Solana has not had any reported outages of its mainnet in two years. In a study published on October 16, the Fidelity Center for Applied Technology estimates that the network has made significant progress in resilience, economic security and performance. However, the observation remains incomplete: staking, hosting of validators and their software remain concentrated between relatively few players.

In brief
- Solana has not had any reported mainnet outages in two years.
- Around 18 validators bring together more than a third of the SOL staked.
- Fidelity especially monitors the concentration of staking, hosting and software clients.
Solana is gradually erasing its old problem of breakdowns
For a long time, network availability was one of the main arguments used against Solana. The picture has changed. Fidelity notes two consecutive years without a reported mainnet outage. The success rate of validator votes will be around 99.7% in 2026, while the average vote latency is around 1.08 slots.
The network is also speeding up. Solana has already reduced its block creation time to 250 milliseconds, after several successive reductions since August. Performance comes with much greater financial activity. According to the Fidelity Center for Applied Technology report, the value of real-world assets tokenized using Solana increased from approximately $1 billion to $4.4 billion in one year.
The network also processed over $2 trillion in stablecoin volume in the first quarter of 2026. Solana still has capacity. After increasing the maximum block size to 100 million compute units in late July, Fidelity observed utilization between approximately 30 and 45 percent over its study period. The blockchain therefore runs faster without constantly operating at its limit.
18 validators concentrate a third of the staking
This is where the report becomes much more cautious. Around 18 validators are currently enough to raise more than a third of the SOL staked. In the Solana consensus system, this threshold has particular importance: if this amount of stake simultaneously stopped participating, the finalization of blocks could be interrupted.
This would not allow these validators to steal user assets or rewrite already finalized transactions. The concentration nevertheless remains visible. Solana had already lost 68% of its validators in three years, going from around 2,500 operators to less than 1,000. Fidelity adds another nuance: several validators can belong to the same company. For example, Figment uses several important identities. Kraken and Coinbase also have several validators.
Counting validators is therefore not exactly the same as counting independent operators. The physical infrastructure is also concentrated. Nearly 59% of the stake would be hosted in Germanyin the Netherlands and the United States. Frankfurt and Amsterdam alone would represent around 40%. A hosting provider also accounts for around 29% of the stake, above the 25% objective set in the Solana Foundation delegation program.
Fidelity also monitors validator software
The third risk is in the code. On paper, Solana uses multiple client configurations today. Diversification is therefore better than a few years ago. But a large part still shares the same software base. Jito variants represent approximately 63% of the stake in the data studied by Fidelity. Other solutions, such as Harmonic, also remain largely linked to the Agave code.
A bug present in this common database could therefore affect several categories of validators at the same time. Firedancer makes a real difference. Developed by Jump Crypto on an independent C++ basis, it represents around 15% of the stake when its different configurations are grouped together.
This diversity becomes important as Solana seeks to accommodate more institutional finance. The network recently surpassed $4 billion in tokenized assets, with increasing flows into tokenized treasuries and stocks. Fidelity therefore no longer describes Solana as a fragile network. The problem has moved. Breakdowns have become much rarer. The next step is mainly to prevent a handful of operators, data centers or software from becoming common points of weakness.
Maximize your Tremplin.io experience with our ‘Read to Earn’ program! For every article you read, earn points and access exclusive rewards. Sign up now and start earning benefits.
