to issue its own token" and instead joined Render, citing "Render’s strong team, infrastructure, and community." This choice means Salad forgoes independent token value capture, binding its compute supply to Render’s BME model.
Meanwhile, decentralized compute isn’t Render’s exclusive domain. Akash Network’s open market for general containerized applications and io.net’s focus on AI compute scheduling both overlap with Render. As Salad integration pushes Render to larger scale, competition boundaries among DePIN compute protocols will become more complex.
Behind the Numbers: Three Layers of the 60,000 GPU Narrative
In crypto, narrative often precedes fundamentals. "60,000 GPUs" is a powerful headline, but it warrants a layered breakdown.
Layer One: Are 60,000 GPUs Real? Salad’s official data says "60,000 daily active machines across 180+ countries." Other sources note Salad’s ecosystem has over 450,000 registered nodes. The 60,000 figure comes directly from Salad and has been confirmed by at least six independent sources. But given consumer GPU network traits, daily active numbers may fluctuate, and actual concurrent online counts differ from registered active devices.
Layer Two: Can These GPUs Be Used by Render? (Hypothetical, based on proposal) The integration plan makes Salad Render’s "exclusive subnet," with all payments through Salad gradually migrating to RENDER on-chain settlement. This economically binds these GPUs to Render. Technically, however, consumer GPUs face offline risk, network latency, and compute volatility—structural traits that can’t be fully eliminated. Salad’s support docs clearly state that, due to distributed and interruptible network nature, hardware ROI isn’t guaranteed and income may fluctuate daily. Whether these GPUs can reliably serve commercial AI and rendering tasks depends on Salad’s scheduler and Render’s task integration.
Layer Three: Does More GPUs Automatically Mean Higher Network Value? (Opinion) It depends on two conditions: whether these GPUs consistently receive paid tasks, and whether those tasks convert to token value via the BME model. The transmission chain has multiple variables—client acquisition speed, task pricing, competition pressure—and lacks sufficient verifiable data for a definitive conclusion.
Industry Impact: From Integration to Substitution
DePIN Track Accelerates Integration
RNP-023 marks DePIN compute’s shift from "independent project development" to "scale integration." Salad’s decision not to launch its own token but join Render may signal that smaller compute networks will increasingly integrate with leading protocols rather than compete alone. If this model proves viable, DePIN’s Matthew effect will accelerate.
Complementary, Not Disruptive: The Real Shift in Cloud Services
Can decentralized compute really "shake up" AWS/GCP? It depends on how "shake up" is defined. If it means "replacing centralized cloud in all GPU compute scenarios," the answer is clearly no. As Securities.io’s comparison report notes, for large-scale synchronous foundation model training requiring ultra-low latency interconnects, centralized clusters remain the only viable architecture.
But if it means "diverting incremental demand in cost-sensitive scenarios," the answer leans yes. Decentralized networks offer 65%–75% discounts, with consumer GPU scenarios saving up to 90%.
Decentralized compute’s market entry is more about "complementary diversion" than "disruptive replacement." This judgment (opinion) is based on verifiable logic: consumer GPUs’ low-cost advantage is real in inference and batch processing, but high-end training needs low-latency interconnects, SLA guarantees, and data governance—requirements that distributed consumer networks can’t physically meet.
New Variables for the BME Model
Salad’s integration introduces a new burn source for the BME model. Structurally, this expands RENDER token demand from "render task payment" to "on-chain payment for consumer GPU compute," broadening token utility. Salad’s founder emphasized the deliberate design of "burning more than minting," and Salad’s post-integration revenue will structurally impact token supply and demand. But actual impact depends on sustained network usage growth and needs long-term observation.
Conclusion
Render Network’s integration of Salad Network’s 60,000 consumer-grade GPUs via RNP-023 is one of the most significant scaling events in the 2026 DePIN sector. It demonstrates the feasibility of decentralized compute networks achieving supply-side scale—a core bottleneck for the sector.
But the real value of "60,000 GPUs" isn’t in the number itself. It depends on whether Render can convert these GPUs into sustainable network usage and token value capture. As of May 19, 2026, Render’s circulating market cap is about $946 million, with RENDER trading at $1.8254. The compute supply surge from Salad integration is reflected in network fundamentals, but revenue scale, client acquisition, and BME burn data still require a longer window for validation.
From an industry perspective, the relationship between decentralized compute and AWS/GCP is best described as "cost substitution in specific scenarios," not "full competition." This isn’t a failure for decentralized compute—quite the opposite. In a market dominated by a handful of hyperscale cloud providers for two decades, any ability to break through on cost is a structural experiment worthy of serious attention.
